## The Zen Anti-Interpretation of Quantum Mechanics

As I lay bedridden this week, knocked out by my second dose of the Moderna vaccine, I decided I should blog some more half-baked ideas because what the hell? It feels therapeutic, I have tenure, and anyone who doesn’t like it can close their broswer tab.

So: although I’ve written tens of thousands of words, on this blog and elsewhere, about interpretations of quantum mechanics, again and again I’ve dodged the question of which interpretation (if any) I really believe myself. Today, at last, I’ll emerge from the shadows and tell you precisely where I stand.

I hold that all interpretations of QM are just crutches that are better or worse at helping you along to the Zen realization that QM is what it is and doesn’t need an interpretation.  As Sidney Coleman famously argued, what needs reinterpretation is not QM itself, but all our pre-quantum philosophical baggage—the baggage that leads us to demand, for example, that a wavefunction |ψ⟩ either be “real” like a stubbed toe or else “unreal” like a dream. Crucially, because this philosophical baggage differs somewhat from person to person, the “best” interpretation—meaning, the one that leads most quickly to the desired Zen state—can also differ from person to person. Meanwhile, though, thousands of physicists (and chemists, mathematicians, quantum computer scientists, etc.) have approached the Zen state merely by spending decades working with QM, never worrying much about interpretations at all. This is probably the truest path; it’s just that most people lack the inclination, ability, or time.

Greg Kuperberg, one of the smartest people I know, once told me that the problem with the Many-Worlds Interpretation is not that it says anything wrong, but only that it’s “melodramatic” and “overwritten.” Greg is far along the Zen path, probably further than me.

You shouldn’t confuse the Zen Anti-Interpretation with “Shut Up And Calculate.” The latter phrase, mistakenly attributed to Feynman but really due to David Mermin, is something one might say at the beginning of the path, when one is as a baby. I’m talking here only about the endpoint of the path, which one can approach but never reach—the endpoint where you intuitively understand exactly what a Many-Worlder, Copenhagenist, or Bohmian would say about any given issue, and also how they’d respond to each other, and how they’d respond to the responses, etc. but after years of study and effort you’ve returned to the situation of the baby, who just sees the thing for what it is.

I don’t mean to say that the interpretations are all interchangeable, or equally good or bad. If you had to, you could call even me a “Many-Worlder,” but only in the following limited sense: that in fifteen years of teaching quantum information, my experience has consistently been that for most students, Everett’s crutch is the best one currently on the market. At any rate, it’s the one that’s the most like a straightforward picture of the equations, and the least like a wobbly tower of words that might collapse if you utter any wrong ones.  Unlike Bohr, Everett will never make you feel stupid for asking the questions an inquisitive child would ask; he’ll simply give you answers that are as clear, logical, and internally consistent as they are metaphysically extravagant. That’s a start.

The Copenhagen Interpretation retains a place of honor as the first crutch, for decades the only crutch, and the one closest to the spirit of positivism. Unfortunately, wielding the Copenhagen crutch requires mad philosophical skillz—which parts of the universe should you temporarily regard as “classical”? which questions should be answered, and which deflected?—to the point where, if you’re capable of all that verbal footwork, then why do you even need a crutch in the first place? In the hands of amateurs—meaning, alas, nearly everyone—Copenhagen often leads away from rather than toward the Zen state, as one sees with the generations of New-Age bastardizations about “observations creating reality.”

As for deBroglie-Bohm—well, that’s a weird, interesting, baroque crutch, one whose actual details (the preferred basis and the guiding equation) are historically contingent and tied to specific physical systems. It’s probably the right crutch for someone—it gets eternal credit for having led Bell to discover the Bell inequality—but its quirks definitely need to be discarded along the way.

Note that, among those who approach the Zen state, many might still call themselves Many-Worlders or Copenhagenists or Bohmians or whatever—just as those far along in spiritual enlightenment might still call themselves Buddhists or Catholics or Muslims or Jews (or atheists or agnostics)—even though, by that point, they might have more in common with each other than they do with their supposed coreligionists or co-irreligionists.

Alright, but isn’t all this Zen stuff just a way to dodge the actual, substantive questions about QM, by cheaply claiming to have transcended them? If that’s your charge, then please help yourself to the following FAQ about the details of the Zen Anti-Interpretation.

1. What is a quantum state? It’s a unit vector of complex numbers (or if we’re talking about mixed states, then a trace-1, Hermitian, positive semidefinite matrix), which encodes everything there is to know about a physical system.
2. OK, but are the quantum states “ontic” (really out in the world), or “epistemic” (only in our heads)? Dude. Do “basketball games” really exist, or is that just a phrase we use to summarize our knowledge about certain large agglomerations of interacting quarks and leptons? Do even the “quarks” and “leptons” exist, or are those just words for excitations of the more fundamental fields? Does “jealousy” exist? Pretty much all our concepts are complicated grab bags of “ontic” and “epistemic,” so it shouldn’t surprise us if quantum states are too. Bad dichotomy.
3. Why are there probabilities in QM? Because QM is a (the?) generalization of probability theory to involve complex numbers, whose squared absolute values are probabilities. It includes probability as a special case.
4. But why do the probabilities obey the Born rule? Because, once the unitary part of QM has picked out the 2-norm as being special, for the probabilities also to be governed by the 2-norm is pretty much the only possibility that makes mathematical sense; there are many nice theorems formalizing that intuition under reasonable assumptions.
5. What is an “observer”? It’s exactly what modern decoherence theory says it is: a particular kind of quantum system that interacts with other quantum systems, becomes entangled with them, and thereby records information about them—reversibly in principle but irreversibly in practice.
6. Can observers be manipulated in coherent superposition, as in the Wigner’s Friend scenario? If so, they’d be radically unlike any physical system we’ve ever had direct experience with. So, are you asking whether such “observers” would be conscious, or if so what they’d be conscious of? Who the hell knows?
7. Do “other” branches of the wavefunction—ones, for example, where my life took a different course—exist in the same sense this one does? If you start with a quantum state for the early universe and then time-evolve it forward, then yes, you’ll get not only “our” branch but also a proliferation of other branches, in the overwhelming majority of which Donald Trump was never president and civilization didn’t grind to a halt because of a bat near Wuhan.  But how could we possibly know whether anything “breathes fire” into the other branches and makes them real, when we have no idea what breathes fire into this branch and makes it real? This is not a dodge—it’s just that a simple “yes” or “no” would fail to do justice to the enormity of such a question, which is above the pay grade of physics as it currently exists.
8. Is this it? Have you brought me to the end of the path of understanding QM? No, I’ve just pointed the way toward the beginning of the path. The most fundamental tenet of the Zen Anti-Interpretation is that there’s no shortcut to actually working through the Bell inequality, quantum teleportation, Shor’s algorithm, the Kochen-Specker and PBR theorems, possibly even a … photon or a hydrogen atom, so you can see quantum probability in action and be enlightened. I’m further along the path than I was twenty years ago, but not as far along as some of my colleagues. Even the greatest quantum Zen masters will be able to get further when new quantum phenomena and protocols are discovered in the future. All the same, though—and this is another major teaching of the Zen Anti-Interpretation—there’s more to life than achieving greater and greater clarity about the foundations of QM. And on that note…

To those who asked me about Claus Peter Schnorr’s claim to have discovered a fast classical factoring algorithm, thereby “destroying” (in his words) the RSA cryptosystem, see (e.g.) this Twitter thread by Keegan Ryan, which explains what certainly looks like a fatal error in Schnorr’s paper.

### 317 Responses to “The Zen Anti-Interpretation of Quantum Mechanics”

1. Moshe Says:

Much of this is the viewpoint of many working physicists (to paraphrase, to find an interpretation of QM you first need to work on your idea of what would an “interpretation” might mean such that it is not incompatible with the facts of QM). Welcome!

2. AtomZ Says:

Scott: considering your third paragraph, have you ever given any thought to quantum logics? If we go down the rabbit hole of QM interpretations often being the result of our attempts to impart classical logic where it doesn’t belong, who’s to say that all of logic itself might not really be empirically based on our classical, macroscopic experience? Are things like paraconsistent logics or other multi-valued logics worth considering?

3. Partisan Says:

I tend to see arguments about the ‘right’ interpretation of QM as basically equivalent to arguing about whether we should think of ourselves as stationary or as rotating about the Earth’s axis (or perhaps also about the sun-Earth center of mass and the Milky way and….). Except that for some reason, in the first case we all agree that you can choose whatever perspective you want, as long as you’re consistent, and that different perspectives might be more or less useful in different situations, while in the second case we’re supposed to pick one and loyally defend it from any who would besmirch its honor.

I don’t know if attitude this is quite the same as the ‘Zen’ school of thought, but adopting it has certainly made me feel more Zen!

4. Moshe Says:

Let me also advocate an attitude that is less common. To compare classical and quantum, it is useful to have a formalism which can include both. That might be harder for deterministic classical systems, but there are presentations of classical stochastic systems which have mathematical elements very similar to QM (operators, Hilbert space, path integrals). Familiarity with this is useful to distinguish issues of language from more substantive issues. For example, lots of puzzles presumably about QM are really puzzles about the meaning of classical probability (for example your point 7 would be identical, in fact more justified, if the world was described by a classical probability distribution).

5. Anthony Aguirre Says:

Love it! Someone should write a book.

6. Pedro Says:

To me this is more like being agnostic about interpretations than transcending interpretations.

7. Jon Awbrey Says:

❝If exegesis raised a hermeneutic problem, that is, a problem of interpretation, it is because every reading of a text always takes place within a community, a tradition, or a living current of thought, all of which display presuppositions and exigencies — regardless of how closely a reading may be tied to the quid, to “that in view of which” the text was written.❞

8. Boaz Barak Says:

I agree and I think I see the distinction between this and shut up and calculate.
I tried to make a similar point toward the end of my quantum skepticism blog https://windowsontheory.org/2017/10/30/the-different-forms-of-quantum-computing-skepticism/

I think Democritus didn’t posit a physical theory as much as give a framework for physical theories, but it turns out that this framework is not enough to capture the actual world we live in. It still had a pretty good run

9. Scott Says:

Jon Awbrey #7: Wha? 🙂

10. Scott Says:

Pedro #6: “Agnosticism” sounds very far from the position I expressed here! It’s not that I don’t know which interpretation is the true one, due to insufficient evidence. Rather, it’s that I feel confident that all of them capture aspects of a situation that can be made familiar through years of working with QM, but can’t (so far as I know) be precisely articulated in words.

11. Boogabaz Says:

As always, when you speak of physics, your words cause the blooming of the dawn. Illusions crumble, fogs lift, wavefunctions- never collapse, they just go to the big infinite square well in the sky.

You write about quantum mechanics the same way Moldbug writes about politics. This is the highest praise I’m able to give.

12. Scott Says:

Boogabaz #11:

You write about quantum mechanics the same way Moldbug writes about politics. This is the highest praise I’m able to give.

Err .. thank you, I think, although I probably won’t request a recommendation letter from you anytime soon. 😀

13. David Karger Says:

Sidney Coleman was my freshman physics teacher; I credit him with making me the computer scientist I am today (by scaring my away from physics).

Have you read Greg Bear’s Moving Mars? QM plays a fun side role.

14. Ryan O'Donnell Says:

Hey Scott, maybe you can give your thoughts on something I (a person who doesn’t know physics) have yet to figure out.

So a major source of difficulties in QM interpretation seems to be the fact that on one hand we have unitaries acting reversibly on quantum states, but on the other hand measurements are some other, non-reversible thing… and how are we supposed to decide what’s what?
But there’s a completely fine and enjoyable setup where everything is consistent: quantum states are trace-1 PSD operators, and every transformation (unitaries, measurements, whatever) is a CPTP map. That’s it. Everything’s unified and consistent and nice. Why is this not satisfactory?
Here’s the complaints I think people make:

Complaint 1: It’s not reversible!
Response: Why is that bad? I don’t see why people want to hold on to reversibility so much.

Complaint 2: Mixed states are probabilistic! (The worry being, I guess, that this is somehow unphysical, and that probabilities are only supposed to reflect lack of knowledge.)
Response: Well the cat’s kind of already out of the bag with admitting probability into the world, with measurements of pure states.

So what’s wrong with just declaring: mixed states are first-class objects, the true representation of states; CPTP maps are the allowed transformations; and everything’s fine?

15. Shmi Says:

Hope your second-dose malaise goes away shortly, leaving long-term immunity in its wake. At least most US states seem to finally have their gotten act together. Here in Canada someone under 60 without extreme health issues is not expected to get even the first dose until July or August.

Also, would it be fair to call your state “Quantum Enlightenment”?

16. Daniel Harlow Says:

We are pretty much in agreement here Scott, especially about the silliness of the ontic/epistemic discussion and the acknowledgment that many of the confusions which are supposedly about quantum mechanics are really about consciousness (and also about classical probability I would add). I think point 3) however shows that you aren’t really an Everettian: a true many-worlder would say that there is nothing to quantum mechanics beyond the Schrodinger equation, while you have (correctly) emphasized that the probabilistic interpretation is an additional ingredient which is just as fundamental.

17. J Says:

Scott, how charming! Hope you’ll get better soon. Two questions:

1) on first read “QM doesn’t need an interpretation” seems to contradict “interpretations of quantum mechanics are not jealous, use them all -at least I do”. But actually I feel like your philosophy is best described as in the same vein rather than opposite to Shor. Wrong reading from me?

2) Speaking of zen things, surely you’ve read the Tao of quantum interrogation by Kwiat?

https://physics.illinois.edu/people/Kwiat/Interaction-Free-Measurements.htm

The maths are easy (or just very very well explained), but I always wondered which interpretation(s) would be the most natural/helpful to grasp what these interaction-free measurements means. What’s your own take?

18. Scott Says:

Ryan O’Donnell #14: The short answer is, no one has ever observed a CPTP map in nature that wasn’t part of a larger unitary transformation. And if such a CPTP map did exist, it would kinda break the entire structure of physics since Galileo (conservation of phase space volume, the Euler-Lagrange equations, etc etc), all of which, even long before QM, was built up on a foundation of time-reversibility.

In the 70s, as you might know, Hawking proposed that black holes do exactly the thing you ask to infalling quantum states: namely, apply CPTP maps that fundamentally destroy the information. But he and everyone else knew exactly how high the stakes were, that this would be the end of reversibility in physics. And today, because of AdS/CFT, string theory, yadda yadda, there’s a near-consensus that Hawking was wrong and that even black holes should just apply unitary transformations, leaking out the infalling quantum states in scrambled form in the Hawking radiation (Hawking himself conceded a bet about this in 2004).

19. matt Says:

I have always found it funny that people spend so little comparative time worrying about “interpretations of classical mechanics”. If you believe that Newtonian mechanics is correct, then the world is described by a bunch of real numbers (position and velocity for every particle) evolving under some partial differential equations as a function of another real variable that we call time. So, uh, why is it that when I look through a telescope I actually “see” a planet at the position Newtonian mechanics predicts? There’s nothing in those partial differential equations that talks about the sensation of seeing. Of course, people did worry about these kinds of things a few hundred years ago, but I don’t see any way that quantum mechanics has worse interpretation problems than Newtown mechanics.

20. Why is P vs NP important? Says:

Is it ok to state no one can state for sure QM *can* succeed in all experiments it is needed to succeed?

21. Scott Says:

Shmi #15: No, as I said in the post, at most I fancy myself on the path to quantum Enlightenment. I’m not going to make the naming mistake that my friends the Rationalists did, where you can’t even use your chosen name without basically daring the world to deflate your pretensions and take you down a peg. 🙂

22. Scott Says:

Daniel Harlow #16: I mean, I’ve talked to self-described Everettians with a variety of views on the origin of the (apparent?) probabilities. Some say Everett already solved the problem in 1957 with his measure argument, others admit that he didn’t but say that Deutsch-Wallace or Zurek or Sebens-Carroll or someone else solved it later, and still others say it remains unsolved. Meanwhile, from my lofty Zen remove, I’m happy to admit all these approaches, and others yet to be discovered, as supplying complementary insights about why, while unitary QM might not logically entail probabilities governed by the Born rule, it certainly yearns for them… 🙂

23. Scott Says:

J #17: I completely agree, seeing the different interpretations as just crutches on the road to Enlightenment is profoundly compatible with Peter Shor’s advice to be an Everettian on Monday, a Bohmian on Tuesday, etc. etc. if that’s what helps the most with your research.

No, sorry, I know Paul Kwiat pretty well but haven’t read the piece by him that you linked.

24. Maggie Says:

I have never understood the appeal of many worlds. It hardly seems natural for the universe to be splitting and interfering and recombining. Copenhagen makes a lot of sense when I’m playing with lasers, nonlinear crystals, and single photon detectors. I’m afraid I’m missing something, though, being a naive experimentalist.

25. Greg Kuperberg Says:

Ryan #14 – Your comment is actually very close to my real thinking on this. I think that absolutely nothing is wrong with your model, and that is a very good mathematical framework for interpreting quantum mechanics as “neo-Copenhagen quantum Bayesianism”. But, two things. (1) it is even better to construct this model as the category of CPTP maps between von Neumann algebras in general, rather than just (the operators on) Hilbert spaces. Then all of classical probability (with a Bayesian slant) becomes a natural subworld (or rigorously speaking, a full subcategory) of quantum probability. (2) On the other hand, even though I like this model very much, in another sense it can’t fully supplant Hilbert spaces and unitary (or subunitary) operators. Namely, you can always dilate all of the TPCPs to unitary operators after all, and in QCQI and QM that’s what you often do.

If Scott has gotten my attention here with blanket flattery (I think that labeling it as such makes it more reasonable), then the one thing that I would add is that I prefer to evaluate QM interpretations through the lens of pedagogy and comprehension rather than through the lens of philosophy. I.e., an interpretation is good if it helps you understand the material. From this viewpoint, many worlds is not all that enlightening in my opinion, except maybe as an intuitive counterpart to Feynman path summation, if you want to think of each path in the sum as a trajectory of the world. Quantum Bayesianism or neo-Copenhagen, on the other hand, is really useful in my opinion for understanding QCQI in particular.

I will just concede the irony that cosmological many worlds could be an inevitable part of inflationary cosmology. Cosmologists like many-worlds thinking for their own reasons, and at least the cosmological version might be a lot more useful than the quantum version.

26. Scott Says:

matt #19: Of course people did talk about the mind-body problem, which is a pretty close fit to your “problem of the interpretation of classical physics,” at least since Democritus in 400BC. My position—the Zen position—is indeed that, once all the other stuff has been cleared away, once decoherence and pointer bases and so forth are understood, what remains of the problem of the interpretation of QM is basically just a quantum version of the mind-body problem. (To wit: “but is there really anybody home in the other branches of the wavefunction? How could we possibly know, if we don’t know why anybody is home in this branch?”) However, like with much else in QM, even when the final destination is vaguely familiar from the classical case, it takes a lot longer to get there than it did classically!

27. Scott Says:

Why is P vs NP important? #20:

Is it ok to state no one can state for sure QM *can* succeed in all experiments it is needed to succeed?

Sorry, I couldn’t parse your question. (I don’t mean “why is P vs. NP important?”—for that see my survey—I mean your other question!)

28. Scott Says:

Maggie #24: The appeal is the conceptual simplicity.

You don’t have to postulate that “worlds are splitting” (and sometimes “recombining,” but only if they’re extremely close together), you can derive all that from the Schrödinger equation, as applied not just to lasers but to everything in the universe including yourself. And that leads, pretty naturally, to a picture where your experience in the lab is exactly as it would’ve been under the Copenhagen interpretation, except now you don’t need to introduce “measurement” as an unexplained primitive that stands apart from other physical laws; instead it’s just a logical consequence of the interactions among the photons, the photodetectors, and ultimately your retina and your brain.

The many-worlders love to use the example of Copernicus and Galileo, who introduced a bizarre and extravagant hypothesis—that the earth, far from being stationary, circles the sun at unimaginable speed—but then argued that, starting from that hypothesis, you could derive that our experience on earth would be just like it actually is, and this time with no need to postulate epicycles.

There are sophisticated arguments against many-worlds, but I think that a good first step along the Zen path, for someone who already knows QM, would be to understand the strength of the case in many-worlds’ favor.

29. Zvi Says:

Thanks for the blog Scott. Feel well. I don’t agree with you though on this one, I think we have simply failed miserably to explain how quantum mechanics models a world in which cats are always either alive or dead. We should acknowledge our failure honestly.

To me your proposal is like trying to reach a zen interpretation of why heavy and light things appear to fall at the same speed even though we know from Aristotle that they don’t, or like Einstein trying to reach a Zen-like acceptance that the Michelson–Morley experiment is consistent with the aether.

We don’t need any interpretation or Zen, we need a theory which successfully models the world we see both microscopically and macroscopically in one consistent theory. We shouldn’t let ourselves off the hook by convincing ourselves that QM has achieved this.

By the way years ago I wrote my Masters thesis on “The Nine Lives of Schroedinger’s Cat” (https://arxiv.org/abs/quant-ph/9501014) but honestly all the nine interpretations I surveyed completely fail to deliver this.

Keep up the great writing

30. Raoul Ohio Says:

“is what it is, ain’t what it ain’t” in other contexts.

When wearing a physicist hat, it is natural to try to figure out the WHY for everything. Usually this can be done, at least roughly. When all attempts fail, you have to consider the possibility of “is what it is, ain’t what it ain’t”.

This comes up a lot in cosmology. For example, many state that the question of why there is more matter than antimatter is the deepest question in play, and propose one wacky idea after another to explain the discrepancy.

While an obvious first guess on the issue is “sure, there is probably just as much antimatter as matter”, this possibility appears to be in severe conflict with observation. The reasonable next thought is “darn! Looks like I was wrong, there seems to be more matter than antimatter– who’da thunk it?”.

Apparently lots of cosmologists reject this view, and insist that matter and antimatter started off the same, until something rigged the system.

Well, maybe, but I am not convinced that anyone has an accurate score for the state of the universe at the t = 0 singularity, and think “what you see is what we got” is the simpler answer.

31. gentzen Says:

AtomZ #2:

Scott: considering your third paragraph, have you ever given any thought to quantum logics? If we go down the rabbit hole of QM interpretations often being the result of our attempts to impart classical logic where it doesn’t belong, who’s to say that all of logic itself might not really be empirically based on our classical, macroscopic experience? Are things like paraconsistent logics or other multi-valued logics worth considering?

If you are interested in how the logic of quantum mechanics differs from classical logic, then looking at the Consistent Histories approach to quantum mechanics would be a good way to learn about that subject. If you don’t like reading SEP articles, a short technical summary including many formulas can be found in section “Consistent histories interpretations” of Maximilian Schlosshauer’s article Decoherence, the measurement problem, and interpretations of quantum mechanics.

I say this because the consistent histories approach stays as close as possible to classical logic, and thereby makes it very clear in which respect quantum logic is actually different. (You may think that I am just advocating an interpretation here, but my impression is that it is not a complete interpretation anyway, and its initial proponents were honest about this.)

32. Ivo Says:

That famous quote by Feynman about not understanding QM has done an enormous amount of damage, convincing smart people they didn’t understand QM and less smart people they understand exactly as much, or more, about it as those smart people.

Feynman was simply wrong: we understand QM perfectly fine, thank-you-very-much. Not being willing to accept its predictions or attempting to interpret them as seems ‘intuitive’ or ‘similar to something else I already understand’ is the problem. QM is like nothing else: it’s its own thing, much like David S. Pumpkins.

33. B. Says:

Regarding 5., can you say a little bit more or give pointers? I guess my two questions are: 1. How can we describe measurements as some kind of entanglement between the observer and the observed system? 2. What makes the measurements irreversible in practice if they are in theory?

34. LK2 Says:

Something I was never able to clarify: do we really need the wavefunction? It seems to me that in the Heisenberg formulation we have only operators (matrices..) and no wavefunctions. So: does it make sense the debate about what the WF is? The two approaches look to me like when in probability theory you talk about probability densities or their moments: it is just a matter of convenience. Or not? I’m I missing something and the WF is really a necessity in the QM math ?
BTW, great post Scott.

35. bertgoz Says:

So…just wait for an AI running on a quantum computer to tell us how QM feels?

36. murmur Says:

I feel that decoherence has some fundamental relation with the accelerated cosmic expansion. Some galaxies are receding at faster than light speed from us. So after a certain amount of time has elapsed two different branches of the wavefunction cannot be brought into superposition even in principle. Scott, I think you have written along these lines before. If you can tell me your thoughts about this again that’ll be nice. (Incidentally, this seems similar to a version of Mach’s principle that the distant stars are responsible for inertia at earth. I don’t know if this is just a superficial similarity or if there’s a deeper connection.)

37. Mateus Araújo Says:

Daniel Harlow #16: I’m afraid you are factually incorrect on this point, about true many-worlders saying that there is nothing to quantum mechanics beyond the Schrödinger equation. I’m a many-worlder myself, although I don’t know whether I qualify as a “true” one. I hope the definition is not like the Scotsman’s.

The best counterexample are the worlds themselves. They are not in the Schrödinger equation, they are not part of the fundamental ontology, but rather of the emergent ontology. We do derive their existence from complex quantum systems interacting and undergoing decoherence, which leads to these quasi-independently evolving quasi-classical domains, but at the end of the day they are just a tool we use to make sense of the mind-bogglingly complex universal wave function. Nevertheless, one cannot imagine doing quantum mechanics, or Many-Worlds for that matter, without the worlds.

Ditto for probability. It’s not in the Schrödinger equation, it’s not part of the fundamental ontology, and anybody that worked on this deriving the Born rule business can tell you that it *cannot* be derived from the unitary part of quantum mechanics alone. Nevertheless, it’s part of the emergent ontology, and it’s unthinkable to do quantum mechanics without it.

38. Meow Says:

Scott, Is there a possibility of any future meaningful application or theory about quantum neural networks (QNNs) or just another hot word? I don’t understand why this combination is necessary.. what do you think? And how about Microsoft’s Topological approach to Quantum Computing? I am lost in the maze, can’t see the big picture

39. Mateus Araújo Says:

Ryan O’Donnell #14: The problem is that measurements are not CPTP maps, as the collapse of the wavefunction is not even linear. In any case, the debate on interpretations of quantum mechanics is not about achieving a unified representation for your transformations, the debate is about what is going on in a measurement. Is it a unitary (or a CPTP map, it doesn’t matter), or a collapse? It can’t be both!

40. gentzen Says:

Partisan #3:

I tend to see arguments about the ‘right’ interpretation of QM as basically equivalent to arguing about whether we should think of ourselves as stationary or as rotating about the Earth’s axis (or perhaps also about the sun-Earth center of mass and the Milky way and….). Except that for some reason, in the first case we all agree that you can choose whatever perspective you want, as long as you’re consistent, and that different perspectives might be more or less useful in different situations, while in the second case we’re supposed to pick one and loyally defend it from any who would besmirch its honor.

Even your example with the Earth is not as clear cut as it may appear to the layman (i.e. myself some month ago, before I dived a bit into the most basic introductions to cosmology and general relativity). It seems that we are moving relative to the universe at the speed of ~ 600 km/s. This is the speed of our galaxy relative to the cosmic microwave background. Maybe you object that me defending the “cosmic microwave background rest frame” as the preferred frame of reference is the sort of pointless debate your example tried to highlight.

However, there is something deeper going on. Your argument just looks at the physical laws, and ignores the importance of the initial state. The big bang at the beginning of time (of our universe) is that initial state in our case. It is very special and defines a preferred direction of time, namely away from that initial singularity. An the spatial directions “orthogonal” to it are the preferred rest frame, i.e. the CMB rest frame in our case.

41. zhaphod Says:

Shut up “while” calculating seems like a sane position to have.

42. Andrei Says:

Scott,

I would like to know your take on the EPR-Bohm experiment (a pair of entangled particles are simultaneously measured by two distant detectors on the same axis). QM predicts and experiments confirm that a perfect anticorrelation (if using spin-entangled electrons) will be observed.

What do you think happens here? Are the measured spins predetermined or not? If not how do you think the results are anticorrelated?

Thanks!

43. Mitchell Porter Says:

I would distinguish between “how to understand quantum mechanics” and “how to think about reality, given the success of quantum mechanics”.

How to understand quantum mechanics – and I mean understand it as a physical theory, not “how to understand the math used in quantum mechanics”:

The main thing to emphasize is that in the theory, the observables are what is physically real, and that the theory is a recipe for predicting the values of observables at a later time, given the values of other observables at an earlier time.

How to think about reality, given the success of quantum mechanics:

This is where I think pluralism of viewpoint is appropriate; but I have a different goal than our host.

For Scott, the contending “interpretations” are apparently useful insofar as they ultimately enhance one’s understanding of quantum mechanics per se, by highlighting different aspects of the theory and its workings.

For me, they are of interest as contending post-quantum theories of reality, inspired by the success of quantum mechanics. (And it’s interesting that they are so ontologically diverse: Everett’s multiverse, Bohm’s nonlocality, Cramer’s time loops, and so on.)

Perhaps this is a difference between a physics outlook and an engineering outlook.

44. Danylo Yakymenko Says:

I don’t agree that QM doesn’t need interpretation.

In the classical world we all agree that there is an objective reality. But QM heavily undermines this belief.

As an analogy, before Einstein we thought that there is an objective, independent time. This turned out to be false.
But we have a new “interpretation” of time. We have spacetime – a very concrete explanation of what is non-objective time.

We need something similar for QM. And I don’t think Everett’s relative state formulation is there yet.

45. Leo Says:

I don’t understand this post even after pondering the FAQ. I can think of three possible interpretations (ha):

One, maybe you mean that the really confusing questions, such as “If I open the box and see the cat is alive, should I be relieved, or remain sad because half of the cat’s measure has died?”, lack answers within QM. They are interesting questions, but we haven’t yet discovered what kind of theoretical framework can answer them. Check back in a couple centuries.

Two, maybe you mean that the questions do have QMal answers, but that they cannot be phrased using only ordinary concepts. “Is the baseball game REALLY a baseball game, or just a meaningless assemblage of particles?” has a satisfactory answer, but not a simple one: it requires an explanation of reductionism. Likewise, “Is the train moving, or the platform?” requires an explanation of Newtonian mechanics and frames of reference.

If this is what you mean, then I’d expect that I know enough QM to see it leads to such answers, even if they’re far ahead of me. Instead, I just see more calculation and engineering. This could be explained as a flaw in my education. I would also expect to see many more attempts to popularise the enlightened answers.

Third, maybe you mean that the questions are meaningless and people should stop asking them. In this case, I would call this the actions of a monster bent on snuffing out holy curiosity, and demand that thou be declared anathema by Bell, textbook, and standard candle, until such time as thou repentest thy sins and thine heresies.

Fourth, maybe you mean something else entirely.

46. Boogabaz Says:

Scott #12: To be more precise, Moldbug has a take on the history of political ideology (call it political history Zen) that’s analogous to your take on QM interpretations. To understand the war of the roses, you have to be able to think like a Lancastrian- to be able to *be* a Lancastrian, able to walk into one of their parties and make small talk- and equally be a Yorkist.
You have to encapsulate both of these sorts of people, both of these worldviews- be able to emulate them, come up with all the same predictions and results they would- and then transcend them.

This is easy to do for something that happened 500 years ago; most people find it impossible to apply to the time they live in (especially the recent past!), because we have a module in our brains screaming at us that becoming able to emulate and encapsulate the Enemy is a Really Bad Idea for laypeople (who are liable to get themselves lynched by “their own side”); understanding the 20th century like a historian understands the Roses-era house of Lancaster is a bad idea unless you can keep your mouth shut. (It might be safe in another hundred years.) However, as with many forbidden, dangerous arts, if you can do it without destroying yourself, you can attain enlightenment. To find the prima materia, you must first wade through filth.

Your post here comes across as a similar project. Understand all ‘interpretations’ of QM, and thus transcend them. Given the existence of political-history Zen, quantum mechanical Zen, and programming Zen (see the Tao of Programming, very funny: https://www.mit.edu/~xela/tao.html) there should be Zen-paths for all fields. Cool stuff.

47. Why is P vs NP so important? Says:

Comment#27:

I meant what if bell inequality turns not violated 1. in a multiparty communication or crypto setting? 2. Or if the entangled particle is is in a galaxy far away? 3. Or the entangled particle is in a critical system (of say bosons) where one particle spinning in a different direction collapses the state of the system by no direct action on the system but by a measurement outside the system?

The survey gives lot of importance to P not NP which is false. I do not think one can identify a proof of poincare conjecture unless you coded up all the constraints in the clauses. What is unclear is unlike meaningless factoring problems which are simple FO statements which could be ac0[6] finding a proof of a theorem is a different matter and identifying the grammar of the mathematics involved perhaps reveals the proof directly through a derivation in the grammar (I do not see how sat in O(n) time can speed up getting the constraints in the first place).

48. Boogabaz Says:

(Excerpts from the Tao of Programming:)

BOOK 1, SECTION 2:

The Tao gave birth to machine language. Machine language gave birth to the assembler.

The assembler gave birth to the compiler. Now there are ten thousand languages.

Each language has its purpose, however humble. Each language expresses the Yin and Yang of software. Each language has its place within the Tao.

But do not program in COBOL if you can avoid it.

BOOK 2, SECTION 4:

A novice asked the Master: “Here is a programmer that never designs, documents or tests his programs. Yet all who know him consider him one of the best programmers in the world. Why is this?”

The Master replied: “That programmer has mastered the Tao. He has gone beyond the need for design; he does not become angry when the system crashes, but accepts the universe without concern. He has gone beyond the need for documentation; he no longer cares if anyone else sees his code. He has gone beyond the need for testing; each of his programs are perfect within themselves, serene and elegant, their purpose self-evident. Truly, he has entered the mystery of Tao.”

49. gentzen Says:

Scott, you wrote: ” Today, at last, I’ll emerge from the shadows and tell you precisely where I stand.” And then you mentioned Sidney Coleman, which is a sufficient excuse for me to also emerge from the shadows. My goal is more modest, I will try to state my opinion (and thoughts) on an old question implicitly raised by Coleman, and explicitly raised by Scott Aaronson (i.e. you) a long time ago.

Scott Aaronson described the difficult question as making sense of “Many Worlds minus the Many Worlds” in an old comment (2011):

Tim Maudlin #6: The view that I take Banks to be defending here is actually one I’ve found extremely common among physicists, so maybe it would be worth philosophers trying to understand it sympathetically and seeing how much sense they can make of it. I like to think of this view as “Many Worlds minus the Many Worlds” — i.e., many worlds without calling it that, or even acknowledging a need to discuss that apparent implication of what you’re saying.

My opinion is that Arnold Neumaier had a valid answer already in 2007. It is an “inconvenient” answer (in my opinion) that implicitly invokes (a valid form of emergent) superdeterminism (in the sense that the assumption of statistical independence of the initial conditions from the measurement setup are violated), but still can’t prevent Many Worlds completely. His answer only seems to succeed to prevent Many Worlds for our world today, but doesn’t seem to exclude the possibility that the world initially splitted many times before our current macroscopic world emerged.

Since this is your blog, and I refer to your old comment, let me first continue to quote more from it (also because the Internet Archive Wayback Machine is sometimes a bit slow):

On the one hand, you view a measurement as just an ordinary, unitary physical interaction, albeit one that “looks and smells measurement-like”—i.e., that exponentially suppresses the off-diagonal terms in a suitable density matrix, because of decoherence theory. On the other hand, you view the reduction of the state vector as completely analogous to ordinary Bayesian conditioning. What are you conditioning on, in this case? Well, presumably, which block of the now-block-diagonal density matrix you’re now “in”! So basically, you get to play a double game: treating the state vector “realistically” for the purpose of understanding unitary evolution (including the entanglement of the system and measuring device that causes decoherence), but then “ontologically” for the final step of the Born rule and state-vector reduction.

So one part of the question is to make it precise how the “correlation without correlata” character of QM is coped with, i.e. how to make sense of conditioning in a statistical interpretation. Then you go on to describe how the existing answers are unsatisfactory:

The gap—i.e., the obvious disanalogies between what we’re doing now and ordinary Bayesian conditioning—are bridged over by
(1) stressing just how drastically the macroscopic interference terms are suppressed, and therefore how unlikely it is that we’ll ever run into problems in practice, and
(2) saying “well, this is quantum mechanics, a perfectly-natural non-commutative generalization of ordinary Bayesian probability theory. If you find it unintuitive, then the problem is with your intuition.”

After all those words, Neumaier’s answer is basically guaranteed to disappoint. I will probably have to explain later why his answer is not disappointing to me. Here is the translation of the relevant part of Neumaier’s “inconvenient” answer:

You cannot superpose entire universes. In any case, I don’t see how that should be prepared. There is only _one_ state in the thermal interpretation, that of the entire universe. Everything else is derivatives.

The superposition principle only applies to systems that are so small that they can be produced in practically any number and manipulated within this universe. Macroscopic systems are definitely no longer one of them!

This limitation brings down Wigner’s classic argument, which proves the incompatibility of unrestricted unitarity, the unrestricted superposition principle and the collapse of the state during a measurement.

The implicit superdeterminism in this argument is that whenever we prepare a small system and measure it, the state of the measurement device together with the rest of the universe will be such that the measurement device ends up in a valid (i.e. non-superposed, neither coherent nor incoherent) macroscopic state. It is a valid form of emergent superdeterminism, because the macroscopic observables emerged such that they will never encounter superpositions from the evolution of the _one_ state of the universe.

There is also an (old) official English translation of Neumaier’s argument: What becomes of the superposition principle?. However, it contains a mistranslation, namely “This restriction brings into play Wigner’s classic argument ” would have been better translated as “This breaks Wigner’s argument” or as “This brings down Wigner’s argument”.

50. Mateus Araújo Says:

Scott, about your point 2, you are confusing the ontic/epistemic distinction with the fundamental ontology/emergent ontology one. We are almost always talking about emergent ontology, be it the quantum state of an atom, of a qubit, or a book, a basketball game, or the Moon. Just because they’re not fundamental entities it doesn’t make them any less real.

The crucial distinction between ontic versus epistemic entities is that the latter don’t have to respect the laws of physics. A basketball game that’s played only inside my head can have players moving faster than light, or playing with a ball made of solid iron.

And this is why physicists are interested in the ontic/epistemic distinction in the first place. If quantum states are epistemic, the unphysical collapse is not a problem. If they are ontic, well, we can’t have them collapsing, can we?

I think this is a rather substantive question, and you dodged it.

51. Scott Says:

David Karger #13: No, I haven’t read Greg Bear’s Moving Mars. If it’s QM-related sci-fi that’s any good, it would be one of very few examples that I know about!

52. Scott Says:

AtomZ #2:

Scott: considering your third paragraph, have you ever given any thought to quantum logics?

Sorry I forgot to answer this earlier. Whatever its interest as a technical topic, as an account of QM I’ve always found quantum logic to obscure 10,000x more than it illuminates. It’s (some of) the rules of probability, not the rules of logic, that I see QM as challenging. And indeed, quantum logic severely crippled itself from the beginning by throwing away the entire probabilistic part of QM, and looking only at the lattice of subspaces of Hilbert space (i.e., probability 0 and 1 statements)! One might even say that modern quantum information theory is what quantum logic should have been; maybe von Neumann would’ve invented it as well if he’d had more time. 🙂

More fundamentally, though

(1) I don’t see a need to tinker with the rules of logic, given that we have several ways to talk about QM (Everett’s, for one) that leave those rules totally untouched, and

(2) even if we did tinker with them, I don’t see how it helps in the slightest to answer the questions people want answered (“do I have equally-real doppelgangers in other branches of the universal wavefunction, or not?”) It’s not like I have definite answers to those questions either, but at least I can talk about them clearly!

53. Scott Says:

B. #33:

Regarding 5., can you say a little bit more or give pointers? I guess my two questions are: 1. How can we describe measurements as some kind of entanglement between the observer and the observed system? 2. What makes the measurements irreversible in practice if they are in theory?

Try Sean Carroll’s Something Deeply Hidden, or my own Quantum Computing Since Democritus or undergrad lecture notes.

Briefly, though, the Everettian/decoherentist account of measuring a qubit is that it’s an ordinary unitary interaction, which maps an unentangled state like

(a|0⟩ + b|1⟩) |You⟩

to an entangled state like

a |0⟩ |You0⟩ + b |1⟩ |You1⟩,

where |You⟩ means the state of your brain (well, and also your whole environment), and |You0⟩ and |You1⟩ mean you having registered the measurement outcomes 0 or 1 respectively.

This is reversible in principle because all unitary transformations are. It’s irreversible in practice essentially because of the Second Law of Thermodynamics (or to spell it out: in order to “erase” the entanglement, you’d have to painstakingly erase it from every single subatomic particle in your brain and the wider environment that recorded whether the qubit was 0 or 1).

54. Scott Says:

LK2 #34:

Something I was never able to clarify: do we really need the wavefunction?

I mean, you could replace wavefunctions by operators a la Heisenberg, density matrices a la von Neumann, or (getting rid of the “states” entirely) sums over histories a la Feynman, but you’d still face basically the same questions in a slightly different mathematical context!

55. Scott Says:

LK2 #35:

So…just wait for an AI running on a quantum computer to tell us how QM feels?

Sure, we could try that, but what if the quantum AI couldn’t explain the true nature of its inner experience, any more than I can explain my inner experience to you or vice versa? After all, any verbal account of its experience would just be the outcome of some measurement on its state… 🙂

56. maline Says:

When people say that “Quantum physics needs no interpretation”, I usually take that to mean “Quantum physics is a well-defined system for making predictions; it works well; and that’s all I care about”. Which is true, and fine as a matter of personal interest, but utterly abhorrent as a normative philosophical stance.

But you seems to have something else in mind. You do care about “how nature actually works”, and you think it is well-described by a state vector evolving unitarily. In particular, the idea that decoherence explains why we see a classical world is important to you; you would not be satisfied with just a predictive statement about “what we find when we measure”.

So your whole claim here is that by maintaining a Zen-like attitude, you can avoid the “metaphysical extravagance” of the MWI while maintaining the ecact same physics.

This claim is false. If our own realitily is one out of many similar branches, and somehow or other we have conscious experience, then there is no reason at all to doubt that the same is true for the other branches. You might as well doubt whether any minds other than your own exist. Remember that the people in the other branches (or at least their wavefunction descriptions) have philosophical discussions as deep as ours!

So as soon as you reject the possibility of objective collapse (which you do; this would be an in-principle-measurable physical effect that would set fundamental limits on quantum computing), you are necessarily committed to the full nine yards of MWI. Including the requirement to derive the Born rule in terms of self-locating probability.

57. mjgeddes Says:

So, OK, I admit it, foundations of QM is a tough one even for a post-human like me 😉

I do get the distinct impression that there’s two ultimate principles at work here for explaining the macroscopic world we actually see, quantum mechanics *and* statistical mechanics. And the latter (stat mechanics) hasn’t been emphasized enough.

The heart of the mind-body puzzle is the nature of *time*. Pure mathematics might be completely timeless, but *applied* mathematics (information science) is the opposite, it’s all about events in time, which is more like the nature of our experience as observers. So you have a clear duality between a timeless picture of reality in terms of mathematical spaces and a picture of reality in terms of events (computation).

On the physics side, stat mechanics seems like one bridge , and on the math side, mathematical logic (category theory, type theory) seems like the other bridge. Both stat mechanics and category theory seem like they’re starting to “close the gap” between the timeless and timeful pictures of reality.

58. Scott Says:

maline #56:

If our own realitily is one out of many similar branches, and somehow or other we have conscious experience, then there is no reason at all to doubt that the same is true for the other branches. You might as well doubt whether any minds other than your own exist. Remember that the people in the other branches (or at least their wavefunction descriptions) have philosophical discussions as deep as ours!

Yeah, this is specifically the part that I now reject. I readily grant that people other than me have minds partly because I can see that if I don’t, those people will call me an asshole (we adopt, as Turing put it, “the polite convention that everyone thinks”). 🙂 But other branches of the wavefunction? QM itself explains why we can never talk to them, or they to us! And it seems like too much to take “appearance in some mathematical description of the universe” as a sufficient condition for consciousness, unless you want to go full Tegmark and ascribe consciousness to every conceivable mathematical structure that acts like it’s conscious.

For more: my old post Why Many-Worlds Is Not Like Copernicanism, and my review of Tegmark’s book

59. Craig Gidney Says:

> [bertgoz 35] So…just wait for an AI running on a quantum computer to tell us how QM feels?

Do you expect it to say anything philosophically relevant that you couldn’t predict already? We have the mathematical model that describes its state evolution. For example, you can work out on paper why taking a reversible classical circuit that estimates probabilities and running it in a quantum context results in the circuit’s probability output register containing the number predicted by Born’s rule (or rather an approximation of it that gets better as sample count increases).

60. fred Says:

Scott, about point 6.
In the Schrodinger cat setup, when you open the box and see a dead cat, is the wave function collapse absolute or relative?
Relative in the sense that if you’re yourself in a box, and there’s an outside observer, and that observer opens the box you’re in, can that observer see an alive cat? (I don’t mean observer as conscious being, but simply as a complex system that gets entangled with an observed, simpler, system). What does the Copenhagen interpretation have to say about this?

I have a couple of questions regarding the concept of branch:

– With the Schrodinger cat story, we’re told there are two alternatives – the cat is alive or the cat is dead. Which makes it sounds like there are two branches. But, in reality, the particle that triggers the death of the cat could decay at any moment in time (existing on a continuum or near continuum at the plank constant), so there is a near infinite possibilities of dead cat (all slightly different), no? Does it mean that an infinite number of branches are in superposition?

– Is it correct that the only time we have actual evidence of branching is when the alternatives are indistinguishable, like in the electron slit experiment, and interference is some sort of evidence that the branches are interacting and therefore “exist”?
But when the alternatives aren’t indistinguishable, like in the Schrodinger cat setup, the branches, don’t interfere, and therefore there’s really no evidence of their existence.

– Sabine Hossenfelder (btw, in a recent video, she confirmed that she still favors superdeterminism…) claimed that the many world interpretation doesn’t really solve the measurement problem at all (i.e. the difficulty that it introduces a non-linear event). She said that instead of having one wave collapse to explain, we know end up with multiple wave collapse events, one for each branch. Does that make any sense?

– In the many world interpretation, the notion of probability becomes tricky, i.e. assigning a probability to a branch is kind of meaningless.
E.g. say there are two clear alternatives, alternative A has an associated square amplitude of 99.9999%, alternative B has an associated square amplitude of 0.0001%, but once we admit the existence of both A and B, it really doesn’t matter what the probabilities are. What are your thoughts on this?

61. Gabriel Conroy Says:

I haven’t had a chance to read this yet (but I will eventually), but I just wanted to say congrats on getting the vaccine.

62. Jelmer Renema Says:

@ Scott OP:

I 100% agree with this approach, but I think this way of thinking is much more broadly applicable than to quantum mechanics.

What I think you have re-invented here is essentially the humanist way of dealing with ambiguous, self-contradictory concepts, that deeply involve (as you put it) the philosophical baggage that you yourself bring along. Particularly the bit about how true enlightenment comes from being able to predict a discussion between different viewpoints reminds me very much of how e.g. philosophers are taught to think (at least if the experience of my local environment is representative).

63. fred Says:

Scott #58

“Yeah, this is specifically the part that I now reject. I readily grant that people other than me have minds partly because I can see that if I don’t, those people will call me an asshole (we adopt, as Turing put it, “the polite convention that everyone thinks”). But other branches of the wavefunction? QM itself explains why we can never talk to them”

I don’t see how the notion of multiple consciousness for “you” in different branches is farfetched.
We already have evidence of “independent” consciousness:

– in the dimension of time: there’s the consciousness of you when you were 10 year old, and the consciousness of you as of now. You can also ask how come those don’t bleed into one another.

– in the dimension of space: there’s the consciousness of you, and then the consciousness of your wife. Both existing at once and separate. You can also wonder how come those don’t bleed into one another.
There’s a more “personal” version of this – people who had the connection between the two hemispheres of their brain severed. There are experiments strongly suggesting that those individuals are the seat of two somewhat separated consciousness, which evolve in a way to stay as coherent as possible (because they’re still both share the same body and subjective history, they just have access to a different sub-state of the brain).

– then there’s “branchial wave function” dimension. One separate consciousness for each branch.

Note: above, by “consciousness”, I mean more simply subjective experience (the content of the brain). Since the content of experience is attached to a particular brain, as a system, it’s obvious that the subjective experiences can be different in time (my brain now is different from my brain a year in the past), space (the current state of my brain is different from the state of my wife’s brain) , and branchial dimension (the state of my brain in a branch were Trump died when Air Force One crashed is different from the state of my brain in the branch where I’m writing this).
Personally I suspect that consciousness *is* the formation of memories. Those memories are related to *any* data content (audio, visual, touch, feelings, thoughts, sense of familiarity, sense of space,…). Once two different brains (in time, space, or branchial space) form different memories, they “realize” different identities. Those identities extend as far as the temporal/spatial/branchial interconnections of those memories.

64. Concerned Citizen Says:

> What is a quantum state? It’s a unit vector of complex numbers (or if we’re talking about mixed states, then a trace-1, Hermitian, positive semidefinite matrix), which encodes everything there is to know about a physical system.

This is a very quantum-information-centric view of reality. 🙂 Complex numbers aren’t strictly necessary to write the equations that govern interactions, they just happen to embody one of the symmetries that is observed experimentally. Other times, other algebraic objects (like spinors) happen to have the right properties. Quantum states also don’t have to be unitary – it is equally valid, and sometimes preferable* to say that a quantum state is the equivalence class in projective space of a\psi for all a. What’s the point of all this? I guess it’s just to say that there are a lot of equally correct ways to write the foundations of quantum mechanics, some in the language of groups, or algebra stuff, or partial differential equations, or anything you want, enough so that it becomes obvious that we are speaking in many different languages about an object we observe.

It is not philosophically bothersome that a red fruit on a table in front of someone could be correctly called either an Apple or a Pomme, so I don’t see why it should be philosophically bothersome that there are many different equivalent ways to write quantum mechanics.

* Mildly, if you define it like that, wavefunctions without an L2 norm, like plane waves in infinite space, or any scattering state really, make sense by default instead of being a tacked-on exception in “rigged” Hilbert space. Another reason that’s nice is that it legitimizes the practical fact that you tend to write down a lot of non-normalized wavefunctions while solving a problem, because under linear operations you know you can save normalization to the end.

65. Greg Kuperberg Says:

Mateus #39 – “The problem is that measurements are not CPTP maps”

Mateus, actually measurements certainly are CPTP maps. Suppose $$\mathcal{A}$$ is the algebra of operators of a system, perhaps all of the operators on a Hilbert space $$\mathcal{H}$$. Suppose that $$\mathcal{B}$$ is the algebra of operators of an observer modeled as a second system, for simplicity a commutative algebra if we model the observer as classical. Let $$\mathcal{A}^\Delta$$ denote the space of states on $$\mathcal{A}$$, including the mixed states, in density operator form. Then there is a perfectly valid CPTP map $$\mathcal{M}:\mathcal{A}^\Delta \to (\mathcal{A} \otimes \mathcal{B})^\Delta$$ to model each physically valid measurement. I.e., there is such a CTPT $$\mathcal{M}$$ for each spectral projection of an operator on $$\mathcal{H}$$ or each POVM on $$\mathcal{A}$$ otherwise, if in addition you decide how to encode the measurement outcomes into states of $$\mathcal{B}$$.

Of course you can still obsess over how such an $$\mathcal{M}$$ can happen in the real world, but that’s not any different from analyzing how any other CPTP map can happen.

66. Scott Says:

murmur #36:

I feel that decoherence has some fundamental relation with the accelerated cosmic expansion. Some galaxies are receding at faster than light speed from us. So after a certain amount of time has elapsed two different branches of the wavefunction cannot be brought into superposition even in principle. Scott, I think you have written along these lines before. If you can tell me your thoughts about this again that’ll be nice.

Absolutely! I somehow neglected to mention this in my post, but thinking about the deSitter cosmology of our universe, and how, as Bousso and Susskind pointed out, it leads to decoherence that’s irreversible even in principle (e.g., from light escaping from the earth that will never bounce back) … this was a huge factor that led me down the Zen path over the last decade. For it alerted me to the possibility that, as diametrically opposed as the Everettians and Copenhagenists seem in their rhetoric, there is a sort of synthesis of their views possible. In that synthesis, we don’t give up on unitary evolution applying always and everywhere, but we also say that, without changing any known laws of physics, there is something fundamental and non-emergent that it means for a measurement to have definitely, irreversibly happened and yielded a definite outcome. Namely, it means that the records of that outcome are headed in all directions toward the deSitter horizon of the observable universe at the speed of light.

I wrote more about this in Ghost in the Quantum Turing Machine.

67. Scott Says:

Meow #38:

Scott, Is there a possibility of any future meaningful application or theory about quantum neural networks (QNNs) or just another hot word? I don’t understand why this combination is necessary.. what do you think? And how about Microsoft’s Topological approach to Quantum Computing? I am lost in the maze, can’t see the big picture

I’ve also never seen the point of quantum neural networks, but maybe a point will be discovered in the future.

Topological QC is beautiful theoretically, it’s been extremely influential even for other forms of QC (through, e.g., Kitaev’s surface code), and because of its natural fault-tolerance properties, it’s an exciting approach in itself for further in the future. From the standpoint of today’s hardware, though, the problem is that no one has yet succeeded in creating even one topological qubit! So there’s a huge amount of catching up to do, even just to get to where superconducting QC or trapped ions are now.

And sorry, but please, no more drive-by off-topic questions!

68. Greg Kuperberg Says:

Scott #67 – “From the standpoint of today’s hardware, though, the problem is that no one has yet succeeded in creating even one topological qubit!”

Actually by a different standard, they almost have. The dominant presence of topological QC so far is not through creating a physical material in a topological phase, but rather through using a noisy non-topological QC to simulate topological QC as a software method of quantum fault tolerance. This includes additive surface codes as a special case, and there have been or could easily now be experiments to do that. I think that by now the only thing missing in such a QC demonstration is not even valid quantum error correction or construction of a qubit, but rather quantum error correction with gain. If so, then maybe they have already made a topological qubit via quantum software, only (a) only one in an abelian anyonic phase, not QC-universal; and (b) not a logical qubit that is better than the physical qubits simulating it. You might know more about the details of the status of this than I do. Here I am more arguing how to interpret the status quo, that you can take existing QC to be more topological than you might first think.

69. fred Says:

Scott #55

“Sure, we could try that, but what if the quantum AI couldn’t explain the true nature of its inner experience, any more than I can explain my inner experience to you or vice versa? After all, any verbal account of its experience would just be the outcome of some measurement on its state…”

That’s really at the core of the hard problem, isn’t it?
The fact that we are all discussing the topics of consciousness and subjective experience is some sort of proof that they’re a thing.
On the other hand, such discussions are supposedly entirely driven by the mechanical state of the brain.

There are different possibilities, all of them self-contradicting:

Consciousness isn’t some sort of extra ingredient on top of the mechanical state of the brain, it’s just some side effect of it… but then, how come we’re talking about consciousness as if it’s an actual thing that’s somewhat truly emergent on top of the physical laws? If it were truly emergent, it wouldn’t be explained by the physical laws alone (in the bottom up way of science). So no simulation of a brain would ever come up with a coherent discussion of consciousness – if they would, then consciousness can’t be a special ingredient (to be truly emergent, a phenomena can’t be explained by underlying causes).

Another option is that consciousness is some external ingredient, which influences the state of a brain enough so that this brain would follow a state evolution that’s not entirely explainable by physical laws alone. In other words, if we tried to simulate a human brain, from a given starting state (say, in the middle of a conversation about consciousness), that simulated brain would diverge from the state of the real brain in a way such that the simulated brain would no longer be able to make any sort of coherent conversation about consciousness (“consciousness? what’s that again?… I don’t understand”). So any coherent discussion about consciousness wouldn’t be based off known physical laws. This sounds implausible, but at the same time, discussions on consciousness are very self referential and hard to follow (a bit similar in taste to Godel’s principles).
And if consciousness was some external ingredient, the problem is that this ingredient should not evolve according to the type of physical laws we’re used to, otherwise it could be deconstructed and integrated in the mechanical laws that dictate the evolution of the state of the brain. Which is why some ppl suggest that this external ingredient should be based off some non-computable evolution. But what would we even mean by “non-computable”, non-computable in the sense of the Halting problem? (Penrose seems to think that) or non-computable in a practical sense, e.g. my cousciousness would depend on the entire state of the universe, or a very particular unique evolution since the big-bang (I believe Scott came up with some idea like that). But this only explains why a brain wouldn’t be computable or clonable, it doesn’t explain why there is subjective experience and whether it’s a necessary ingredient or not.

Yet another idea is that consciousness is not needed and has no causal effect on the brain whatsoever. It’s an epiphenomenon.
So, a bit like superdeterminism in QM, there’s some sort of incredible conspiracy at play: consciousness is a thing, yet it has no effect (if it weren’t there it wouldn’t make a difference) and the fact that we do discuss about consciousness as if it were a thing is just some coincidence (and totally explainable by the physical laws that direct the evolution of a brain that’s discussin consciousness). But it’s very hard to make a thing both real and have no causal effect.
Doug Hofstadter suggested that consciousness could be an epiphenomenon, but when there’s a feedback loop, even an epiphenomenon can have some subtle real indirect causal influence, which leads to its own creation. A bit like noting that a camera can’t possible take a direct photo of its own shape (which is behind the lens), but if you point the camera at its own shadow or at a mirror, the shadow or reflected image will influence the state of the camera. Of course this is a problematic explanation because such analogies still rely on physical interactions (shadows and reflected images are physical manifestations that can be included in the model of the camera by extending the boundaries of what constitute a camera). So an epiphenomenon that has a causal effect (no matter how subtle) is not a true epiphenomenon.

70. lewikee Says:

Regarding question 5, why are you assigning the “recording” of an observation to the entanglement of two quantum systems? I would think entangling two quantum systems would merely lead to a more complex wavestate, not some irreversible thing we call observation. The irreversible thing happens later down the line, near some fuzzy boundary we consider as dividing “micro” from “macro”. And true, that isn’t easy to pin down, but surely it doesn’t happen as early as mere entanglement?

71. Aspect Says:

Hello Scott,

I suspect that a certain physicist/blogger would be happy with the bold letters in the third paragraph of your text…

Jokes aside, I have a question that is kind of OT but I wasn’t sure what was the right place to ask since I don’t want to spam your inbox. I’ll go ahead and ask here but feel free to leave it in the queue if you feel that it’s derailing the thread.

In your chapter “Quantum” of QCSD, exercise 6 asks us to prove that any norm preserving map on N dims can be implemented by a continuous motion in N+1 dims. I was wondering, do you have any reference where I can read more about this? I’m aware of certain cases (like in computer graphics) and how it works in those, but after failing my first few tries to prove the general statement I’m here asking for hints/pointers.

72. scottd Says:

Having just stumbled onto Relational Quantum Mechanics (via a brief Wikipedia-surf from Scott’s link to Wigner’s friend), now I’m curious what people here think about it. Briefly: “an interpretation of quantum mechanics which treats the state of a quantum system as being observer-dependent, that is, the state is the relation between the observer and the system”.

As far as I can tell, it’s a relatively new interpretation that is getting some nonzero amount of serious attention; at some level it seems like the logical conclusion/combination of all the things in both special relativity and quantum mechanics that make my brain want to seize up because WHAT ABOUT OBJECTIVE REALITY?!, although at this point I have mostly made peace with the fact that the universe is not obligated to make any sense to me.

73. Bram Cohen Says:

What we really want to know is what’s your opinion of what’s the one true inertial reference frame

74. Scott Says:

Bram Cohen #73: Oh that’s easy! It’s the rest frame of the CMB.

75. Peter Morgan Says:

Mateus Araújo #39: “Is [collapse] a unitary (or a CPTP map, it doesn’t matter), or a collapse?”
I suggest that an alternative way to think about “collapse” is that it isn’t a dynamical process (so not unitary, et cetera): it is instead an algorithm for constructing joint probabilities even for non-commuting self-adjoint operators (so not a collapse of the state, but instead it’s a (nonlinear) reconciliation of measurement incompatibility).

Because we use this kind of reconciliation of incompatibility to construct joint probabilities, we can equally well use quantum non-demolition measurements (as in Tsang&Caves in PRX 2012) and a different state to construct the same joint probabilities. This elementary mathematical reconstruction looks to be quite telling at least for MWI and for consistent histories (and, but this is my thing, so to speak, as part of the connection between quantum mechanics and Koopman’s Hilbert space formalism for classical mechanics.)

There are also contextual circumstances in which we use non-commuting self-adjoint operators to construct probability distributions that do not admit joint probabilities, as for experiments in which Bell inequalities are violated. In such cases, we of course do not use such reconciliation.

FWIW, I describe the mathematics in more detail in https://arxiv.org/abs/2101.10931. It’s a small but telling shift of perspective.

76. Mateus Araújo Says:

Greg Kuperberg #65: You haven’t specified what the CPTP map *is*, only where it’s acting, so I don’t know what you’re talking about. Do you have in mind something like $$\rho \mapsto \sum_i \Pi_i \rho \Pi_i \otimes |i\rangle\langle i|$$? That’s a CPTP map alright, but it represents a Many-Worlds measurement. I don’t think this is Ryan O’Donnell had in mind, and anyway if you’re doing that you might as well stay with unitary measurements.

What I meant is the usual collapse sort of measurement. It’s either non-trace preserving $$\rho \mapsto \Pi_k \rho \Pi_k$$, or non-linear $$\rho \mapsto \Pi_k \rho \Pi_k/\mathrm{tr}(\Pi_k \rho)$$. Either way not a CPTP map.

77. Greg Kuperberg Says:

Mateus – I guess I can’t speak for Ryan, but that is what I had in mind. My point is that a CPTP map like that one — except written slightly differently to express that the system $$\mathcal{B}$$ is commutative — is a perfectly valid CPTP model of a measurement. You can call it “many-worlds” if you like, but it isn’t any more “many-worlds” than any probability distribution on a set is many-worlds. Would you say in general that Bayesians are “many-worlders” because they assign probability distributions to objects? I do consider myself a Bayesian and a quantum Bayesian, but as Scott said, I consider many-worlds to be, not exactly wrong, but overwritten and melodramatic. If I say that a coin that has been flipped but not observed is half heads and half tails, then I consider it melodramatic to summon two universes into existence, one in which the coin is heads and one in which it is tails. Among other reasons, because a Gibbs state of a gas would be exp(Avogrado) number of distinct future worlds.

I think what might bother people, and what I think fits the description of what Scott calls “Zen”, is that I feel that I understand the material better if I simply accept non-determinism. I agree with Scott that the residual angst here is close to just the mind-body problem in philosophy of science.

Anyway, for the second part of what you said, it indeed seems the most natural to me to model collapse onto one term as a sub-trace-preserving CP map, as you say a term in the larger sum. After all, it is exactly the same thing as a Bayesian posterior state. Here as well, calling this “collapse” is not wrong, but also melodramatic, because you can perfectly well interpret it as an update of knowledge, just like in Bayes’ formula otherwise. If you specifically mean it not to be an update of knowledge, then you are modelling the effect of (say) me measuring your system without telling you the measured answer. That too, as it sounds like you know, is given by a CPTP map.

78. Scott Says:

Andrei #42:

I would like to know your take on the EPR-Bohm experiment (a pair of entangled particles are simultaneously measured by two distant detectors on the same axis). QM predicts and experiments confirm that a perfect anticorrelation (if using spin-entangled electrons) will be observed.

What do you think happens here? Are the measured spins predetermined or not? If not how do you think the results are anticorrelated?

Why even talk about EPR-Bohm when for 57 years, the Bell inequality has thrown the same issues into much sharper relief? The Bell inequality shows that the measured spins cannot be “predetermined,” under minimal assumptions of no superluminal signalling and no insane cosmic conspiracies.

My short answer as to why the spins are nevertheless anticorrelated is: because QM is true, and because QM straightforwardly predicts exactly the correlations we observe. One can give longer answers, from an MWI perspective or a Copenhagen perspective or a Bohmian perspective or an interpretation-agnostic perspective, but they’re all going to boil down in the end to that short answer.

More broadly, the idea that there are only two possibilities,

(1) local hidden variables or
(2) superluminal signalling,

with nothing intermediate between the two, is a perfect example of the “classical baggage” that the Zen anti-interpretation of quantum mechanics counsels us to discard.

The reality or unreality of the other Everett branches still occasionally keeps me up at night, but I confess that 2-particle entanglement hasn’t kept me up at night for at least ~15 years. This is one of those things that I really believe you can just accept as part of the world, swallow and move on.

79. Mateus Araújo Says:

Greg Kuperberg #77: I think you’re demonstrating very well why Scott’s answer to his point 2 is actually dodging a crucial question.

If you attribute a probability distribution to a coin that has been flipped but you don’t know the result, you’re not being a many-worlder, because the coin is in a definite state, the probability distribution is only in your head.

Interpreting a wavefunction collapse as an update of your knowledge only makes sense if you believe in hidden variables. The physical system was in a definite state all along, the quantum state was only in your head, and your ignorance collapsed upon learning in which state it was.

I’m sure you’re going to tell me that you don’t believe in hidden variables. This is what makes me angry with quantum Bayesians. They want to use the hidden variables solution to the quantum paradoxes, while simultaneously evading the deep problems of hidden-variable theories simply by declaring “no! hidden variables don’t exist!”. You can’t have your cake and eat it too.

80. Gerard Says:

Scott

Regarding point 5 in particular isn’t the problem with reducing measurement to decoherence that the QM formalism from which decoherence is derived already assumes non-unitary measurements/the Born rule ?

Doesn’t that mean that the QM formalisms we have are incomplete and/or circular at some level ?

81. fred Says:

Scott #66

“Namely, it means that the records of that outcome are headed in all directions toward the deSitter horizon of the observable universe at the speed of light.”

Say an observer is falling towards a black hole along with a Schrodinger cat box.
What’s the difference between the observer opening the box right before passing the event horizon and the observer opening the box right after passing the event horizon?

82. Mateus Araújo Says:

Peter Morgan #75: I’m talking about what is going on in the laboratory, not about anybody’s description. An actual photon comes out of an actual laser, passes through a bunch of actual mirrors and actual beam splitters, and ends up in an actual photodetector. I’ve done this experiment personally (well I turned on the laser, an actual experimentalist prepared everything), it is something that actually went on in reality. What happened to this photon? Did it interact unitarily with the photodetectors, or did the photodetectors cause its state to collapse?

83. fred Says:

Scott #78

“and no insane cosmic conspiracies.”

The actual assumption is independence of choice, not the non-existence of insane cosmic conspiracies.

84. Scott Says:

Zvi #29, Danylo Yakymenko #44, Leo #45: Hopefully the following will clarify my position. My feeling is that philosophical analysis has probably already done nearly all it can for the interpretation of QM. I’m not optimistic that any new interpretation, or any new argument for an existing interpretation, is going to come along and make everyone slap their foreheads and exclaim “it’s all so clear now! why didn’t any of us think of that earlier??”

Of course, if some unexpected empirical finding (related to quantum gravity or cosmology?) were to reveal a deeper structure beneath QM, or really any new fact relevant to the interpretation of QM, that would completely change the conversation, but I’m not inclined to put my money on that either!

So to my mind, that leaves only two major categories of question still standing under the “interpretation of QM” umbrella:

(1) Questions like “how can we understand ‘measurement’ as just a special case of unitary interaction among system, measuring apparatus, and environment?” This sort of question can actually be addressed with concrete physics calculations, and I think that to a large extent, decoherence theory has successfully done it over the past half-century.

(2) Questions like Leo’s “If I open the box and see the cat is alive, should I be relieved, or remain sad because half of the cat’s measure has died?” Just like the mind-body problem, these are profound questions, which engage not only metaphysics but also morality and one’s entire attitude toward life. I don’t dismiss them at all. But I feel like by this point, they might be subjects not only for philosophical arguments that endlessly go around in circles, but for poetry and fiction and other imaginative works.

85. Scott Says:

gentzen #49: Your question (was it a question?) is a little too complicated for me to disentangle, but I can certainly tell you that no “solution” involving superdeterminism has ever had the slightest appeal for me.

86. Job Says:

I have a similar zen attitude towards QM interpretations, but maybe for different reasons.

It’s just too speculative, it needs to be more tangible.

Plus, i’ve kind of put it off until QCs are factoring large integers.

87. fred Says:

Usually it’s guru quacks who misappropriate QM concepts to sell us the proper way to understand the mind.

Now, for a change, we got the reverse – actual mind introspection concepts being wrongly applied to sell us the proper way to understand QM.

Scott A. has officially become the Deepak Chopra of QM!

88. Scott Says:

Mateus Araújo #50: Sorry, I still disagree. Why not say that quantum states are multipurpose constructs? There’s the state that I ascribe to the system I’m measuring, which does collapse when I measure it, and then there’s the state that I’d hypothetically ascribe to the entire universe including myself, which never collapses. You could call the first state “more epistemic” and the second “more ontic” if you wanted, but why not ditch those terms (which I always disliked, given how so many concepts have both “ontic” and “epistemic” components) and just talk directly about what properties the two states satisfy?

89. Scott Says:

fred #82: Ironic, then, that the actual working physicists who’ve commented on this post are generally the ones most strongly in favor of my Zen Anti-Interpretation! 🙂

90. Scott Says:

fred #60:

In the Schrodinger cat setup, when you open the box and see a dead cat, is the wave function collapse absolute or relative?

Unless we’re modifying QM, “wavefunction collapse” is always relative to the specific measuring apparatus, observer, and environment that the measured system is interacting with. Even modern proponents of Copenhagen would agree about that.

91. 1Zer0 Says:

I do tend to mostly agree with the arguments presented in the “Zen approach” and would like to add a few things.
I always considered the question which interpretation of QM is true to be a question of mathematical ontology.
In the end, you try to give a meaning to the formalism.
The claims based on the mathematics of QM are the predictions
The claims about the meaning of the mathematics are the interpretations

I wonder why physicists don’t ask question’s about for example the “true interpretation” of a
solution to the Maxwell equation like a Sinus function.
Or what the “true nature” of the Schwarzschild Radius is.

So you got these purely formal statements like for example a solution to the Einstein field equations on the one side, and you can properly calculate (if you are lucky) and predict things with them yet the relation between the formalism and reality is unknown.
Furthermore the relationship between platonism and reality is also hard to understand.

I take Gödel’s approach and make a division between the “platonic objects” in mathematics and the formal systems used to investigate them.
The inability of sufficiently strong formal systems like ZFC to proof for example the continuum hypothesis justifies that division in my eyes. I do believe in quite a lot of “ontologically true” statements in standard math which cannot be proven…. and would love to add them as axioms (although this will obviously not make the formal system decidable).
Formal systems appear to be just a tool to explore the nature of the platonic realm.
Phenomenological Tools like our senses which transmit, somehow, qualiafied information to our mind are quite similar in that regard.

Did physicists by now derive the Schrödinger Equation axiomatically, as demanded by Hilbert’s sixth problem, somehow? I could only understand QM if there were axioms or an axiom schema to derive all derivable theorems.

Thus far I am agnostic about interpretations of QM, though
From my limited understanding of physics Penrose’s Objective Collapse “Interpretation” can be verified, so that’s something to consider; If I had to choose, I choose objective collapse of the wavefunction by gravity.

92. Peter Morgan Says:

Scott #79: “My feeling is that philosophical analysis has probably already done nearly all it can for the interpretation of QM.”
It’s more that the mathematics of Classical Mechanics now includes a realization that the classical analysis of experimental raw data is enough more complicated in its Liouvillian, probabilistic form that CM’s algebra of observables is naturally noncommutative. We can say, loosely, that there are hidden observables that extend CM to what could be called CM+ (in contrast to the almost century-long search for hidden variables that might make QM more like CM). The Classical Mechanics we’ve been saying is underpowered has been a straw man. The difference between QM and CM+ is not a triviality, but there is a much closer relationship than the mathematically crude relationships of quantization and the Correspondence Principle, and there is much to be learned about both QM and CM+ from that relationship. As an aside, the consequences of using Quantum Non-Demolition measurements, as described in a paper by Tsang&Caves in PRX 2012, is not a bad starting point.
There’s a dynamical systems conference in April, for example, https://www.we-heraeus-stiftung.de/veranstaltungen/seminare/2021/koopman-methods-in-classical-and-classical-quantum-mechanics/main/. As to philosophical analysis of this mathematics, to my knowledge there’s precisely one paper in the philosophical literature (Craig Callender’s https://doi.org/10.1007/s11229-014-0582-3) that discusses Koopman’s Hilbert space formalism for CM, in the 3 page subsection 5.2, so there’s some catch-up for philosophers to do.

93. Scott Says:

Greg Kuperberg #25:

I would add is that I prefer to evaluate QM interpretations through the lens of pedagogy and comprehension rather than through the lens of philosophy. I.e., an interpretation is good if it helps you understand the material. From this viewpoint, many worlds is not all that enlightening in my opinion…

I completely agree with you about evaluating interpretations through the lens of pedagogy rather than metaphysics! But in those terms, I believe that I can make the case to you, based on 15 years of teaching quantum information to undergrads, for why many-worlds is extremely enlightening.

Students, I find, are utterly confused by “measurement.” They’re particularly confused when you tell them that, when you apply a CNOT gate from qubit A to qubit B (initially in the |0⟩ state), the effect on qubit A is exactly as if someone had measured qubit A in the {|0⟩,|1⟩} basis. You can prove it, of course, using density matrices and partial trace, but it doesn’t much help: why should it have been that way?

But when you tell them, “look—why not say that qubit A was measured, and qubit B was the one that measured it! what did you think measurement was, anyway? When you measure a qubit, do you think that’s anything other than a giant CNOT gate, with the qubit as the control and your own brain and the entire rest of the environment as the target?”—then suddenly it clicks for them. But if they then start asking a bunch of followup questions, it seems to me that your choices are either to deflect the followup questions with witty remarks, or else (with a resigned sigh, if you must…) start talking Everett to them.

94. Scott Says:

Jelmer Renema #62:

I 100% agree with this approach, but I think this way of thinking is much more broadly applicable than to quantum mechanics.

What I think you have re-invented here is essentially the humanist way of dealing with ambiguous, self-contradictory concepts, that deeply involve (as you put it) the philosophical baggage that you yourself bring along. Particularly the bit about how true enlightenment comes from being able to predict a discussion between different viewpoints reminds me very much of how e.g. philosophers are taught to think (at least if the experience of my local environment is representative).

I’m so glad that you and I “100% agree” on this! At risk of burning the goodwill that we now share: the ability to see the same issue from multiple viewpoints, correctly anticipating and clearly explaining each of them, aiming toward a synthesis (whether explicitly stated or not) that transcends all the viewpoints—this was perhaps the central feature that attracted me to Slate Star Codex, just like it attracted me to so many of my favorite writers of ages past.

95. Scott Says:

lewikee #70:

The irreversible thing happens later down the line, near some fuzzy boundary we consider as dividing “micro” from “macro”. And true, that isn’t easy to pin down, but surely it doesn’t happen as early as mere entanglement?

What makes you so sure? What if the entanglement isn’t merely with you, but with your whole environment, including a sphere of photons flying away from the earth toward the deSitter horizon of the observable universe, which can never again be collected even in principle?

96. Scott Says:

Aspect #71:

In your chapter “Quantum” of QCSD, exercise 6 asks us to prove that any norm preserving map on N dims can be implemented by a continuous motion in N+1 dims. I was wondering, do you have any reference where I can read more about this? I’m aware of certain cases (like in computer graphics) and how it works in those, but after failing my first few tries to prove the general statement I’m here asking for hints/pointers.

Look up the “orthogonal group” O(N) and the “special orthogonal group” SO(N). I believe the fact you want is that O(N) embeds into SO(N+1).

97. Mateus Araújo Says:

Scott #83: Both of your examples are trivially ontic states, the state of the system in your branch of the wavefunction, and the state of the whole wavefunction.

You are so absolutely certain that quantum states are ontic that you can’t even understand what people mean when they say they’re epistemic. Epistemic states are a representation of your ignorance about some underlying reality, a probability distribution over hidden variables. Sometimes people also use epistemic to mean a Copenhagen/QBist point of view, where quantum states are your knowledge or belief about measurement outcomes, without reference to an underlying reality. In either case, the quantum state is always epistemic, always just some kind of belief, reality is never described by a quantum state, if there is something as reality at all.

You might want epistemic to mean something else in order to fit your Zen transcendence of interpretations, but this is a matter of using standard definitions so that people can understand each other.

You’re right, by the way, quantum states are ontic.

98. Steven Says:

Has nobody commented on the fact that Claus Peter Schnorr must be the evil twin of Peter Shor

99. Scott Says:

scottd #72: Part of what inspired the Zen Anti-Interpretation, is what I see as a striking lack of innovation in the QM interpretation field over the past half-century. Indeed, I claim that all “new” interpretations that I’m aware of, are simply linear combinations of interpretations proposed in the 1950s or earlier. For example:

“Relational Quantum Mechanics” = (2*Everett + Copenhagen) / 3

“QBism” = 10000 * (Copenhagen – Everett)

“Transactional Interpretation” = 0

😀

100. fred Says:

Scott #86

“Students, I find, are utterly confused by measurement.[…] But if they then start asking a bunch of followup questions, it seems to me that your choices are either to deflect the followup questions with witty remarks, or else (with a resigned sigh, if you must…) start talking Everett to them.”

I don’t understand why you say this when you also wrote
“This sort of question [the measurement conundrum] can actually be addressed with concrete physics calculations, and I think that to a large extent, decoherence theory has successfully done it over the past half-century.”

The mystery of the wave function of a particle being updated by the measuring device (itself made of particles) such that all superpositions but one disappear, a non-linear process, well, either it’s been solved or it hasn’t. What do you mean by “I think that to a large extent”? There are cases where decoherence doesn’t explain the measurement process? There are cases where “concrete physics calculations” don’t work?

101. Kerem Says:

Scott,

I share your disdain for superdeterminism, but personally feel MWI is just as convoluted because of question 7 you ask above.

It seems nonsensical to stay on that train for as long as die-hard proponents wants us to, as they argue on and on about subtleties such as the basis in which the many worlds keep branching while pulling the wool over our eyes on question 7. With all the intellectual high-brow lectures on how MWI is the simplest and most elegant interpretation while avoiding THE question seems like pure gaslighting and reminds me of Bohr’s boring lectures …

Why can’t we more openly say that there seems no “mechanistic” or satisfying interpretation that makes sense (with all the respect to deBroglie-Bohm for its historical significance) and admit that there is a very good possibility that future physicists/computer scientists will look back at these as valiant but ultimately misinformed attempts?

Of all the card-carrying QM authorities who have license to talk about this you are probably coming closest to making such an admission — and as usual you deliver such a liberating piece here — but I am still left with the feeling that the Zen way still isn’t quite that.

102. Ryan O'Donnell Says:

@Mateus Araújo #76: For what it’s worth, $$\rho \mapsto \sum_i \Pi_i \rho \Pi_i \otimes |i \rangle \langle i|$$ is indeed what I had in mind. To the extent that I followed all the details of what Greg wrote, I pretty much agree with it.

103. Scott Says:

fred #93: The relevant question is whether you care that after the measurement, there’s still that other branch of the wavefunction where you perceived the other outcome, and the quantum state has no little tag on your outcome to indicate that it’s more real than the other one. If you can get yourself not to care about that, then decoherence solves the measurement problem. If you can’t, it doesn’t.

104. Greg Kuperberg Says:

I believe that I can make the case to you, based on 15 years of teaching quantum information to undergrads, for why many-worlds is extremely enlightening.

Scott – I can believe you — in the context of teaching undergrads. The issues at the graduate and postgraduate level are different. I have found that it really help break down the belief barrier, especially at the postgraduate level, to emphasize a picture with density operators, or really density functionals, on an algebra of random variables which may or may not be commutative. (In other words, a state is a dual vector on what is called a “quantum system” of operators, which can be defined without first mentioning a Hilbert space.) Even though this shouldn’t be presented as the only fundamental picture (because it isn’t, Hilbert spaces are fundamental too of course), this density-based picture helps people bridge QM/QCQI with prior intuition about classical probability, and helps establish “Zen” in my own thinking.

But certainly you have to have enough mathematical preparation to appreciate this story. At the same time, experienced mathematicians tend to be suspicious of many-worlds stories, which could equally be included in classical probability, but clearly seem melodramatic and unnecessary in that context. For instance, your example with a quantum CNOT is exactly analogous to a classical CNOT with randomized bits instead.

105. Bepis Says:

This post makes me want to read a fictional story about a society of people that grew up in a world where quantum effects were predominant on the macro level, so intuitive to them. However they later came across (what we call) classical mechanics at a very large scale, and it was very confusing and strange to them. Slowly people created “classical mechanics interpretations” and eventually some great physicists reached a “classical zen interpretation” where they could understand what we call classical mechanics on an intuitive level, after years of dedicated study.

106. Scott Says:

Bepis #98: A classical<->quantum role reversal is fun to think about, but if these beings are able even to talk about their “interpretations of classical mechanics” in books, papers, and so forth, then don’t they have stable, copyable records, suggesting that they already knew the basic properties of classical information? 🙂

107. Tim Converse Says:

Scott #10 —

You said ““Agnosticism” sounds very far from the position I expressed here! It’s not that I don’t know which interpretation is the true one, due to insufficient evidence. Rather, it’s that I feel confident that all of them capture aspects of a situation that can be made familiar through years of working with QM, but can’t (so far as I know) be precisely articulated in words.”

I must be having trouble with too much (scientific) realism, or maybe I am just not at all far down the QM path. But I’ll admit that I do have trouble distinguishing the position from agnosticism, and here’s why.

A believer in wave-function collapse says that only this branch exists and the others don’t exist at all
A believer in many-worlds says that all the branches exist, and this branch is no more real than the others

These two statements seem to be direct negations of each other as statements about whether the other branches actually exist (which, as a (scientific) realist I believe is a reasonable thing to make a statement about, even if we ourselves can’t determine the fact of the matter). For me this has the same flavor as Golda Meir’s famous statement about compromise.

An agnostic view would express uncertainty about which of these statements hold, but that’s not your view, right? Instead you believe that EXISTS and ~EXISTS (as applied to the other branches) both capture aspects of the situation?

If someone said to me about the question of God’s existence: “I am not an agnostic, but instead I believe that both theism and atheism capture aspects of the situation”, I think my response would be “OK, I am not sure what label applies here, but the one thing I’m sure of is that you’re not an atheist” (and would privately label this as a form of agnosticism).

108. Scott Says:

Mateus #97: So, if (as it sounds) I was charitable enough to unwittingly error-correct an obviously hopeless position (quantum states are secretly probability distributions over “ontic states”) to something slightly less hopeless (quantum states “live only in observers’ heads,” like Chris Fuchs says), is that good or bad? 😀

109. Jeremy Says:

Scott #55 –

Is there any toy model of what interference would do to an AI running on a quantum computer? I get that we could in principle calculate its behavior without having a real quantum computer. But I feel like it would illuminating to have a toy model of a simple algorithm which is then interfered, and how it would respond to queries about its past.

110. Scott Says:

Jeremy #109: So you’re effectively asking, about whichever specific quantum algorithm happens to run in this AI, how or why does the algorithm use interference? Damned if I know! Maybe it includes Grover search as a subroutine? 🙂 Or maybe interference between two computational paths was just deliberately set up, so that the AI could experience what it felt like.

111. Dgodric Says:

Scott #56 & #66:

“… Yeah, this is specifically the part that I now reject. I readily grant that people other than me have minds partly because I can see that if I don’t, those people will call me an asshole (we adopt, as Turing put it, “the polite convention that everyone thinks”). 🙂 But other branches of the wavefunction? QM itself explains why we can never talk to them, or they to us! And it seems like too much to take “appearance in some mathematical description of the universe” as a sufficient condition for consciousness, unless you want to go full Tegmark and ascribe consciousness to every conceivable mathematical structure that acts like it’s conscious.”

I think there is a tension between your two comments in 56 and 66 that points to a thought experiment that speaks to the heart of whether the “many” other worlds actually exist. On the one hand it sounds like you take the MWI at face value right up until you decide simply to refrain from answering that question, as simply a moot question given the everlasting non-interaction between those worlds and yours (somewhat mine too, hopefully).

On the other hand, you find the decoupling of future interaction because of the continuous flow across our cosmic event horizon compelling (and I thank you for that insight connecting that with many-worlds issues, that is opening more doors in my head as we speak) evidence of a fundamental kind of irreversibility that changes the picture.

But I would look at that combination of ideas from a different angle, and ask: do you actually not strongly believe that the objects in the space beyond the event horizon have a real existence, despite the fact that we will never again interact? And if so, why is that not basically the same structure of argument for the “existence” of the many worlds; they should be just as real (and no more) in our minds, as objects in THIS world but beyond the horizon. It is just decoupling by event horizon instead of decoupling by decoherence.

In both cases, if we take the past state of the world, and evolve it forwards using the dynamics we believe to be our best truth, it generates a world that matches the world we can see, but also includes other generated regions which are beyond our reach, for interaction and confirmation, things separated from us either by space or branching of the wave function, but which are inevitable outcomes of the evolution of our past according to the dynamics we believe in.

I’ve challenged other physicists with this analogy before, and I haven’t gotten a good answer back yet. How can you justify ascribing more reality to things beyond the cosmic veil than to things beyond the ineffable parting of the ways as decoherence splits the wave function’s symmetry in myriad directions?

Both are generated in our belief systems through the same kind of process, in which we give them the simple reality that they are necessary components of an overall model that has minimal complexity; if we try to get rid of them our description of the world becomes more complex; we would have to add additional rules to exclude them for no purpose other than to deny their reality- add a collapse postulate, or picture a kind of cosmic event horizon firewall, a different(!) kind of firewall that by fiat ends the existence of objects as Hubble tears them from our grasp forever.

112. Scott Says:

Tim Converse #107: I’m going to answer your critique by going positivist on you. There are some things, like Jimmy Hoffa’s body, for which we understand in principle how someone in the future might find them and then persuade everyone else that they exist. For those things, the question of their existence can pretty much only admit the answers “yes” or “no” (or some straightforward intermediate cases, like “here the body was, but it’s burned to the point where it can be only be ID’ed via laboratory analysis”). But, like, branches of the wavefunction besides ours? What if the only way to verify their existence was to become one with God and examine the source code of the cosmic Program? And what if, when you did so, you discovered that the Program computed properties of the other branches on-the-fly whenever needed, but it saved our branch to hard disk and also sent it to the printer? Knowing everything there is to know about the matter, wouldn’t you be just as confused as before about just what it means for a branch to “exist”?

113. Scott Says:

Greg Kuperberg #104:

At the same time, experienced mathematicians tend to be suspicious of many-worlds stories, which could equally be included in classical probability, but clearly seem melodramatic and unnecessary in that context. For instance, your example with a quantum CNOT is exactly analogous to a classical CNOT with randomized bits instead.

I’d say that, in the context of MWI, the relevant difference between quantum and classical probability is precisely this: that even if we knew the state of the universe to be one specific n-bit string undergoing time evolution, we’d still encode our partial knowledge about the state using a probability distribution over n-bit strings, and would never encode it using a superposition over n-bit strings. In other words, there’s this amazingly powerful pin to deflate claimed multiverses, but it only works for classical multiverses and not for quantum ones.

114. Greg Kuperberg Says:

In other words, there’s this amazingly powerful pin to deflate claimed multiverses, but it only works for classical multiverses and not for quantum ones.

That’s true, but in my mind, this refuge of hidden determinism that is there for classical probability isn’t much of a victory in the real world in the following sense. It is established physics that quantum uncertainty is a massive, eternal wellspring of classical uncertainty. The two types of uncertainty can’t meaningfully be separated in the real world. In the case that I don’t know the outcome of an event, but someone else already does, then I can say that there is hidden determinism — in the literal sense that someone else determined the truth but it is hidden from me. For many other uncertain questions, that’s not possible; there clearly can be stochastic “butterfly effects” that ultimately mix quantum uncertainty into many classical future questions. Yet I have not seen many-worlders be eager to impose the many-world viewpoint on things like the 538 election forecast. Of course the real issue is ideas rather than people, but still, I don’t see how to make a fence between classical and quantum uncertainty; I don’t think that it’s possible. On the other hand, I didn’t think that many-world branching for all uncertainty, quantum and classical, was supposed to be the idea.

For me personally, the truth is that I already liked classical Bayesianism even before I appreciated how convenient it is to combine it with quantum Copenhagenism. Yes, classical Bayesianism is more optional, but I don’t see the need to struggle to avoid either one.

115. J Says:

Tim converse #107,
Maybe we should privately label this: A Path Toward Quantum Syncretism.

116. Greg Kuperberg Says:

In the context of MWI, the relevant difference between quantum and classical probability is precisely this: that even if we knew the state of the universe to be one specific n-bit string undergoing a stochastic evolution, we’d still encode our partial knowledge about the state using a probability distribution over n-bit strings, and would never encode it using a superposition over n-bit strings. In other words, there’s this amazingly powerful pin to deflate claimed multiverses, but it only works for classical multiverses and not for quantum ones.

I guess I’m also wondering whether this particular point is still pedagogy, or back to philosophy. Does this explain why undergraduates get confused by internal measurements with CNOTs? Or is it more that they need more practice reasoning with correlation and entanglement?

117. Dgodric Says:

Scott #112: “And what if, when you did so, you discovered that the Program computed properties of the other branches on-the-fly whenever needed, but it saved our branch to hard disk and also sent it to the printer? Knowing everything there is to know about the matter, wouldn’t you be just as confused as before about exactly what it means for a branch to “exist”?”

So this is part of the same conversation, so I continue by saying that I would answer clearly that I don’t think that is confusing, we just give a modest definition to “exists”, which is whether it is naturally reified as an object in the simplest description of reality we can come up with, and the various answers to those questions just naturally follow.

If we learned that the actual world was one that selects out and saves one special path down the branchings of possibility while lazily evaluating the others, then those other paths would not be reified in the physical state of the world in that dynamics, so they would not be real except as a tenuous, possibly-disconnected set of fragments. But that world is a world that is definitely more complex than the one generated by something like MWI, because it has to have all the things that are in the MWI model but also the addition of both the conceptual mechanism to pick out the path to save, and the additional data encoded in the description of that path.

So that is a truly different world than the one that non-collapse interpretations describe, and so it is perfectly fine that in that world we say that the paths that aren’t followed because of the optimization of lazy evaluation are less real; in that world, that would make sense. From that more lofty perspective we can clearly draw the line between the path recorded and the paths realized only as needed, and they do indeed have a practical, measurable difference of reality in that world.

So although that more-complex-to-describe world would from within look just like our world, we should properly want more evidence to believe in it, which could be provided as you described. But the two worlds are not of equivalent complexity, so absent that evidence, the more symmetrical, simple world would seem to be preferred, the one that just gives all paths through the wave function the same matter-of-fact, contingent-on-interaction sort of reality.

I think the thing that properly rankles some MWI ahem “non-proponents” is the overblown interpretation implied by the “world” formulation. A more humble way to put it would be that they aren’t entire universes conjured out of a single interaction, they are propagating, expanding, LOCAL bubbles of exploration of the particular consistent alternative consequences of measurements and choices coming together, entangled, through the dynamics.

Rather than elevating all these other worlds to our own naturally immodest level of reality, perhaps we had better instead join them in the more modest reality assignment just-another-possibility-among-many lazily realized probably in many forms as it bumps along successively into other things.

118. Danylo Yakymenko Says:

Scott #84:

So, you’re saying that we can’t deduce anything remarkably new from the laws of QM that would explain the world (and answer those questions about measurements). I also think so.

But my optimistic position is that we can use our imagination, our inventiveness (and also common sense) to figure out what QM is missing. We shouldn’t just wait until we observe the missing phenomena in nature. It could be too late 🙂

119. Matjaz Leonardis Says:

What do you make of this view:

Before a scientific theory evolves to the point where it can play the role of a worldview – when it can be boiled down to a small set of clear principles and concepts for which one can assert “this is the way the world is” – it usually goes through a phase of being a set of vague, confused but still highly useful/suggestive/explanatory ideas.

The problem is QM was forced to play the role of a worldview too early – it is still in its primordial state of being a set of very useful mathematical models but with the relationship between those models and actual world left very ambiguous and unclear.

The interpretation controversy then results from people trying to make QM play a role it is “not yet ready to play”, adding things that patch it up into something that looks like a worldview in a way that is generally not very convincing.

120. The Erogamer 7.! Says:

“I don’t think it makes sense either,” said Charles within the dream. “How could you tell from inside a universe whether it had quintessence or not? Even if you say Snow White’s universe lacks quintessence, from the inside Snow White has no way of telling that.”

“Hm,” said Sonia. “I suppose that’s a fair point. Well then, let us suppose instead that even though everything exists—or equivalently and more simply, nothing exists—there is nonetheless some mysterious factor by which certain things exist more.”

“Bullshit,” Charles said.

“I suppose we could call it that,” said Sonia.

“No, I mean the whole idea is bullshit,” Charles said. “It sounds like a metaphysics George Orwell would invent as a parody. ‘All animals are real, but some animals are more real than others.’ How could they tell?”

“You claimed yourself that Snow White exists as much as we do,” said Sonia. “Then why don’t we run into her at the corner shop? Why do I find chocolate biscuits there, instead of Snow White? There is a story in which the two of us meet, and the people within that story have no way of knowing themselves to be unreal. And yet I find myself discovering chocolate biscuits instead. Clearly, there is some factor that the possibility containing myself and chocolate biscuits possesses in greater quantity, compared to the less real possibility containing myself and Snow White. Even if we reply to the great question by answering that nothing exists, some zeroes are greater than other zeroes and quantitative ratios may be established between them. Like any other self-observing structure, Snow White finds herself to be exactly as real as herself, a ratio of one to one, and in this sense her existence is locally an absolute. But to say this does not say whether Snow White is more or less encounterable than other things. There is some quantitative degree to which our universe is looking more towards the chocolate biscuits. Some essence of how much something is observed or experienced, which the chocolate biscuits have more of than Snow White. We could call it quintessence, or propensity, or mana, or the blood of God.”

“Or bullshit,” Charles said.

“I suppose that term is as good as any other,” Sonia said. “By whatever name, it is the single, sole, and only truly universal form of money. And we can extend the same logic further. Having postulated the notion of bullshit, it would follow that things are more real only to the extent that they are, in some sense, more bullshit. Or rather, by definition, anything that makes a possibility more encounterable is exactly what we are calling bullshit. Possibilities are experienced by conscious beings in exact proportion to the total bullshit breathing fire into those possibilities. Then to whatever extent a mathematical model is not describing bullshit, it is mere math. Only to the extent that a mathematical structure is about bullshit does that structure form an encounterable part of its universe. It follows that every sapient species, if they investigate the physics of their universe far enough, will eventually find a level at which physics seems solely to describe the arrangement of pure bullshit—some physical quantity whose apparent meaning is making possibilities more encounterable. By your own argument, the presence or absence of bullshit must be falsifiable for anyone inside the universe to notice a difference. Then the eventually-discovered laws will show that variations in bullshit are experimentally observable and cause other events to proceed differently. Indeed, variations in bullshit will be the only causally potent factor. Across every universe with conscious life, any inquisition into physical law, sufficiently advanced, will render everything into bullshit, which is and must be the sole constituent of experienceable causality.”

“Can you simplify that?” said Charles.

“No, but I’ll do it anyways,” said Sonia. “Whatever it is that makes one possibility more encounterable than another, we are calling that bullshit by definition. Then whatever bullshit is, if something isn’t made of it, we won’t encounter it. And if we couldn’t detect something experimentally, we’d have no way of encountering it. So bullshit has to appear to us as a quantity in our equations—in fact, the only thing structured by our equations, because we can’t encounter anything else. ‘By convention there is sweetness, by convention bitterness, by convention color, in reality only atoms and the void.’ And underneath atoms and the void, it’s all just bullshit.”

“Can you simplify that some more?” said Charles. He felt ashamed, but he wasn’t at his mental best while asleep.

“The only thing that truly basic physics equations can be about, is the stuff that breathes fire into equations, the stuff that makes things be more real,” said Sonia. “That’s what I’m taking as my character’s second Element, if the DM lets me get away with that.”

121. Steven Evans Says:

Scott Says:
Comment #85 March 5th, 2021 at 3:14 pm

” I can certainly tell you that no “solution” involving superdeterminism has ever had the slightest appeal for me.”

It’s not about appeal, though, is it? Superdeterminism is not ruled out by the evidence.
For MWI, doesn’t the “breathing fire” suggest the other-branch mes experience consciousness similar to this-branch me, for example? Based on the evidence, as superdeterminism isn’t ruled out, MWI may just be providing a blue-print for other-branch mes, but there may be no implementation as MWI may not be physical.
(Maybe another-branch me might explain themselves better 😉

122. bertgoz Says:

Craig #59: I expect to gain some insights on the measurement problem, i.e. whether there are parallel branches and how it “feels” when your consciousness in each branch merge with all the others. This assumes that consciousness is not a phenomenon dependent on decoherence.

123. Peter Morgan Says:

Mateus Araújo #82: “I’m talking about what is going on in the laboratory, not about anybody’s description. An actual photon comes out of an actual laser, passes through a bunch of actual mirrors and actual beam splitters, and ends up in an actual photodetector.”
I would prefer not to mention photons or their properties in a description of what is going on in a laboratory until we are certain that saying “photon” will introduce no inconsistencies whatsoever. I hope we can do that for a moment, but because it will be somewhat unfamiliar I will have to be a little long-winded.
As you say, you “turned on the laser”. The experimentalist had designed and built (or bought off the shelf) the electronic device that drives the signal that hardware and software monitored for a thermodynamic transition that we would call an “event”. Whenever an event happens, the hardware and software acquires some number of bits from clock hardware and records those bits as the time of the event. While the laser is off, the hardware records “dark rate” event statistics. When you turned on the laser to drive the electromagnetic field contained by the apparatus, the event statistics changed. Had the “bunch of actual mirrors and actual beam splitters” been set up slightly differently, the event statistics would have been slightly different.
Once we had designed and built hardware that records such event timings, we discovered that we can’t control individual event timings at all, but we can more-or-less engineer statistics. So, we apply systematic statistical analysis to event timings out of an experiment. Any combination of algorithms whatsoever is acceptable in the search for systematic measurement results that we can control, which includes incompatible measurements that in some cases cannot be modeled by commuting operators, so the classical physicist introduces the methods of classical signal analysis, which fortunately includes fourier analysis, Wigner functions, and Koopman’s Hilbert space formalism for classical mechanics.
This may be too philosophical for other people here, but let’s call it electromagnetic engineering and talk to the experimentalists. Instead of asking “What happened to this photon? Did it interact unitarily with the photodetectors, or did the photodetectors cause its state to collapse?” we could ask instead “What happened to the electromagnetic field that is driven by the laser and that drives the electronic device that drives the signal that … results in an “event” being recorded?” We could, for example, ask the experimentalist to record the signal level at GigaHz rates, so that we can have a closer look at what’s happening to the EM field in the apparatus (which is varying at PetaHz optical rates and QFT would say that quantum and thermal fluctuations are present at all frequencies, but we do what we can) instead of watching the signal line for events at MegaHz rates. We could introduce more devices and move them around so that we can get slightly closer to tomographic knowledge of the EM field state as it is guided by the apparatus (that’s infinite-dimensional, so, yeah, the idea of close is problematic, but, again, we do what we can and we work with an ansatz that is as unrestrictive as possible, until we have an ansatz for which there are no inconsistencies).
Once we discuss quantum fields (and random fields), I have found the question of “collapse” of the state to look somewhat different, insofar as for the experimentalist everything is about comparing probabilities with statistics, and that sometimes requires reconciliation of measurement incompatibilities in the theoretical models. That isn’t to dismiss what happens between what the experimentalist knows from actual results, because we want to predict what would happen if we performed some experiment that was between the experiments we’ve previously performed, but we can be ever so very slightly agnostic about whether the moon is there when we aren’t bouncing a laser beam off it.

124. Mateus Araújo Says:

Scott #108: I think you should ask the people you have been charitable to, if they found it a good thing. I have a feeling they will just feel insulted.

125. Mateus Araújo Says:

Ryan O’Donnell #102: In this case going for mixed states and CPTP maps doesn’t get you anything different from pure states and unitaries. It’s just a different representation of the same thing, because of the various purification theorems. It’s often very convenient for calculations, but has zero consequence for foundations.

126. fred Says:

Haha, Scott, you’re in Sabine’s new video about complex numbers (which, obviously, I won’t link).

127. Scott Says:

Steven Evans #121:

It’s not about appeal, though, is it? Superdeterminism is not ruled out by the evidence.

It absolutely is about appeal. Russell’s teapot also isn’t ruled out by the evidence. But we have no positive reason whatsoever for supposing it to exist, in exchange for the massive conceptual problems it would introduce if it did exist. I feel exactly the same way about superdeterminism.

128. Scott Says:

The Erogamer 7.! #120: Where is that from?!? Is it an original composition? Is it an extract from something longer? (I’d happily read more of it.)

129. Scott Says:

Mateus Araújo #124: You’re telling me that, even in the wake of PBR and the other no-go theorems, there are still living, breathing epistemicists (in the sense of “|ψ⟩ represents a nontrivial probability distribution over hidden ontic states”)? Who are they? Do they live in special nature preserves or something? 😀

130. hwold Says:

How do you reconcile Scott #53 and point 7 of the blog post ? Isn’t “a |0⟩ |You0⟩ + b |1⟩ |You1⟩” meaning that You are a coherent superposition of You0 and You1 ?

(What is a coherent superposition technically, BTW, as opposed to a “simple” superposition, without the “coherent” modifier ?)

131. Scott Says:

hwold #130: The adjective “coherent” just means that there are no additional environmental degrees of freedom that we don’t have control over and with which the branches of the superposition are fully entangled.

All knowledgeable people, not just the Everettians, will agree with the formula I wrote in #53, provided we replace “You” by “He” or “She”—i.e., by some other agent who you get to stand apart from and regard quantum-mechanically. The crux of the debate is whether it even makes sense to talk about a “quantum state of the universe” that includes yourself as part of it. To whatever extent you say “yes,” you’re going to be pulled in an Everettian direction; to whatever extent you say “no,” pulled in a Copenhagen/QBist direction; to whatever extent you say “mu,” pulled in a Zen direction. 🙂

132. Scott Says:

I just thought of another way to express the Zen position. It’s like, I’m nerdy enough to believe that there are many, many important insights about the nature of reality that can only be achieved by careful, systematic clarification and analysis of concepts, rather than (say) sitting around listening to the Beach Boys’ “Good Vibrations” while stoned out of one’s mind. Having said that, whenever I see the returns from further conceptual clarification in some given direction falling below the “Beach Boys threshold,” I tend to favor just cutting our losses in that direction. In my view, the question of the reality or unreality of other you’s in other branches of the wavefunction has probably fallen below the Beach Boys threshold since the 1970s or 1980s, at least for people who already understand all the relevant facts about decoherence and so forth.

133. Wyrd Smythe Says:

I’ve long thought the key is determining exactly why the classical world is different from the quantum world. Is there a Heisenberg Cut and, if so, what’s responsible for it.

Your comment #53 embodies a question I’ve long struggled with: How is the quantum state of a tiny quantum system able to affect the quantum state of |You⟩? That seems like throwing ping-pong balls at an oceanliner. It probably has some effect, but seems like it would be entirely swamped out.

I also struggle with the idea that Everett doesn’t seem to solve the measurement problem. In a spin measurement, for instance, we end up with α|Self+Lab⟩|Up⟩ and β|Self+Lab⟩|Down⟩ — two branches — but doesn’t the wave-function in both “collapse” to a known state?

I also wonder how we can experience “particles” since those seem to represent a “collapse” of the eigenstate for position.

134. Greg Kuperberg Says:

Mateujs #125 –

In this case going for mixed states and CPTP maps doesn’t get you anything different from pure states and unitaries. It’s just a different representation of the same thing, because of the various purification theorems. It’s often very convenient for calculations, but has zero consequence for foundations.

I don’t really understand your stance here. You first argued that measurements can’t be represented by CPTP maps. After Ryan and I pointed out that you actually can, you say that it isn’t a convincing answer because you can dilate it to pure states and unitaries. Why does that mean that it’s not a convincing answer?

I argue that a version of this answer isn’t just “convenient for calculations”, but is also a correct, accurate model of real life. For clarity, I hope that we can allow a classical computer as an agent that can witness a measurement of a quantum system. (This is to get away from the mind-body problem, using the precept that people are evolved, biological, classical computers.) If so, I would write your measurement formula as $$\mathcal{M}:\rho \mapsto \sum_k P_k \rho P_k \otimes \mu_k$$, where $$P_k$$ is the projection operator of the outcome $$k$$ and $$\mu_k$$ is an appropriate mixed state of the classical computer having recorded the answer $$k$$. I don’t want to just say $$\mu_k = |k\rangle\langle k|$$, because a realistic classical computer (or any other observer) has a vastly larger physical Hilbert space than just the number of outcomes of the measurement. Instead of orthogonal states, the correct rule is that the measurement records are nearly disjoint mixed states, i.e. that $$\mu_j \sim\perp \mu_k$$ when $$j \ne k$$. Although the specifics of the mixed states $$\mu_k$$ are negotiable (as well as complicated), and therefore that $$\mathcal{M}$$ is not completely specified, I would argue that the CPTP $$\mathcal{M}$$ is a truthful summary of a classical computer measuring a quantum object.

I have faith in this as the answer to the question, “What kind of animal is measurement? What is state collapse?” Ryan endorsed this type of answer too. Yes, it means that the universe stays in a state of uncertainty. But in my view, that’s how Bayesianism works (both quantum and classical Bayesianism), and (as Scott said) leaves us with the mind-body problem as the remaining philosophical question. Meanwhile what remains of the measurement problem as a physics question is which CPTP $$\mathcal{M}$$ might be, and how you might decompose it dynamically. That’s a valid topic, but it is just a special case of CPTP modelling of open quantum systems in general.

135. Daniel Matysiak Says:

In this vein and of possible interest:

https://www.wolfson.cam.ac.uk/about/news/there-just-one-fundamental-thing-exists-jrf-proposes-new-ontology-quantum-mechanics

(Direct link to journal article can be found within.)

136. Scott Says:

Ward Smythe #133:

How is the quantum state of a tiny quantum system able to affect the quantum state of |You⟩? That seems like throwing ping-pong balls at an oceanliner. It probably has some effect, but seems like it would be entirely swamped out.

This is a completely resolvable confusion. The main point here is ordinary chaos, or sensitive dependence on initial conditions. What actually happens is that (say) a single photon strikes a photodetector, and that sets off an exponential cascade of more and more particles banging into each other (almost like in a nuclear chain reaction), until enough has happened that you can see it with your eye.

I also wonder how we can experience “particles” since those seem to represent a “collapse” of the eigenstate for position.

I dunno, have you ever “experienced a particle”? I’m told that it might be possible for the human retina to detect a single photon, but only under extreme conditions (and even then, presumably your retina isn’t the locus of your experience, but something in your brain). Or have you only experienced particles’ amplified effects, as in the example above?

137. JimV Says:

Beautiful writing. Thanks.

138. Mateus Araújo Says:

Scott #129: Sure, Bohmians are the original epistemicists, and they will never die. Tim Maudlin and Travis Norsen are alive and kicking, for example.

There are also non-Bohmian epistemicists, like Rob Spekkens and Matt Leifer. They freely admit that they don’t have a working theory, but are sure that the quantum state will turn out to be epistemic if they ever manage to build one.

I don’t think the PBR theorem changed anybody’s mind. Bell’s theorem already showed that psi-epistemic models must be nonlocal; PBR showed that you can’t have independent preparations either, which doesn’t seem to be a problem if you already accept nonlocality. Bohmian mechanics survived the PBR unchanged, for instance.

Also, I don’t think your concept of epistemic state, which is just epistemic because of being part of the emergent rather fundamental ontology, actually captures what QBists and Copenhageners mean by epistemic states. It’s rather hard to know what they mean, so you’ll have to ask them (not me) if you’re interested.

139. Wyrd Smythe Says:

What adds to my confusion is that your second answer seems to contradict your first. With regard to particles: cathode ray tubes, the photo-electric effect that won Einstein’s Nobel, and what they do at CERN and other particle accelerators, seems to give weight to the notion of particles.

With regard to the photodetector and the exponential cascade, isn’t that an energy thing, where the system is configured to supply its own energy to amplify the effect of a single particle? It’s not clear to me that the detector is in some critical quantum state where detection of a single event can affect the quantum state of the detector.

Isn’t the detector’s quantum state the combination of all the particles of that detector? What does the solution for the Schrödinger equation look like for an entire detector? I would imagine it contains a vast number of contributing terms, and I find it hard to accept that adding a single new quantum state can have that much effect. The ping-pong ball versus oceanliner analogy seems apt.

As you suggest, when the photon’s wave-function “collapses” by being absorbed by some electron in the detector, isn’t that a “measurement” with a definite outcome? I can see that first electron, and even nearby electrons entering a superposed state of |detection⟩ + |nothing⟩ but it seems that superposition must be quickly damped out by the rest of the detector let alone the environment.

It also raises the question of how physical matter is able to coincide. Why doesn’t the Pauli principle prohibit fermions from coinciding? The putative overall wave-function might give them a superposition, but aren’t the individual electrons occupying all possible quantum states they can have? I don’t see how the detector, let alone the cat or Wigner’s friend and lab, are allowed to physically coincide.

140. Mateus Araújo Says:

Greg Kuperberg #134: I thought Ryan meant a *collapse* measurement, as that would make a difference, and is impossible. After you and him clarified that he meant a *non-collapse* measurement, I said sure, of course you can, but it doesn’t make any difference.

Ryan was proposing to take CPTP maps as the set of allowed transformations in order to solve the incompatibility between unitary evolution and measurements. Well it turns out that he solved it by claiming that a measurement is a unitary! Sure that is a solution, but has nothing to do with CPTP maps, that’s just obscuring what the question is.

Your CPTP map does truthfully model a measurement, in my opinion, but it is literally the same way Many-Worlds does it! Let me be more precise. The unitary doing the measurement is given by
$$W = \sum_k \Pi_k \otimes U_k \otimes V_k,$$ where $$\Pi_k$$ is a projector acting on the quantum system, $$U_k$$ is a unitary recording the measurement result on your computer, and $$V_k$$ is a unitary uncontrollably recording it on the environment. Acting on a intial state $$\rho \otimes \mu \otimes e$$, we end up with
$$\sum_{kl} \Pi_k \rho \Pi_l \otimes U_k \mu U_l^\dagger \otimes V_k e V_l^\dagger.$$ We don’t have access to the environment by assumption, so we should trace it out. The usual decoherence result is that $$\mathrm{tr} (V_k e V_l^\dagger) \approx \delta_{kl}$$, so we end up with
$$\sum_{k} \Pi_k \rho \Pi_k \otimes U_k \mu U_k^\dagger.$$ Define $$\mu_k = U_k \mu U_k^\dagger$$, and we get the same CPTP map as you. Of course, this is the situation where you don’t know the result of the measurement, otherwise you would just use the relative state in your branch, $$\Pi_k \rho \Pi_k \otimes \mu_k/\mathrm{tr}(\Pi_k \rho)$$.

Maybe it’s better to look at what other solutions of the measurement problem look like. It’s either adding hidden variables (Bohmian mechanics), or changing the Schrödinger equation to make collapse a physical process (collapse models). These are genuinely different solutions! Now, representing the same physics via unitaries or CPTP maps? No, come on.

141. Greg Kuperberg Says:

Mateus –

I thought Ryan meant a *collapse* measurement, as that would make a difference, and is impossible. After you and him clarified that he meant a *non-collapse* measurement, I said sure, of course you can, but it doesn’t make any difference.

But I don’t see the distinction. A measurement is a measurement; the only difference between the “collapse” and “non-collapse” kinds for me is if I get to learn the answer. I specifically offered that the measuring agent is a classical computer as a stand-in for a person; what difference would it make if it really were a person? Does that make me “many worlds”? I guess I can start to see it, even though I don’t like to think of it that way. But two things: (a) I only accept it as a strictly solipsistic many-worlds model that applies only to the self and not anyone else. I.e., I accept it only within the context of the mind-body problem. (b) I can only accept it as a many-worlds model that makes no distinction between classical and quantum uncertainty.

Your CPTP map does truthfully model a measurement, in my opinion

Good, I’m glad that we agree on that much.

Maybe it’s better to look at what other solutions of the measurement problem look like. It’s either adding hidden variables (Bohmian mechanics), or changing the Schrödinger equation to make collapse a physical process (collapse models). These are genuinely different solutions!

I’m not sure if Bohmian “mechanics” really counts as geniunely different; I will grand that changing the Schrödinger equation probably would. Either way, it leaves me cold, because it strikes me as an attempt to use physics to solve a philosophy or psychology question — the mind-body problem. If we’re willing to leave a classical digital computer in a mixed state when it does a measurement, then I’m also willing to leave homo sapiens in a mixed state when it does a measurement, including me, except in my own perception. I guess it can start to feel many-worlds-ish, but there are reasons that I don’t like that terminology, one of them being that the density matrix form makes a lot more sense to me as a description of the non-determinism than images of futures with exp(Avogadro) bifurcations.

142. A1987dM Says:

@Leo #45:

“If I open the box and see the cat is alive, should I be relieved, or remain sad because half of the cat’s measure has died?”

Scott’s point 7 sounds applicable here — we can’t tell you whether to care about the life of the cat in other branches until we know why you care about the life of the cat in this branch.

143. Tu Says:

Scott,

I am sorry that you are feeling a bit under the weather, but very glad to hear that you are fully vaccinated. I know that you were thrilled to see Governor Abbott set Texans free once and for all– I expect to see you back out and about on the Austin nightclub scene shortly.

Thank you for writing this post. It takes a lot of courage, given your readership, to write a blog post on the philosophy of QM that is not a rigid affirmation of the virtues of the many worlds interpretation.

After several years now of studying QM, I feel myself coming to rest in a similar state of acceptance/anti-interpretationism. QM puts substantial restrictions on what can qualify as a fundamental description of reality. At the same time, some things present in the theory do not have obvious physical correlates. In the words of the very underrated John Bell, “a curious situation indeed.”

I like to think that in 100 years you will not be remembered for your contributions to quantum computing, computational complexity, etc, but will rather be remembered as the revolutionary philosopher of science responsible for “Aaronson’s Beach Boys Threshold.” People will no longer debate whether a hypothesis is falsifiable in principle, or falsifiable in practice, but rather if such hypotheses would render our picture of the universe sharper than simply smoking some weed and listening to the beach boys would.

Jokes aside, I think the BBT is right on the money here. I think one thing that helped me get out of my obsession with this topic/depression over its intractability was the realization that interpreting QM is maybe not the deepest, most important question. Even if we accept one of the candidate interpretations, important mysteries remain. Is space a continuum? What about time? What happened before the big bang? What if someone spills Dr. Pepper on the server our simulation is running on? Once we venture into the wonderful world below the Beach Boys Threshold, I prefer to spend my time on questions like these. 🙂

144. A1987dM Says:

@Fred #69:

> Consciousness isn’t some sort of extra ingredient on top of the mechanical state of the brain, it’s just some side effect of it… but then, how come we’re talking about consciousness as if it’s an actual thing that’s somewhat truly emergent on top of the physical laws? If it were truly emergent, it wouldn’t be explained by the physical laws alone (in the bottom up way of science). So no simulation of a brain would ever come up with a coherent discussion of consciousness – if they would, then consciousness can’t be a special ingredient (to be truly emergent, a phenomena can’t be explained by underlying causes).

Huh… The rules of chess don’t mention the concept of pawn structure. Like, you never need to know what pawn structure is to know whether a chess move is legal, and *in principle* you never really *need* to know about pawn structure to know whether a chess move is a good move — but this doesn’t mean it never makes it much much easier *in practice*.

Consciousness sounds like this to me — even if there’s nothing that can’t *in principle* be explained without it by an omniscient being, doesn’t mean there aren’t lots of things which *in practice* would be way way more unfeasible for actual human beings to explain without it than with it.

145. Andrei Says:

Scott,

“Why even talk about EPR-Bohm when for 57 years, the Bell inequality has thrown the same issues into much sharper relief?”

There is a very good reason for doing that. The EPR-Bohm experiments presents a different argument than Bell’s. It provides a rock-solid proof that, indeed, “there are only two possibilities,

(1) local hidden variables or
(2) superluminal signalling”

The argument is presented below:

At two distant locations (A and B) you can measure the X-spin of an entangled pair. QM predicts that:

P1: If you measure the X-spin at A you can predict with certainty (probability 1) the X-spin at B.

Let’s exclude non-locality:

P2: the X-spin at B is not determined by the measurement at A.

From P1 and P2 it follows that there was something at B that determined the result of the B measurement. EPR named that “something” an “element of reality”. So:

P3: There is an element of reality at B that determines the measurement result at B.

You may observe that there is no other logical option available (unless you think that it’s by pure luck that we manage to always predict with certainty the X-spin at B, which is rather absurd).

From P1 and P3 it follows:

P4: there was an element of reality at A that determined the measurement at A.

This is because once we’ve established that the measurement at B was fixed (P3) it’s impossible that the measurement at A could have been different, right?

OK, so P3 and P4 lead to:

C: The X-spin of both particles, at A and B were determined before any measurement took place (deterministic hidden variables).

I think that ignoring EPR-Bohm and going directly to Bell is a big mistake. Bell’s theorem was created as a refinement of EPR (look at the name of the original paper), not a replacement. Bell agreed that EPR proved that locality can only be saved by hidden variables, and his theorem was supposed to test the viability of those hidden variable theories. The conclusion is that they fail and QM is non-local. This conclusion is still true (with the exception of superdeterminism).

Now, if EPR is ignored you can say that non-deterministic theories like Copenhagen can be local because they are not ruled out by Bell. This is false. They have been ruled out by EPR and Bell does not address them at all. Bell theorem is only defined for the remaining class of local theories, the hidden variables ones.

“My short answer as to why the spins are nevertheless anticorrelated is: because QM is true, and because QM straightforwardly predicts exactly the correlations we observe.”

Yes, but is this local or not? I think the answer to this question is extremely important. If there are hidden variables we should find that correct theory. If non-locality is true, again, we should look for that non-local theory.

“The reality or unreality of the other Everett branches still occasionally keeps me up at night, but I confess that 2-particle entanglement hasn’t kept me up at night for at least ~15 years.”

So, in the end you just don’t want to present your understanding of this simple experiment. Why? If it’s parallel worlds why not saying it out loud?

146. Why is P vs NP so important? Says:

Scott coherent answers for #47 (which was reply to your #27)?

147. AJD Says:

Sidney Coleman’s QMIYF lecture is truly fantastic. A recording can be found at
https://www.youtube.com/watch?v=EtyNMlXN-sw

I naturally read it as fully endorsing MWI, but I’m sure proponents of other theories interpret it as fully supporting their own favored interpretation. Or lack thereof, pace Scott.

I have found one thing truly lacking in the discussion: there haven’t been enough slogans thrown around. Where’s “Church of the larger Hilbert Space”, and “Hilbert Space is a big place”? Sure, Mateus has used those arguments with math, but what’s the fun in that?

148. gentzen Says:

Scott #85:

gentzen #49: Your question (was it a question?) is a little too complicated for me to disentangle, but I can certainly tell you that no “solution” involving superdeterminism has ever had the slightest appeal for me.

I guess the question is whether I am expected (by you?, by Sabine Hossenfelder?, by Arnold Neumaier?, by Nicolas Gisin?) to “emerge from the shadows”. Sounds strange? This is not namedropping, even so I never met any of them in person, and I didn’t even try to contact Nicolas Gisin. (“emerge from the shadows” here means to actively defend my opinion)

I think part of the reason why you wrote this post is that currently some people like Sabine Hossenfelder, Sandro Donadi, Tim Palmer, Marc-Olivier Renou, Nicolas Gisin, or Miguel Navascues really mean business with respect to making progress in our understanding of quantum mechanics. OK, that list of people is very arbitrary, but something is in the air.

I once wrote: “A real number can contain an infinite amount of information, but nature probably doesn’t contain an infinite amount of information in a finite volume. So nature has to use QM (or something similar) to blur details enough such they no longer contain an infinite amount of information.” Gisin expressed similar thoughts (and still cares today):

Given the complexity of the (many-) worlds, it had to be encoded in some infinitesimal digits of some quantum state, possibly in the billionths of billionths decimal place. I am always astonished that some people seriously believe in that. … Do these infinitesimal digits have a real impact on the real world? Is this still proper physics? For sure, such assumptions can’t be tested. Hence, for me, hyper-determinism is a non-sense …

I read Arnold Neumaier’s book Coherent Quantum Physics. (Part of the reasons why I read it was to learn details of his interpretation of probability, which also fixes this “infinite information” issue.) I was blown away by the many remarks and observations that are spot on both practically and philosophically. His thermal interpretation is a “valid” clear and unambiguous interpretation that deserves to be known and discussed, just like Copenhagen interpretations, the deBroglie-Bohm interpretation, Many World interpretations, or the minimal statistical interpretation are known and get discussed. But who should care about my opinion? And who should care about interpretations anyway?

Well, maybe Scott Aaronson still cares about his thoughts on “Many Worlds minus the Many Worlds”. Of course I get that the following was also directed at me:

I’m not optimistic that any new interpretation, or any new argument for an existing interpretation, is going to come along and make everyone slap their foreheads and exclaim “it’s all so clear now! why didn’t any of us think of that earlier??”

The truth is that I am grateful that Scott tried to answer me. Perhaps I also hoped for a reaction by Mateus Araújo, like the one in #37. This is a hard one, he certainly had an impact on my thinking about these issues, but I also somehow fear his reactions. I expect new interpretations to raise new questions, not to “make everyone slap their foreheads”. (Examples of such questions for the thermal interpretation would be whether spacelike or timelike quantum correlations are observable, not just by fiat but also within the interpretation/QM theory itself by a suitably modeled measurement setup. My guess is that only spacelike quantum correlations are observable in that sense. Or that question about nonlocal randomness which made me read Gisin’s book Quantum Chance.)

The thermal interpretation is a deterministic interpretation, just like deBroglie-Bohm and Many World. Unlike deBroglie-Bohm, it doesn’t artificially break existing physical symmetries, and is not more nonlocal than QM itself. This raises the question how it got around “Wigner’s classic argument, which proves the incompatibility of unrestricted unitarity, the unrestricted superposition principle and the collapse of the state during a measurement” (or around a similar theorem that Mateus Araújo called Maudlin’s trilemma). It is done by requiring that there is only one state, the state of the entire universe, and by explaining that you cannot superpose universes.

Is this a valid answer? OK, there was more, the thermal interpretation also explained what are its macroscopic observables and why, and that made it a bit easier to accept this answer. But in the end, just like the nonlocality is the point where you either accept or reject deBroglie-Bohm, this answer is the point to either accept or reject the thermal interpretation. So what are the drawbacks of this answer, and what is the unintuitive and undesirable thing to swallow? It is that the initial state is no longer (statistically?) independent of the measurement setup. And this is basically the superdeterminism loophole (which is apparently not just a “known” lookhole for the Bell test, but also an “overlocked” loophole for Wigner’s argument and Maudlin’s trilemma).

So at that point, I remembered that Sabine Hossenfelder was very excited about superdeterminism, and read her paper with Tim Palmer. Some of their points were very convincing for me, at others I feared they don’t realize just how bad and devastating (for reductionism, for science, for …) superdeterminism really is. But Bohr’s complementarity is also devastating for reductionism (and science), just more ambiguous and less clear.

But let me come to the inconvenient parts. The thermal interpretation basically managed to get “Many Worlds minus the Many Worlds” working consistently (even so comparing it to Many World might feel insulting, both to many worlders and to Neumaier), but by doing so it also showed the price you have to pay. At least that is my opinion. But it gets worse, once you accept superdeterminism, it also makes sense to start looking for non-linear (even stochastic?) “low dimensional” models, whose statistical behavior gets described by Schrödinger equation. And then you are really in the realm where people consider you to chase a hopeless dream, even if you would be a Nobel laureate like Gerard ‘t Hooft.

149. Mateus Araújo Says:

Greg Kuperberg #141: The distinction is merely whether you get several worlds or not! I’m afraid you are so deeply certain that collapse is not physical that you forget that some people actually believe there is a physical collapse happening upon measurement that erases the other possibilities from existence. This is what the debate is about. Or was about, given that you and everybody else in this thread is not even acknowledging that alternative anymore.

Learning the answer is merely eliminating your self-locating uncertainty; if you don’t know in which branch you are you’re better off using that mixed state to represent your ignorance, but once you learn the answer you can safely ignore the other branches, as decoherence will prevent them from fucking you up in the future. This applies irrespectively of whether “you” are a human or a computer, so I don’t see why you’re bringing up the mind-body problem here.

I also don’t see why you think that Bohmian mechanics has anything to do with the mind-body problem. It was definitely not developed for that, the goal was to explain what happens in a measurement (and consequently have a quantum theory that was valid at all scales), irrespectively of whether this measurement was done by a human or a computer. Mind you, this was the 50s, people still took physical collapse seriously back then.

I’m curious about what why would anybody do a many-worlds model of classical uncertainty. Really classical uncertainty, not these “classical” uncertainties that boil down to quantum effects, but something like the uncertainty about what is the 100th digit of the Tsirelson bound of the I3322 inequality – stuff that is perfectly determined by mathematics but nobody yet knows.

150. Paul Says:

Can I send you a new yet still under review interpretation? I quote you (AAronson) in it.

151. JimV Says:

Now having skim-read the comments, I’m a bit disappointed in your summary dismissal of super-determinism (not that anybody should care about my disappointments). I am suspicious of it myself, but there are at least a few good minds who do consider it a live possibility, having to do with chaotic sensitivity to initial conditions (at the point of experimental measurements) which would be hard to distinguish from actual randomness. That is, there might be other conceivable mechanisms (at least one) for it than conspiracy theories. If that were true, than dismissing it on the basis that it must be a conspiracy theory would be a fallacy.

Personally, I think a lot of determinism (perhaps of a statistical nature) and a little bit of randomness is a good way to run a universe; but chaotic pseudo-randomness might work as well. (Complete determinism can produce loops which there is no way out of. A little bit of randomness adds a way out.)

152. Scott Says:

Tu #143: It took me just a day to recover and be fine. And yes, of course, I’m now partying like it’s 2019!

Mere hours after I enunciated my Beach Boys threshold, you have the honor of being the first to put it in its proper context, as what will surely be my most enduring contribution to intellectual life.

Actually, if you consider a typical quantum foundations paper mashing together relationalism, Wigner’s friend, EPR, and the words “ontic” and “epistemic” into a slightly different concoction, I feel it might have a hard time competing for insight even just against the isolated line “I dunno where but she takes me there.” 😀

153. Scott Says:

JimV #151: No, superdeterminism is much more insane than you and most others realize it is. It’s not about anything benign like chaotically amplified randomness from the initial state. Instead, it’s necessarily about a conspiratorial fine-tuning of the initial conditions to make it look like the Bell inequality is being violated even when it isn’t. By the same device, one could also produce (the appearance of) superluminal signaling or basically any supernatural phenomenon one wanted, although of course the superdeterminists never spend time worrying about that.

154. Scott Says:

Paul #150:

Can I send you a new yet still under review interpretation? I quote you (AAronson) in it.

You can send it, but no promises that I’ll read it!

—“AAronson”

155. Paul Says:

email it to what address?

156. Scott Says:

Paul #155: Seriously, you missed my last post?? scott@scottaaronson.com

157. AJD Says:

Greg Kuperberg wrote:

I guess it can start to feel many-worlds-ish, but there are reasons that I don’t like that terminology, one of them being that the density matrix form makes a lot more sense to me as a description of the non-determinism than images of futures with exp(Avogadro) bifurcations.

I honestly don’t see an important difference. The density matrix in that case is also of size exp(Avogadro). The exponentially large state space is part of quantum mechanics. The earlier forms of MWI exposition may emphasize it by talking about noninterfering terms of an exponentially large equation describing a pure state. But it can equally well be described as drawing lines around some of the entries of a density matrix saying “the future evolution of these entries is only sensitive to others in the same bucket” — as if they’re evolving in entirely separate worlds. In modern formulations it of course points to decoherence arguments as the justification. CPTP maps do rather a similar thing to the same space: apply some unitary, and zero out a whole bunch of cross-terms, leaving a blocked structure.

What I do see as the difference is not in analyzing or describing the outputs after the measurements, but in how fundamental measurements and their description is. The MWI makes the astonishingly bold claim that measurements are merely normal quantum evolution. The description in terms of CPTP maps instead is completely agnostic to what they are, or to how the state changes during them, only describing their idealized effects. This is indeed a very useful way to parameterize and classify the entire space of measurements possible. It is a fine characterization suited for many purposes. But it doesn’t even try to answer “what’s going on”, only answering “how can we practically model what we observe”. And for that, the extra blocks that MWI insists on keeping around are superfluous.

158. Greg Kuperberg Says:

Mateus –

The distinction is merely whether you get several worlds or not! I’m afraid you are so deeply certain that collapse is not physical that you forget that some people actually believe there is a physical collapse happening upon measurement that erases the other possibilities from existence. This is what the debate is about. Or was about, given that you and everybody else in this thread is not even acknowledging that alternative anymore.

No, no, I understand why some people might want a separate physical theory of state collapse, only I personally don’t see that it is necessary. But also, we have mostly exchanged our views at this point and we can agree to disagree about some of this.

Just to summarize my own position, I guess I believe that decoherence due to hidden measurement, or what you call non-collapsing measurement, is roughly speaking “ontic”, while collapsing measurement is basically “epistemic”. (Or another way to say it is that we don’t collapse, but we rationally perceive it.) Since it is more-or-less epistemic, I’m happy to model a collapsing measurement (plus anything that follows one) by a “CPSTP” — a completely positive, sub-trace-preserving map. As you would know, if it is from a projective quantum measurement, then it is even nicer, a subunitary operator.

I also don’t see why you think that Bohmian mechanics has anything to do with the mind-body problem.

I agree that Bohmian mechanics might not do much for the mind-body problem. Actually, in the category of divergent interpretations of standard QM, I don’t know what Bohmian mechanics really accomplishes. If I understand correctly, in this theory a particle surfs its wave, and it just strikes me as an artificial mixture of probability and geometry.

I’m curious about what why would anybody do a many-worlds model of classical uncertainty.

To be clear, and again I should mainly defend my own position rather than trample anyone else’s, the “many-worlds” label is not my favorite and I would only accept an epistemic version of it. If that is understood, then here I make a distinction between classical uncertainty for real-life questions like the 538 presidential forecast, and classical uncertainty for mathematical questions like the trillionth digit of $$\pi^{\sqrt{2}}$$. I argue that the former type of classical uncertainty is thoroughly entangled (I guess pun intended) with quantum uncertainty. If I am supposed to imagine two separate worlds in which the cat lived and died, then it is just as necessary to imagine a separate world in which Trump won reelection.

Now, it’s an interesting question to analyze what is meant by placing bets on the trillionth digit of $$\pi^{\sqrt{2}}$$. I do think that someone could rationally place such a bet, but I’m not completely sure why it’s rational! (Actually it’s a conjecture that $$\pi^{\sqrt{2}}$$ is irrational, but what I mean is that the bet might be rational.) Regardless of the answer to that, it’s in a different category, having to do with limits on computational power and mathematical knowledge, rather than limits on observation of the real world.

159. The Erogamer 6.8.d: One-Sided Conversation Says:

“Is Bob other things too?” Charles said.

“Like what?” said the preacher.

“Uh, say, could Bob be this sidewalk,” said Charles. It was the first thing that had come to mind.

“Anything can be Bob if Bob wills it,” the preacher said solemnly. “But I don’t see why Bob would want people walking on him all day long.”

“Does Bob sometimes will to be other people?” said Charles. “Could Bob be part of other people, or could somebody be Bob and not Bob in some way that’s hard for mortals to understand?”

“Well, as to that…” the preacher said, and paused.

“As to that?” Charles said.

“As to that, I’d say your question is… total nonsense! Everyone is themselves. If you were someone else you’d be someone else instead of you, and then you wouldn’t be you being someone else, you’d just be someone else being someone else! Bob could be talking to you through me without you knowing, but Bob couldn’t be me with or without me knowing! The distinction between myself and you isn’t that I’m up here and you’re down there, it’s that I’m what’s up here and you’re what’s down there. Asking whether somebody is themselves or somebody else is like asking why this is Norville instead of Norway, or whether the Earth is upside down, or why it’s today instead of yesterday, or how fast time is going right now, or whether you exist or don’t exist, or if there’s something rather than nothing. It’s all about where the question is being asked, see? And what it is that’s asking.” The preacher turned away and began addressing the crowd once more. “Do you have questions in your heart? Bob has answers in his Pipe! Bob put all the questions and all the answers into his Pipe and smoked it!”

“No, wait,” Charles said, once his mind had caught up. “Wait, I understand why you can’t ask how fast time is going without being outside of time, but how is my existing or not existing like that? That’s an absolute if anything is an absolute! Cogito ergo sum, I think therefore I am!”

“Yeah?” said the preacher, turning his head again. “You, whoever you are, mister…”

“Charles,” said Charles.

“Mister Charles, suppose I told you that you, right now, didn’t exist. Say, you’re only an animation in a Disney movie. How could you tell?”

“I… what?” said Charles. “If I know literally anything, I know that’s false! I have to exist in order to ask whether or not I exist!”

“Aha!” said the grey-bearded preacher, holding up a finger. “But maybe you don’t exist and therefore you’re not asking whether you exist. How can you tell the difference? If you were an animated drawing, you’d be saying just the same thing.”

“That… honestly sounds to me like nonsense,” said Charles.

“Exactly, mister gentleman Charles sir! It is nonsense just as you say! Maybe somebody outside of existence thinks that we don’t exist, but from the inside of reality we’ve got no way to know that and no reason to care. It’s the same way with Disney princesses. You can say all you want that Snow White doesn’t exist, but she can’t hear you, so as far as she knows, she exists. She’s got just as much proof of her own existence as you have.”

“But…” Charles said. “But, accepting that for the sake of argument, how do we go from that to Snow White becoming real here?

“Friend,” said the preacher, “it’s not like I haven’t enjoyed our conversation, but I fear it might not be entertaining the rest of my audience as much as it’s entertaining us.”

“Yeah,” said Charles. He was aware that he was flushing. “I’ll go now.”

“Go with the blessings of Bob, my friend whose reality or unreality is ultimately a nonsensical question!”

(Excerpted from The Erogamer, a 660,000-word NSFW quest on Questionable Questing that kept getting deeper and deeper until the deepness went completely out of control. Link in commenter URL. QQ requires a login to read.)

160. Michael Weissman Says:

On #4, I think that your phrase “under reasonable assumptions” is doing all the heavy lifting. The “reasonable assumptions” are sort of crazy-sounding in the context of the plain theory, although virtually necessary in our ordinary frames of thought. I think there’s been only one semi-plausible bridge from the plain theory to Born, suggested by Jacques Mallah, who called it the “many computations approach”.
https://ui.adsabs.harvard.edu/abs/2007arXiv0709.0544M/abstract

161. Felix Says:

I think the issue of ontic/epistemic interpretations has a more significant formulation than the way you frame it here.

Specifically, we can ask does quantum theory require an external classical observer, so that all quantum are open systems and only have relevance insofar as external classical agents extract information from them?  Or can we include the observer as a part of the quantum system so the whole universe is a closed system?  I think you more or less made this version of the ontic/epistemic distinction at the end of your “Get Real” article on PBR.

What is even more interesting to me is when ontic approaches end up sneaking back in a strict observer/target system distinction. Not in the classical/quantum form, but by mixing two very different types of quantum systems.  For example, Unruh Dewitt detector models have a “particle/atom-like” observer and a field-like/local subalgebra target system.  I think the real burden of proof for ontic approaches is to do away with this, so the subsystems we label observer and target is not given a priori, but this is especially challenging in a field theory. I would say this, rather than the Born rule, is where the action is.

162. Tim Converse Says:

Scott #112: Thanks for the response. Yes, I was starting off with a scientific-realism stance (roughly: that there is stuff that exists out there, and that existence is independent of facts about whether we can observe it or not).

And yes, you can definitely counter that with a move to positivism (roughly: that we can only talk about things that we could in principle observe, and so we make an executive decision that
1) stuff-that-exists and
2) stuff-that-we-could-in-principle-observe
are exactly the same thing).
“Going positivist” is sort of the standard move to make in response to a realist stance about unobservable things, right?

I’m curious about your comfort level with the positivist stance in another case of non-observability. Speed-of-light limits imply that our observable universe has a finite extent. Stars and galaxies that fall within that spatial area are in principle observable by us; if there were, say, galaxies that were a little bit too far away to be be in our observable universe, then we’re not going to be able to interact with them, or observe them, ever.

Nevertheless, taking the realist stance, I believe that:

a) there is *some* fact of the matter about what exists that is just a little too far away to be in our observable universe. And what’s more:

b) I believe that conditions just over that border are probably a lot like conditions just this side of that border, because it would be awkward for cosmological theories if they weren’t. In particular, just like there are galaxies that we can observe (because they’re close enough), I think it’s likely that there are galaxies that we can’t observe (because they’re too far away), but those unobservable galaxies exist even so.
[I’m aware that the language I am using implies a notion of consensus simultaneous time that doesn’t hold over these spatial scales (e.g. talking about what “exists” in a present tense), but I hope that the broader point survives that oversimplification)]

Do you believe or disbelieve in the existence of galaxies that are too far away for us to observe? I can imagine being agnostic about this (and saying that we don’t have enough information to answer the question), but I’m especially curious if in this case you would also want to say that you reject all three of
1) EXISTS,
2) ~EXISTS,
3) agnosticism about EXISTS/~EXISTS,
and say that you believe
4) that both EXISTS and ~EXISTS capture aspects of the situation(?)

I’m guessing that you would take the QM/many-worlds question to be interestingly different from the cosmology/distant-galaxies question in this respect, but if so it would be really interesting to understand the difference.

By the way, I really liked your thought experiment about our branch being saved to hard disk and sent to the printer, where other branches might have a more on-the-fly existence, and what do we make of hard EXISTS/~EXISTS questions in that case? I remember a money quote from Deutsch’s Fabric of Reality from long ago, which I would paraphrase as “this quantum-computational work is happening somewhere, and if you don’t think that it’s happening across parallel worlds, then where do you think it is happening?” 🙂 But yes, maybe other branches are lazily evaluated as needed, and get spun up like virtual cloud-computing instances and then are deallocated. This is indeed an in-between answer to questions of real existence.

If so, though, note that you have assigned a different ontological status to our branch than you have to others (I think). So instead of opposing

1) Everettians (many-worlds believers) who say EXISTS about other branches
2) Copenhagenists (wave-function-collapse believers) who say ~EXISTS

let’s make the distinction be between:

A) Everettians (many-worlds believers) who say that other branches exist and there’s nothing special about ours, and it’s all quite symmetric
B) Non-Everettians who believe that while the ontological status of other branches may be nuanced, *our* branch is special (including as a subset the Copenhagenists who believe it’s so special that it’s the only one).

If you believe in ontological symmetry between the branches, then it stands to reason that you would believe that other branches are “real” in the same sense that ours is (even though the others may be forever unobservable).

I am not sure how you would classify your own stance in terms of (A) and (B) above. I fear, though, that you might say that you’re not agnostic between (A) and (B), but you think that both (A) and (B) capture aspects of the situation. :/

163. Mark Says:

Two thoughts:

(1) It seems Einstein would have been a fan of MWI. Perhaps, if he were around today, he would’ve amended his famous statement to “God only *appears* to play dice to our feeble minds”.

(2) To me, MWI is as much about parallel worlds as quantum computing is about trying every possibility in parallel. Just as you can interpret a single universal wave function as meaning there are many parallel worlds, you can interpret quantum algorithms like Shor’s as splitting into many different worlds, and making a single trial in each world.

But the results of the various trials are then isolated in the different worlds, which isn’t very useful on its own. Instead, to get anything useful out of the quantum computer, you need to re-cohere the results of the various worlds to get a single answer in a single world. This re-coherence is where the truly interesting stuff happens. Along the same lines, the many worlds of MWI can in principle re-cohere. They many also may be blurred at the edges, rather than truly discrete.

My preferred view is instead that there is a single “world”, namely that of the universal wave function. The many supposed worlds are just an artifact of our ability as observers to only view a single basis state.

164. Steven Evans Says:

Scott Says:
Comment #127 March 6th, 2021 at 8:26 am

“Russell’s teapot also isn’t ruled out by the evidence, in exchange for the massive conceptual problems it would introduce if it did exist.. ”

Based on our cultural knowledge we can conclude with the same confidence of a Physics experimental result or Mathematical proof that Russell’s teapot isn’t there (with significantly more confidence in the case of proofs of the ABC Conjecture). The clinching evidence is not from Physics.

“I feel exactly the same way about superdeterminism.”

Why? Physical evidence is the only relevant evidence here. MWI, superdeterminism and pilot-wave theory all potentially solve the perceived measurement problem. From a human psychological perspective, distinct selves exist across the time dimension so why not across quantum branches; and apples always fall and never rise, so equally why would we be troubled by a superdetermined universe? I sometimes see people stating they “like” or “favour” MWI, but I don’t understand what is meant by comments like that – it’s Physics not Facebook.

165. Steven Says:

Nobody has talked about how Claus Peter Schnorr is the evil twin of Peter Shor. Honestly, I am shocked.

166. Steven Evans Says:

Greg Kuperberg Says:
Comment #104 March 5th, 2021 at 7:56 pm

“experienced mathematicians tend to be suspicious of many-worlds stories, which could equally be included in classical probability,”

But no-one has ever observed physical evidence that a die has a superposition of faces facing upwards prior to checking the result of the roll, so there is no need to include MW in classical probability qua theory of die rolls. Physical evidence of “wave collapse” has been observed, though, so there is a motivation for suggesting MW in complex probability qua theory of sub-atomic physics.

167. Omer Says:

Why isn’t the fact you’re saying “A quantum state is a wave-function that evolves according to the Schrodinger equation; the rest is commentary” makes you simply a many-worlder (albeit in denial….)?

Don’t you agree that this is the claim to fame of the Everettians? All other interpretations suggest that something is missing in the theory (whether the wave-functions are an incomplete description of the physical reality, or the Schrodinger equation is an incomplete description of the way they’re evolving).

168. wolfgang Says:

Scott,

so far you left out (*) the one obvious interpretation: solipsism.

I am “the classical observer” of Copenhagen, there is no other.
Everything else that appears in my experience is described by quantum theory and only real to the extent that I observe it.
This is the simplest interpretation using Occam’s razor; zero worlds are easier to believe than many worlds.

There is one unexplained mystery: Where did I come from?
But every theory and philosophy is incomplete (where did your universe and quantum theory come from?) – postulating the existence of a football sized classical neuronal network would be sufficient to explain my experience. But why postulate things I cannot observe, even in principle?

You asked what “breathes fire” into quantum theory to make things real and the obvious answer is that I am what makes things real; I can make things exist.
E.g. my “tooth brush” does not exist now (i think it is in a quantum superposition) but if I would “walk” into my “bathroom I could make it exist in my conscious experience.
Btw i call this power “free will” but this is another topic …

(*) But I just made it appear on “your blog” 😎

169. ppnl Says:

Scott,

late to the party but…

Does wave collapse ever have to happen? The thing about decoherence is that it does not cause collapse exactly. It only causes a decoherence in the wave relationship. This makes observing coherent quantum effects impossible.

Say you have a cat in a box in a superposition of alive/dead. How can you tell? You can’t really. But say you have two boxes with cats that are not only alive/dead but their states are entangled so that when one cat is alive the other is dead. Now you can in principle devise an experiment to detect the superposition. What happens if something corrupts that coherent phase relationship between the cats? The cats become indistinguishable from classical cats that are in a particular state.

Maybe all cats everywhere are in a superposition of many states but because there is no coherent wave relationship between them they seem classical. Maybe the macroscopic world around is is composed of a vast network of superpositions and entanglements that appears classical simply because it is so complex.

We don’t need no collapse.

How does many-worlds deal with subjective collapse? For example if you have a cat in a box then the state of the cat is not determined until you open the box and the universe splits. But what does the cat think of this? From its perspective it was always in a particular state and there was never any reason to split the world. Surely it is entitled to think of itself as an observer making a measurement. So does that alternate reality exist or not?

Many-worlds is a good way to visualize things. But I think it has problems with subjective collapse and I think we need subjective collapse.

170. Aspect Says:

Scott #96:

Thank you! This seems like a simple yet important concept, to me at least 🙂

171. Mateus Araújo Says:

Greg Kuperberg #158: Historically, what Bohmian mechanics accomplished was to demonstrate that one could have a quantum theory that applied at all scales, even if it was not particularly appealing. Remember, back then the consensus was that quantum mechanics only applied to the microscopic domain, while the macroscopic domain had to be classical. A more modern view is that Bohmian mechanics is just Many-Worlds but with the hidden variables breathing the fire of reality in one of the branches of the wavefunction. See Deutsch’s take, for example. Incidentally, I think this is what it takes to consider some branches real and others not. Simply declaring that some are not real without having any physical difference doesn’t cut it.

About classical many-worlds, I think I agree with the underlying reasoning, but I don’t see the point of it. Pretty much any macroscopic event that we see as random derive its uncertainty from quantum mechanics, if we go back far enough in time. The 2020 US election might have depended on a mutation in a virus in a bat in China, a clearly quantum event. But after it happened, it is a matter of fact that we already are in that branch of the wavefunction, the uncertainty that remains is in which branch we are. And that’s classical uncertainty. Or to give a more prosaic example, before we pass an atom through a Stern-Gerlach apparatus there’s quantum uncertainty on its spin, but after it has hit the screen mere classical uncertainty remains about which branch we are in. Sure, it derives ultimately from quantum uncertainty, but why model it that way explicitly? Seems like a lot of work for no gain.

About the digits of your favourite transcendental constant, that are the same in all worlds, I think the blasé attitude of Bayesians really helps here. They don’t care whether the event is repeatable, or even actually random, these degenerate gamblers say its rational to bet on anything that doesn’t lead to a sure loss.

172. ARaybold Says:

Scott #58:

When I read “And it seems like too much to take ‘appearance in some mathematical description of the universe’ as a sufficient condition for consciousness”, I nodded in agreement, but then I thought, are there other words we could substitute for ‘consciousness’ here, such as ‘real’ or ‘physical’, to yield a proposition that is no less well-defined, and no less (or more) plausible? In particular, if one were to accept that ‘appearance in some mathematical description of the universe’ is a sufficient condition for being physical (in whatever sense) but not for being conscious, then would these other versions of ourselves be p-zombies in a way that David Chalmers may never have imagined?

173. A1987dM Says:

@Bepis #105: sounds like something Eliezer Yudkowsky could write.

174. Alex Says:

Scott: so would you consider yourself an instrumentalist then? Are you saying you think it’s a mistake to try to figure out whether the formal concepts of Quantum Theory ‘really’ refer to reality or experience or something else (and if so what the nature of those entities might be such that they’re described by QT), except insofar as doing so helps the physicist in their research (i.e. be an Everettian on Monday, a Bohmian on Tuesday, etc.)? That instead we should simply see QT as a useful calculating tool for navigating reality and predicting the results of experiments, and not expect any more from it?

175. A1987dM Says:

@Bepis #105:

176. Scott Says:

Alex #174: No, I’m emphatically not an “instrumentalist.” I believe understanding is the entire point of science. I believe there’s a big chunk of the measurement problem that can be and has been understood, via decoherence theory. I believe there’s another big chunk that can probably be understood only when we’ve understood the mind-body problem or what “breathes fire” into the universe at all, which (alas) might mean never. I believe different interpretations can help one along the path to these realizations, although (as I said) for most students Everett is the best in my experience.

Now meditate on this until you’ve achieved enlightenment. 😀

177. JimV Says:

An analogy occurred to me which might or might not work: you are writing a computer program that has to consider multiple cases and has a tangled flowchart. You wonder if it isn’t possible to do it more simply with fewer branches. If so, you would have a better interpretation of the algorithm, one that is easier to understand and explain to others. However, if you’ve pored over the program a lot and it is burned into your brain, you may decide it works well as it is and you don’t need a better interpretation.

Thanks for the reply on the superdeterminism. Probably you are right, but I am confused as to why other smart people have written papers saying it doesn’t have to be a conspiracy theory. (I don’t think either side has a rigorous proof at this point, and I get that if you have a strong feeling/intuition you don’t see the need to spend time thinking about it.)

178. Scott Says:

JimV #177: FWIW, I’m also confused as to why people write papers about it! But at least it’s useful as a litmus test: once someone starts talking superdeterminism, I know to take less seriously anything else they say about foundations of QM.

179. Scott Says:

On reflection, maybe the best way to state the Zen Anti-Interpretation is that, to whatever extent there’s such a thing as the “measurement problem of QM,” it neatly decomposes into two subproblems:

(1) The problem of understanding “measurement” in terms of unitary interaction among system, observer, and environment — a problem that’s to a large extent already been solved.

(2) The quantum-aware version of the mind-body problem (“but what breathes fire into one of the outcomes and makes it experienced? is fire also breathed into the other outcomes?”)

180. Scott Says:

Omer #167:

Why isn’t the fact you’re saying “A quantum state is a wave-function that evolves according to the Schrodinger equation; the rest is commentary” makes you simply a many-worlder (albeit in denial….)?
Don’t you agree that this is the claim to fame of the Everettians? All other interpretations suggest that something is missing in the theory (whether the wave-functions are an incomplete description of the physical reality, or the Schrodinger equation is an incomplete description of the way they’re evolving).

If I’m just an Everettian in denial, then by the same standard, almost every informed person nowadays is likewise just an Everettian in denial! Besides Penrose and GRW, I don’t know anyone today who thinks of “collapse” as an actual physical thing. Certainly not the Copenhagenists or QBists! The way they’d put it is that for them, quantum states are personal, “collapse” is just Bayesian updating, and there’s no such thing as the “quantum state of the universe.” The way I’d put it is that, if you force them (kicking and screaming) to write down a quantum state that includes human beings, they’ll simply deny that the fire of experience gets breathed into more than one of its branches.

181. Scott Says:

Mark #163:

It seems Einstein would have been a fan of MWI.

No doubt the MWIers would love to think that! All we know is that he hated Copenhagenism and had 30 years to mull it over while being Albert friggin’ Einstein, and he never suggested anything like MWI. (On the contrary, when he wrote to Schrödinger about gunpowder in a superposition of exploding and not exploding — the forerunner of Schrödinger‘s cat — he clearly viewed it as a reductio as absurdum rather than a serious possibility.)

My preferred view is instead that there is a single “world”, namely that of the universal wave function. The many supposed worlds are just an artifact of our ability as observers to only view a single basis state.

That’s completely orthodox MWIism! Their real position, AFAICT, is that “splitting worlds” are just a useful picture for the rubes; what really exists is the universal wavefunction.

182. Scott Says:

Why is P vs NP so important? #47:

Obviously if we observed any violation of standard QM, like Bell inequality violation failing over large enough distances, that would completely change every aspect of this discussion. The discussion is premised on the assumption that QM will continue passing every single experimental test without exception as it has for the past century. If you don’t like that assumption, you should probably find a different thread. 🙂

183. Ted Says:

In the first half of your post “Why Many-Worlds is not like Copernicanism” from several years ago, you point out that the MWI is quite “brittle” to true experimentally measurable deviations from quantum mechanics. Do you think that certain interpretations of QM are more “robust” to experimental deviations from orthodox QM (e.g. a slight deviation from the Born statistics, a slight nonlinear modification to the Schrodinger equation, a dynamical collapse mechanism, etc.)? If so, do you think that’s a reason to favor those intepretations?

More concretely: if tomorrow an experiment demonstrated a slight but robustly significant deviation from the postulates of QM occuring in the real world, then would that make you any less Zen regarding interpretations of QM? (Of course, many even slight deviations from orthodox QM immediately open up a Pandora’s box of “challenging” implications, like superluminal signaling, efficient solutions to computationally complex problems even beyond NP, etc. – so the burden of proof of experimental significance would need to be very high.) If so, which one would you move toward, or would that depend entirely on the details of the experimental discovery?

184. Why is P vs NP so important? Says:

Scott 182. I think you are orthodox if it comes to science. No bell violation inequalities are tested for multi party situations nor is the assumption P is NP is entertained? How can you assume something in theory if the entanglement has not been tested for multiple parties? Even for two party there is a lot of headache about freewill and the theory stands on whatever footing it can hold on to. If three or further parties you are talking of cheaters, colluders and a whole bunch of mightmares. Perhaps it should be correct and I think it should be correct. But think about the impact on science of it fails. What is the standard interpretation in theory if an entangled particle collapses a coherent system (of say bosons) it is part of, if it’s partner entangled particle in a gazillion miles outside the galaxy is manipulated?

185. Why is P vs NP so important? Says:

Perhaps the system the particle is part of causes a chain reaction or may be the entangled particles are coupled in a way such that the system the first particle experiences a chain reaction iff the other does not (these might cause defects in mass and so perhaps there is an effect of time we are missing which can remove simultaneous action at a distance by a ‘causative’ reaction). Perhaps there is a smart person (girl or guy or gender generally great) can concoct something unorthodox and still experimentable which can shred current theory? Why should we believe all future experiments in qm would be consistent to current orthodox knowledge?

186. Scott Says:

Ted #183: Yes, as I just said, if the slightest experimental deviation from QM is ever found, then the entire philosophical discussion changes completely.

In other words, this is not an issue for MWI that’s not shared by literally every other interpretation. It does, though, create a disanalogy between MWI and Copernicanism, as I said in that old post.

187. Scott Says:

Mateus #138:

Sure, Bohmians are the original epistemicists, and they will never die.

Sorry, I had meant to ask you about this earlier. I thought Bohmians believed in the reality of the wavefunction just as surely as Everettians did! It’s just that they view it as merely a “guiding field” for the hidden variables.

Or by “epistemicists,” did you just mean that they ascribe a probability distribution to the hidden variables?

188. fred Says:

To see the crucial role of memory in subjective experience, or at least why it can’t be dissociated from consciousness.

After a night of sleep, how would I really know which one of those two alternatives happened:

1) I was unconscious throughout the night, until the moment I woke up and I was conscious again. I.e. throughout the day I’m able to consistently remember the experience of waking up, but anything before that is a blank.

2) I was actually somewhat conscious throughout the night, having awareness of the noises in the room, my own breathing, and experiencing thoughts and emotions in dreams. But as I woke up all the memories of those experiences vanished entirely.

Similarly, there could be two alternative types of anesthesia during surgery (I don’t mean they’re both real):

1) I’m truly totally unconscious during surgery, i.e. no experience is felt at all. The lights are out.

2) I’m totally paralyzed during surgery, but can still feel every sensation of the surgery. But there’s no long term memory associated with those sensations, i.e. they vanish within seconds. So, after I wake up, I do feel like I was unconscious throughout the operation.

This is meant to illustrate that memory is an important ingredient of consciousness.
Imagine we could only retain experiences for x amount of time. By how much could we reduce x (down to 1 day, 1 hour, 1 minute, 1 sec?) and still claim we’re conscious?

Similarly, what would it feel like to have artificial memories implanted in our brain? I.e. events we never experienced, but we now swear we actually experienced them.

189. Omer Says:

Scott #180:

I would classify any information-centric interpenetration as non-realist, as they all inherently (and proudly) claim to model something else than the objective physical reality (if they even admit such as a thing to exist).

As far as I can see, if you subscribe to such a view you must either denounce the existence (or well-definedness) of “objective reality”, or admit that QM is incomplete (simply since it doesn’t model whatever it is that’s really happening).

But you’ve expressed a disdain towards instrumentalism, and you seem to believe that QM is a complete theory – and therefore (assuming the dichitomy above makes sense) I guessed that (modolu their pedagogical value) you don’t take seriously such interpenetration. Was I wrong?

And as for the realistic interpenetrations, I was under the impression that they’re all (with one notable exception, which is the many-worlds) revolving around some supposedly missing-pieces in QM. Besides Penrose and GRW, you have e.g. the pilot wave theory or superdeterminism (for which I fully share your sentiments…) that (unlike your examples) dismiss the notion of “collapses” and instead suggest that QM is incomplete and some other extra mechanisms is required to explain our observations.

I can’t see how you may escape the unpleasant choice forced upon you by taking QM realistically: either many-worlds is realistically correct, or else QM is obviously incomplete (“obviously” because given any experimental setting – it’s missing the part that predicts or explains your particular observations).

Do you have an example for an objective (e.g. non-Bayesian and non-information-theoretic) interpenetration, which is not essentially Everettian and yet claims that QM is complete?

190. Scott Says:

Tim Converse #162 and others asked me (essentially) whether I ascribe reality to galaxies outside my past and future lightcones—the implied question being, if I do, then why not the other Everett branches?

On reflection, as long as there’s a chain where I can talk to X, who can talk to Y, etc. who can talk to Z, then I ascribe reality to Z even if I can never talk directly to Z. Note that this condition is satisfied for faraway galaxies (and for, e.g., the ancient Greeks and for our far-future descendants) but not for the other MWI branches.

I’m very comfortable with a “physical reality” consisting of the set of things I could stub my toe against, or that a friend told me their cousin stubbed their toe against, or that left a track in a particle accelerator. I’m equally comfortable with a “Platonic reality” consisting of all positive integers, computable mathematical structures, logically conceivable universes, and so on. It’s only when people insist on an intermediate category between those two—i.e., on things that have more than a Platonic reality, even though there’s no chain connecting them to anything I could stub my toe against—that I start getting nervous! 🙂

191. Scott Says:

Everyone: Thanks for the great discussion! You’re welcome to continue it here, but I fear I’m now going to do one of the main things recommended by the Zen Anti-Interpretation, which is to stop talking for a while about the Zen Anti-Interpretation. 🙂

192. Why is P vs NP important? Says:

I think the time is not ripe for a new discovery.

193. Scott Says:

Omer #189: I don’t mind if you call my view “essentially Everettian”! But I also don’t mind if someone else calls it non-Everettian. 🙂

194. Why is P vs NP important? Says:

Can we consider

$BQNC=ClassicalC_NC+Entanglement$
$BQP=ClassicalC_P+Entanglement$

to be correct? If so what are the classes ClassicalC_NC and ClassicalC_P?

In general is $BQ\mathcal C=ClassicalC_\mathcal C+Entanglement$ correct?

195. Scott Says:

#194: I don’t know what it means to “add entanglement” as an operation on classical complexity classes, nor do I know what you mean by C_NC. And please no more off-topic questions!

196. Philosopher Eric Says:

Hello Scott,
A good friend reblogged your post, I suppose to help support his Everettian sympathies given that I and at least another of his friends have been quite skeptical. Anyway the question arose of whether or not you’re a full “Sean Carroll” type of supporter who thus believes in the existence of these other worlds, or merely consider this perspective an effective pedagogical tool?

(My criticism of a beyond pedagogy account is that this runs into the situation of other universes also emerging (or whatever) from the ones that spawn from ours, and so an infinite regress that could only be accounted for non-causally, which is to say, supernaturally. Here’s a link to my commentary, which I provide as proof rather than as something that you or anyone else bother reading. https://selfawarepatterns.com/2021/03/06/an-instrumentalist-everettian/comment-page-1/#comment-136775)

197. 1Zer0 Says:

Fred #188

“Similarly, what would it feel like to have artificial memories implanted in our brain? I.e. events we never experienced, but we now swear we actually experienced them.”

More generalized, is it possible to inject “Qualia information” into a brain which that brain never actually (did experience) LOG_OR (could have have experienced) in the universe? I purposefully formulate the latter that way in order to reference experiences that could have usually not been experienced: Like The mental image of a 34 dimensional space or the impression of a color never seen before in the respective person’s mind? If inside the mind it’s all just turing-computable information (Something I certainly disagree with), new qualia should be injectable into the mind. Good luck with that.

In converse, Semi-Philosophical Zombies like people suffering Aphantasia (https://www.facebook.com/notes/blake-ross/aphantasia-how-it-feels-to-be-blind-in-your-mind/10156834777480504/ ) which has only been started to be investigated academically in recent years certainly lack some phenomenological properties believed to be part of every mind.

198. Shmi Says:

I’m quite surprised that most discussion of the Born rule ignore one apparently essential thing: gravity. (Well, Penrose is one exception.) Though even Sean Carroll, as hard-core Everettian as anyone, sort of acknowledged its importance:

https://twitter.com/seanmcarroll/status/1363611156493950979

So… why is gravity likely to be important in the future progress on the measurement and emergence of apparent classicality? A few reasons:

* it is nonlinear
* it is intimately and mysteriously related to entropy, a property of macroscopic systems that
magically becomes important at the scale where the quantum effects are all but unobservable (somewhere below or around one Planck mass, or maybe around a trillion-fold ground state degeneracy?)
* it is related to holography, connecting local and non-local effects
* it points to various issues in our current understanding of the Quantum (e.g. the black hole information paradox)

I am confused why these experimental and theoretical hints do not seem to create more interest in trying to link the two in the low-energy many-state domains. There is some experimental work on mesoscopic cat states (https://www.nature.com/articles/srep13884) but I cannot find any theoretical work predicting what the experiment might find. Weird.

199. Mateus Araújo Says:

Scott #187: Sorry, Bohmians do believe in the reality of the wavefunction. Leaving that out made my comment rather confusing. But for them it is just a guiding field to evolve the Bohmian positions, which is the observable component. Now these Bohmian positions are distributed with probability density $$|\psi(x)|^2$$, so the quantum state is also a representation of our ignorance about the real position. I think calling them epistemicists is fair, though perhaps an oversimplification.

200. mjgeddes Says:

Scott >>:

>>(1) The problem of understanding “measurement” in terms of unitary interaction among system, observer, and environment — a problem that’s to a large extent already been solved.

>>(2) The quantum-aware version of the mind-body problem (“but what breathes fire into one of the outcomes and makes it experienced? is fire also breathed into the other outcomes?”)

I don’t think (1) and (2) are different mysteries, a solution to one should entail a solution to the other.

Scott >>:

>>I’m very comfortable with a “physical reality” consisting of the set of things I could stub my toe against, or that a friend told me their cousin stubbed their toe against, or that left a track in a particle accelerator. I’m equally comfortable with a “Platonic reality” consisting of all positive integers, computable mathematical structures, logically conceivable universes, and so on. It’s only when people insist on an intermediate category between those two—i.e., on things that have more than a Platonic reality, even though there’s no chain connecting them to anything I could stub my toe against—that I start getting nervous!

There’s three levels here:

(1) The platonic , timeless level of pure math (all logically possible worlds)
(2) The applied math of computer science (constructive math)
(3). The block-universe (including universal wave function) of physics

Computation, minds and dynamical systems (in other words, the world we can consciously experience) are all on level 2. The mind-body problem is how level 2 relates to levels 1 and 3.

I don’t think (1) and (3) have the same status as (2). They only ‘exist’ in a weaker sense. It seems that there has to be some *measure* of the strength or degree to which objects can be said to ‘exist’ (that is to say, the levels mentioned above are really a continuum).

It seems that platonic math worlds and block-universes are more like the scaffolding of reality. The objects referred to perhaps do ‘exist’ in a weak sense, but aren’t a complete picture, that is to say, the world of conscious observers ‘exists’ in a stronger sense. But what sense?

I’d say that the ‘extra’ ingredient is *time*. Whereas levels (1) and (3) are timeless, the world of our experience is all about processes unfolding in time. And we can carefully decompose the concept of ‘time’ into 3 somewhat different concepts, which I’ve already talked about: Causality, Complexity and Compositionality:

CAUSALITY: Probability theory in its fullest sense is really about cause and effect and how to do prediction, retrodiction and imputation. We don’t just want to know about correlations between things, we want to know about causes and counterfactuals, which outcomes are possible, and how would those outcomes change if we intervene in some way.

COMPLEXITY: Coding theory in its fullest sense is about dealing with complexity. We want to compress our representations of the world, to find efficient encodings to deal with limited resources in terms of space and time and limited information. In the real word, we are confronted with complex adaptive systems, and these embody a mix of randomness and determinism that makes them complex. How do such systems achieve open-endedness, efficiently exploring and creating new possibilities ?

COMPOSITIONALITY: Constructive logic in it’s fullest sense is about compositionality: how are large systems built from smaller ones, and going in the other direction, how do we manage to split the world into smaller parts, objects and the relations between them? Mereology studies the relationship between the whole and its parts. We want to know how to engineer and combine ontologies based on the principle of compositionality.

I think that what ‘breathes fire’ into a ‘possible world’ then, is math with the added 3 ingredients I outline above (so computable math structures with causality, complexity and compositionality). Thus, the measurement problem is not really separate from the mind-body problem.

201. Tim Converse Says:

Scott #190:

[Scott makes clear in #191 (two minutes after replying to me and others) that he he’s done with the discussion. Oh no! I do hope that I am not the one who drove him over the edge. :/ So my reply is to whoever is left, and I will address Scott in 3rd person rather than 2nd.]

Scott’s response:

“On reflection, as long as there’s a chain where I can talk to X, who can talk to Y, etc. who can talk to Z, then I ascribe reality to Z even if I can never talk directly to Z. Note that this condition is satisfied for faraway galaxies (and for, e.g., the ancient Greeks and for our far-future descendants) but not for the other MWI branches.”

I don’t understand in what sense this condition is satisfied for faraway galaxies, and suspect that it gets its seeming plausibility from the commonsense notion of “talking to” which is bidirectional.

Take galaxy Z that is outside our observable universe. We can’t “talk to it” (in the sense of observing it directly) because it’s outside our lightcone. But we also can’t observe it in an indirect sense by talking to someone who talks to someone who talks to it, for the same reason (outside our lightcone). Information cannot get from it to us, or even to future us.

Yes, in some sense we can talk to someone at half the distance, and they could presumably talk to galaxy Z, so if we’re OK with a transitive relation of “talking to” that is also loose about directionality, then we’re fine. But note that this already assumes the existence of Galaxy Z. We are assuming that if Galaxy Z exists then our intermediate relay partner could enjoy (or have enjoyed) a conversation with it. But for all we know, galaxies actually wink out of existence the moment they go outside our own personal lightcone (which is the question at issue). Our relay partner will never be able to confirm or disconfirm that for us.

I think that this situation is still analogous to our potential relationship with other branches. If another branch exists that diverged from ours half an hour ago, then I (Tim) have a transitive talking-to relationship with Tim-prime from the other branch. This is because Tim[one hour ago] has “talked to me” (Tim[now]) via memories, and Tim[one hour ago] has also talked to Tim-prime[now] the same way. (Granted, the links are unidirectional in passing information, and don’t make a unidirectional chain that can pass information, but that was also true in the Galaxy Z case.)

So it seems that if Galaxy Z exists then we may enjoy a transitive talking-to relationship with it, and if Tim-prime exists then I may enjoy a transitive talking-to relationship with him. In both cases we have to assume the existence first, and we get no evidence about that existence from the the supposed transitive relationships.

I don’t understand the essential difference between the two cases.

202. fred Says:

1Zer0 #197

People who’ve tried psychedelics often claim to have experienced unknown qualias (like new colors).
Seizures, meditation, or auto-hypnosis can also trigger experiences that are often very hard to translate into normal experiences.
When I was a kid, as a game, lying in bed at night I would often first focus on the sensations in my teeth, which I could then slowly morph into very strange sensations of massive spaces and extreme speeds. It was both scary and exhilarating.

203. Jeremy Says:

Scott #110

That wasn’t what I was trying to get at. I was more trying to get at the “Why can’t we hadamard transform a brain?” question, but substituting a simpler classical AI algorithm for the “brain”.

Imagine that we had a fully classical algorithm that could pass the Turing test. If we ran that algorithm on a quantum computer, and fed it a superposition of inputs, I could imagine that we might be able to observe the interference in some cases, and then query the algorithm afterwards for questions about its subjective experience.

Unfortunately, A) we don’t have any AI algorithm that passes the Turing test, and B) The AI algorithms we have still are probably too complex to reason about their behavior in superposition.

But I am wondering if there is some simpler algorithm we could substitute in place, that still captures some of the fundamental properties we care about (such as ability to remember some experience that it can report on later), but which is simple enough that we could analyze its behavior easily. For example, an algorithm which just stores a few basic facts about the inputs it is fed, and can remember them later.

I am wondering if someone has worked out such a thought experiment, or whether it doesn’t make sense (is it too hard to get interference in a system that is doing anything nontrivial?)

204. AJD Says:

If I’m just an Everettian in denial, then by the same standard, almost every informed person nowadays is likewise just an Everettian in denial!

The somewhat obnoxious response is that many of the informed people are Everettians who are not in denial.

I think it is largely true that most of the people thinking seriously about the foundations have converged to something very close to Everettian program. Nowhere near all, by any means. There actually are people who have thought through and hold very different views about the roles of measurement, information transfer and apparent collapse.

But this large majority who do answer more or less like Everett are for whatever reason is not willing to call it MWI, and unwilling to acknowledge how different the earlier conceptions were from the modern consensus, largely in the ways that Everett talked about.

Of course, I think the interpretational issues are also a bit less interesting than I used to think. After all, quantum mechanics doesn’t actually hold. It’s merely an outstandingly effective theory, and the best ways of sorting what’s epistemic or ontic about it need not correspond well to “what’s really going on”. We really need to look at these questions in the context of quantum field theories and their interaction with GR. But if you think QM has hard interpretational problems, be prepared to be shocked at that lack of interpretation in QFT even in flat space-time, much less on or interacting with curved space-times.

That’s a much harder problem, though what do I know as a failed physicist?

205. Wyrd Smythe Says:

Tim #201: FWIW, I see galaxies beyond our horizon as real because I assume the universe is homogenous and an observer in a galaxy just inside our horizon sees a horizon that includes us in one direction and galaxies we cannot observe in the other. Likewise an observer in that most distant galaxy must see their own domain within their horizon.

A difference in multiple worlds is that these worlds aren’t distant but coincident with us, so the situations seem physically different to me.

206. mjgeddes Says:

Growing Block Universe Model with Reality as a Sphere of Knowledge
—–

Consider reality to be a ‘sphere of knowledge’ (or ball of knowledge). All logically coherent concepts exist as regions on and within the sphere of knowledge.

The surface of the ball is about purely *ontological* concepts (it represents only things that are ‘objective’) and it’s not about particular worlds, it only represents the *multiverse*. It’s the *objective* information content of reality, the ‘scaffolding’ of reality so to speak. But as you move into the interior of the ball,*epistemological* content (cognitive content) is added. So the interior of the ball is a cognitive space that starts to mix in *subjective* concepts. But as you move into the interior, *particular* worlds (or branches of the multiverse) start to get represented.

The crucial part of the model is that the concepts are ordered according to complexity. The surface of the ball represents the simplest layer of concepts, and as you move into the interior layers of the ball , the complexity is going up and up. It’s intended to represent a *growing block universe* , where the interior layers of the ball are still *under construction*. Reality starts with the pure information on the surface layer and the advance of time is equivalent to the *filling in* of the interior layers.

Now the geometry of this would be highly peculiar, because the complexity goes up and up as you move into the interior of the ball. In a very real sense, the ball is ‘bigger on the inside than the outside’ (in the limit that you reach the center, the complexity has gone to infinity). I think perhaps something like a fractal, but more complex.

Now let’s see how quantum mechanics is represented. The universal wave-function is equivalent to the surface of the ball, it’s maximally simple, and it’s pure objective information content. It represents the multiverse as a whole.

Applied math (computer science) is equivalent to the geometry and topology of the ball, it’s about the *paths* from the surface of the ball into the interior. These paths represent both *time evolution* and *cognition* (mental concepts).

Physical reality is the mixed states represented by the interior layers of the ball. As one moves from the exterior of the ball , reality is crystallizing from the pure states on the surface, to the mixed complex macroscopic worlds in the interior. The geometry and topology of the paths from the surface into the interior is cognition, representing the growth of knowledge.

207. Steven Evans Says:

Scott Says:
Comment #127 March 6th, 2021 at 8:26 am

” Russell’s teapot also isn’t ruled out by the evidence. But we have no positive reason whatsoever for supposing it to exist”

As I wrote, we have a positive cultural reason to suppose Russel’s teapot doesn’t exist. Do you have a positive reason for your feeling that superdeterminism can’t be true? I ask out of interest, not just to bug you while you’re recovering from the jab.

I think the point in all this is that the wave function does reflect what we call physical reality, for example it tells us what percentage of electrons will make it over a potential barrier in a chip. But even if one bears in mind details like the decoherence reflected in the density matrix, and so on, there is still a discontinuity in the theory which is a puzzle. One can argue about the extent to which MWI, superdeterminism or wave-pilot theories are currently scientific, but they do fit roughly at least with current empirical evidence, and they are an attempt to answer a real puzzle. None of them have been ruled out and I’ve never heard of any Russell teapot-like issues.
This contrasts with the multiverse, say, which isn’t an attempt to answer any known problem in Physics. If MWI is the result of smoking dope listening to the Beach Boys, then the multiverse theory is the result of a bad trip.

208. gentzen Says:

JimV #177: FWIW, I’m also confused as to why people write papers about it! But at least it’s useful as a litmus test: once someone starts talking superdeterminism, I know not to take seriously anything they might ever say about foundations of QM.

Oh, I didn’t realized this was directed towards me. I was tempted to say some more words on my general attitude towards superdeterminism anyway. So this is as good an excuse as any. To “prove” that I didn’t change my attitude on the fly, I copied below the parts from an email which I wrote to Sabine Hossenfelder and Tim Palmer a long time ago (25. October 2020):

“you recently published a paper about “Rethinking Superdeterminism” with the explicit goal to make progress on the measurement problem in quantum mechanics. I strongly believe that your paper will indeed enable progress, not just in the sense of finding models without a measurement problem which are able to reproduce all currently verified predictions of quantum mechanics, but also in the sense of convincing a significant number of physicists that those are indeed “valid” models.

[…]

However, I write to you because I fear that you opened Pandora’s box by defending superdeterminism (as scientific), without fully realizing the consequences. I realized them after seeing how emergent superdeterminism not just explains the paradoxes related to (the
absence of) “counterfactual definiteness” in quantum mechanics, but also gives plausible explanations for Bohr’s “blurry” complementarity concept, and also plausible explanations for why reductionism fails in general. But just because superdeterminism provides plausible
explanations does not mean that those are scientific explanations. Those explanations are not unscientific, but superscientific (überwissenschaftlich), superphysical (überphysikalisch),
metaphysical, and in the end may belong more to philosophy than to science or physics. (I do like philosophy and metamathematics, so please don’t interpret those words too negative.)

The repeatability of independent experiments is an important ingredient for science. Just as it would be wrong to claim that things must be independent because we could no longer do science otherwise, it is just as wrong to claim that we can continue to do science as usual in the presence of such dependencies (or non-repeatabilities). The placebo effect in medicine is a good example of this. It forces us to go beyond the scientific method(s) by using blind and even double blind studies (in addition to control groups, which are still part of normal science). You might protest that scientific methods are not set in stone and are allowed to adjust to new circumstances. However, one of the goals of science is to achieve objectivity by adhering to methods which allow to converge to consensus in the end. Superdeterminism might allow to explain why certain paradoxes do not need to be resolved, but I fear it does not allow to converge to consensus about the consequences of such explanations in the end. (Sabine Hossenfelder herself explained why the multiverse hypothesis is not science, and explanations invoking superdeterminism might have similar issues.)

[…]”

209. 1Zer0 Says:

Fred #202,

In retroperspective, I do agree. I am a lucid dreamer given that I have enough sleep (9h, rarely lately) and I have to admit that I experience some impressions which I did not experience through my senses awake. Allow me to clarify my previous statement; If the mind is a computational device in the Turing sense and no other ingredient is responsible for qualia, then you should be able to change the Turing-Machine lookup table describing the mind algorithm or construct inputs into that Mind algorithm in such a manner that the internal experience of the respective mind is now a 11D, 13D or 4423D dimensional space.
Or a sound never heart before, or maybe some entirely new sensation which is not visual, a feeling, sound, flavor, smell or any “linear combination” thereof. To make the reductio ad absurdum even more striking let’s say we run that mind algorithm given as TM lookup table, encoded as string or something on a mechanical computer running at 1HZ yet with enough memory. Thus you can give this mechanical computer any sensation, even the ones we humans usually do not experience – An Absurdity of the highest degree, in that sense I meant that you cannot “construct information” that will trigger new qualia.

Did you try if you can still trigger that experience with your teeth?

210. Benjamin Feddersen Says:

I just borrow Freeman Dyson’s general take, which is that classical physics is about the past, and quantum physics is about the future. They both exist, but they’re different kinds of things, and so need to be described differently. It covers all the bases and is perfectly intuitive; it’s only recently that we managed to convince ourselves that there is no such thing as the difference between past and future.

211. ARaybold Says:

Tim Converse #201:

I think those are fair points, and it seems that there’s an element of Occam’s razor here: given some unverifiable claim about the universe, the most parsimonious assumption is that those physical laws, that have been verified to the extent possible up to now, apply in that case also.

From this point of view, any response to many-worlds that assumes some incomplete sort of reality in the alternative worlds is triply extravagant: not only in the number of worlds but also in the number of different physical realities and the number of ways in which one branches from another. Also, if the other realities are restricted versions of ours, what makes ours special? (Though maybe this issue is a form of the anthropic principle: if we were not on the fully-real branch, we would not be contemplating these issues.)

I do not think this is an argument either for or against many-worlds, but more like a possibly-falsifiable reason for caution over partial acceptance of it.

I would also like to add (with my tongue almost, not entirely, in my cheek) that if, per Bostrom, we are in a simulation, then neither the winking-out of galaxies crossing our light cone, nor the loss of information in a black hole, nor either the existence or non-existence of Everett’s branches, would be problematical.

212. J Says:

Happy women’s day everyone!

Wyrd Smythe #205,

“an observer in a galaxy just inside our horizon sees a horizon that includes […] galaxies we cannot observe”

Not exactly. As you know the light we can see from a galaxy just inside our horizon is also as old as it can. At the time this light was emitted, this galaxy couldn’t see anything that is not within our present horizon. Let’s call this galaxy Sarah. Assuming everyone stay still (rest frame == CMB), then at some point Sarah will be (will have been?) able to see galaxies that are outside our present horizon. But at this point either these galaxies will also be inside our horizon or Sarah will have quit.

Tim Converse #201,

Here’s a difference: hypothetically we could start traveling to faraway galaxies with a few g constant acceleration (say we have some help from all these grabby aliens), then, a few dozen years of proper time later (meaning a few billions light-years from here), we could check whether we still see far-away galaxies. To the contrary, there is no physical way to communicate with a different slice of the many worlds.

PS: Yes, quantum suicide: we could sit on our butts and wait, then, a few billion years latter, we could verify that we’re still alive, which would strongly support MWI (because hypothetically MWI doesn’t allow suicide). However to my knowledge this idea has not been proven beyond reasonable doubt (e.g. even if MWI was right it might be that there is just no branch of the universe where someone can survive that long).

213. 1Zer0 Says:

mjgeddes #206,

“Consider reality to be a ‘sphere of knowledge’ (or ball of knowledge). All logically coherent concepts exist as regions on and within the sphere of knowledge.
The surface of the ball is about purely *ontological* concepts… ”

vs

“The universal wave-function is equivalent to the surface of the ball”

Even the whole string theory landscape compared to the full extent of existence described by Modal Realism is virtually a tiny drop in an infinitely deep ocean. The extension of “All logically possible worlds” is so beyond measure no human mind could come close to envision most of the exotic realities that MR enables.
I can construct you an infinite set of logically possible worlds that do not have wave-functions if requested. Therefore the universal wave-function can’t be the surface of the ball in your metaphore.

214. Scott Says:

Benjamin Feddersen #210:

I just borrow Freeman Dyson’s general take, which is that classical physics is about the past, and quantum physics is about the future.

That’s a pithy and Dyson-worthy take! But it needs to supplemented by

(1) An explanation of how decoherence, and the Second Law more generally, create the asymmetry between past and future, even though the microscopic laws are time-reversible

(2) An explanation of how quantum physics reduces to classical physics when decoherence is strong

(3) An explanation for why classical physics already had such striking successes at predicting the future (e.g. eclipses, comets)

215. Scott Says:

Steven Evans #207:

As I wrote, we have a positive cultural reason to suppose Russel’s teapot doesn’t exist. Do you have a positive reason for your feeling that superdeterminism can’t be true?

I don’t say it can’t be true. What I say is … look, do you believe in ancient aliens building the pyramids? In the moon landing or 9/11 being faked? If not, why not?

For me, the overriding reason not to believe any of those things is because what they ask us to accept is overwhelmingly more implausible than the relatively prosaic thing they’re trying to explain in the first place. In other words, it’s not that any of these theories are logically impossible, it’s just that they’re ludicrously overpriced in the plausibility economy, like million-dollar toilet seats.

If that makes sense to you, then think of superdeterminism as a googolplex-dollar toilet seat. The only reason to postulate such a thing is to explain the violation of the Bell inequality—a phenomenon that I regard as not genuinely mysterious in the first place. One just needs to accept that the world is governed by quantum probability! Unlike with (say) Wigner’s friend, the mind-body problem isn’t even roused from its slumber!

And yet, to “solve” this problem, superdeterminism proposes a type of cosmic conspiracy (limiting what experiments we choose to do to those that just happen to have certain answers) that, if upheld, could just as well be used to explain the appearance of superluminal signalling, or precognition, or the efficient solution of the halting problem, or basically anything else. I.e., superdeterminism would allow for an arbitrarily magical-looking world, so that the real mystery to be explained would be why the observed world is not arbitrarily magical.

In the plausibility economy, this strikes me as one of the worst deals ever offered. At any rate, to find its like you’d need to venture far from the arXiv, to the realms of tentpole evangelism and late-night infomercials.

216. Chris Says:

Hi Scott! Is there an example of a n-state quantum system (is n=3?), and associated Hamiltonian, that I could use to understand decoherence and see how the Born rule pops out from it?

217. fred Says:

1Zer0 #209

“Did you try if you can still trigger that experience with your teeth?”

Actually I did once last year, when taking a hot bath and meditating!

The data structure of the sensory stream information must be what dictates the sensation (I guess that’s pretty obvious, but who knows).
That’s maybe one clue we get from analyzing AI neural networks: the structure of the neural net (in terms of connections) must reflect the spatial and temporal dimensionality of data itself (e.g. audio data is more 1D/temporal, visual data is more 2D spatial). Apparently the visual cortex is also laid like a screen (I’m no neurologist, so that’s a simplification).
The brain has many areas that seem very plastic and data agnostic, and they grow based on the data presented to them. Like those cases where blind people were able to “see” by stimulating their tongue with a 2D pixel grid.

There’s also not just the dimension of the data that’s intriguing, but also its range. E.g. we can kind of imagine what it would be like to reduce the resolution of our visual perception (a bit like looking at a coarse grid, although we can still perceive wasted space within each pixel) but it’s much harder to imagine what it would feel like to see with a sudden 10x increase in sharpness/resolution (suddenly we get the visual acuity of an eagle).
E.g. look at the dot on top of this “i”, and imagine you could perceive it’s actually a detailed map of the USA.

Another interesting point (at least to me) is that we often take the mystery of vision for granted. That’s because we so easily confuse what we see with the actual reality out there, as though the eye is really just a keyhole into the world, when in fact what we see is a reinterpretation of the world. We suppose there’s a strict homomorphism between visual perception and the world, be we can’t be sure.
One radical modification of visual data I like to toy with is watching a 3D movie, and then switch the stereo 3D effect on and off, with both eyes open (which is subtly different from just looking at something and closing one eye).

218. fred Says:

(*) by “mystery of vision” I meant that it’s very common to hear people say that consciousness has to be related with how things evolve with time, but, when we consider vision, it also becomes clear that we have to be able to perceive at once multiple things that are spatially separated, which isn’t all that obvious when you think about it.
And it’s actually possible (with mediation) to hold together and at once a seemingly huge amount of very different sensations.

And then, to come back to the MWI of things, who knows if consciousness isn’t also able to somewhat cover multiple branches at once, maybe as long as some entanglement between those branches exist, which is probably on an extremely short time scale.

219. Wyrd Smythe Says:

If I may, I’d like to tack on to what I asked in #139.

Every proponant of Everett I’ve met says the interpretation is not Tegmarkian. Even Tegmark divides them into Level III and Level IV type multiverses. Yet the Everett view is that the Schrödinger equation is The Thing, The Whole Thing, And The Only Thing. Long ago when I first learned about the MWI it was pointed out that multiple worlds are no more mysterious than how polynomials have multiple roots, and for a long time thereafter I saw MWI as essentially a Tegmarkian notion.

But proponants say it isn’t, so where is this universal wave-function? How does a mathematical object translate to physical reality? I can understand the Schrödinger equation as a description of reality (like F=ma), but MWI seems to say it’s much more than that.

My understanding is that, given a wave-function for a system at some time, t, it can be evolved forwards and backwards to any other time — the multiverse described by a universal wave-function would be a deterministic one given MWI has no measurement to ever probabilistically collapse to some measurement?

It seems this universal wave-function contains a lot of information if it can describe all possible worlds. Which presumably includes the parts of this world outside our horizon. It seems an amazing amount of information to pack into a single state vector. When Sean Carroll uses his beam-splitter app in a lecture, and then jumps left or right, the two worlds created are hugely similar, yet two audiences have different memories of what they saw. I’m a bit agog at the ability of a universal state vector to embody this. (Perhaps I just don’t understand the theory well enough.)

Mostly I find MWI very confusing. For me it raises more questions than it answers. I would like to understand its implications better.

For future posts, maybe you could tackle the mathematics, in terms of Everett, of the beam-splitter or two-slit experiment? WRT the former, for instance, is there a superposition before the photon experiences the half-silver mirror or only after? WRT the latter, does MWI apply to the interference between paths or, as I’ve wondered, just in terms of where the photon lands?

I assume, since the photon can land in many places, multiple worlds must be created for each possibility? I envision an experiment that fires particles through slits one at a time, they hit a screen of detector pixels, and that screen is repeated as a large display viewed by a large audience (think scoreboard at a stadium filled with observers). After, say, ten particles, are there many, many branches in which many, many audiences have seen a different pattern build?

I would love to see posts exploring MWI in concrete detail!

220. Tim Converse Says:

Wyrd Smith #205: Yes, the situations of other branches and distant galaxies are physically different. However, I think that they play similar roles in their respective theories.

We can’t observe galaxies that are too far away. You believe that they are real because you *assume* that the universe is homogenous in a particular sense. This is another way of saying that the cosmological theories that predict the existence of those galaxies both explain what we can see *and* seem simpler or nicer or better-behaved or contain fewer extra assumptions (e.g. that there is something special about where we are located) than those theories that don’t.

Similarly, we can’t observe other branches. Many-worlds adherents believe that they are there anyway, because they are a consequence of the simplest and most elegant theory that seems to explain what we can see – i.e. the Schrodinger equation, without an add-on theory about measurement collapse.

221. Tim Converse Says:

ARaybold #211:

Sorry, I couldn’t figure out in which direction you thought Occam’s razor was cutting – in favor of granting existence to other branches, or not, or against granting only partially-real status to other branches (which I agree seems like the most complex answer of all).

I think that the standard answer about Occam’s Razor and Many-Worlds is that the Razor is designed to apply to the number of assumptions in a theory rather than to the size of the universe that the theory implies. Everettian interpretations are a rather extreme example of this tradeoff – razoring QM down to the Schrodinger equation alone (without extra special assumptions about measurement) leads to assuming a rather incredible increase in the “size” and complexity of the multiverse.

222. 1Zer0 Says:

Fred #217,

Yes especially considering how we only see a small part of the electromagnetic spectrum. When I am for example starring at an operating CPU, it glows bright, just not in the visual spectrum but in Infrared, an aspect of reality “right under our nose” that we are not able to experience phenomenologically, only people aware of the theory of electromagnetism know there is something there. Our ability to interpret the formalism of physical theories certainly is strongly influenced by the senses available to us. The senses we use to probe and conceive information from the respective theories domain.

This reminds me, I somewhere read of a study concerning people blinded since birth and whether they could still have experiences of color in their mind. Some blind people were apparently inconclusive about the question as they reported some dream experiences which were neither a feeling, a sound, smell or flavor thus might possibly be visual. But of course as they have no reference to what a visual impression is, they could not confirm. Maybe one day when eye sight can be restored for people blind since birth we will have a definite answer whether visual qualia can be experienced without ever having had a visual input from their eyes.

I find it interesting that we can still somewhat “operate” on Qualia impressions. Like mentally changing the brightness of a “blue” experience, making it more or less bright or comparing it to other colors. Or something like CompareLoudness(Sound1, Sound2) – there is almost an ordering within a specific Qualia phenomena which allows it to be compared with other “members” of the respective qualia category. You can compare the loudness of two sounds or the brightness of two colors but the same function would not work when you compare Qualia of different categories with each other like CompareLoudness(Sound1, Red).

223. Tim Converse Says:

Final comment: of *course* it’s not inconsistent to believe in the reality of faraway galaxies on the one hand, and disbelieve in the reality of other branches (or believe they have a more nuanced ontological status) on the other. The theories implicated are very different, and it makes perfect sense that someone might have differing reactions to the two ontological commitments.

But I will say that I either don’t believe or don’t quite follow the argument that bases a belief in faraway galaxies on some kind of transitive talking-to relationship, and if Scott hadn’t moved on from the thread I would have liked to understand that argument better.

Instead, I think that it’s reasonable to believe in faraway galaxies because (as a non-positivist) I think it’s OK to believe in physical theories that have some non-observable entities as a consequence, which means (in turn) that you’re sort of stipulating to the existence of some things you can’t observe, even in some transitive sense of observation. You’re doing this not because you *like* stipulating to unobservable things, but because in other ways the theory beats all the other theories on offer, and if you accept it then you can’t get away from the predictions about non-observables as well.

224. Wyrd Smythe Says:

Tim #220: I’m sorry, but I don’t agree they are similar situations at all. The former is based on our observations and a notion fundamental to science about the isotropy of physical reality. The latter is an almost metaphysical assumption based on an interpretation of a mathematical equation from a theory that we know to be incomplete (because GR and QM refuse to get along).

225. asdf Says:

At some point QM is supposed to be about the real physical universe, in which objective collapse of the wavefunction either is a thing (contradicting MWI) or else it’s not. So that’s unknown factual info, not just interpretation. Similarly the premise of an infinite-dimensional state space seems like the premise of the universe being infinite in age and extent: we currently think its actual age is 13.8e9 years or something like that. Maybe “I notice I am confused” (per HPMOR) will lead to some new understanding of QM that allows Grover’s algorithm to factor numbers up to 1389 bits but no more than that…

There are publications around that purport to explain wavefunction collapse in terms of decoherence, with nobody saying they are right are wrong. I know physicists like to make fun of (e.g.) social science for being fuzzy and noncommittal, but from a math perspective, physics seems almost the same way…

226. O. S. Dawg Says:

Keegan Ryan’s twitter thread is a lot more convincing that C. P. Schnorr’s most recent claims are flawed than the “if the work in fact destroyes RSA he would have surely included some previously unknown factorizations” line of thought I’ve read elsewhere. I didn’t see many record sized primes in the AKS Primes is in P paper!

Somewhat related question: I have read solving SVP is NP-hard under randomized reductions but really have no clue what that means. If I were to tell you that (at times) I believe SVP is in P, can you tell me what the implications of this are? I have to then believe P=NP? I am a crackpot? Something more interesting albeit less true?

227. mjgeddes Says:

#213

Yes, for full modal realism, there would be still more outer layers corresponding to (for instance) different levels of the Tegmark multiverse. I simply started with the quantum multiverse for illustrative purposes.

#214

Yes, there’s definitely something to the idea of equating different aspects of time to different physics levels. The block universe does seem like it’s about the past rather than the future, which is why I’m more partial to growing block universe models, where the future hasn’t happened yet.

I think that my 3 notions of time , causality, complexity and compositionality, correspond to the geometry , functions and topology of a ‘sphere of knowledge’, respectively.

There’s an enormous hole in your own field Scott I think, it doesn’t deal with complex systems. There’s some sort of link between fractals and algorithmic information theory that no one has followed up!

228. Steven Evans Says:

Scott Says:
Comment #215 March 8th, 2021 at 10:48 am

Cheers. I take your point. And I should point out I don’t think MWI, superdeterminism or pilot waves have been shown to be scientific never mind physical.

But firstly does superdeterminism necessarily imply all the implausible things you claim it does? The universe could just be superdetermined to be understandable up to a certain point by the human brain. Superdeterminism could have explanatory power as part of a sub-atomic theory but limited scope. Then, there is a loophole to be explored so there is no harm in analysing it – as you pointed out, pilot wave analysis led to Bell’s insights. Thirdly, it’s a better motivated idea than the multiverse, and if you consider the amount of resources that have been wasted on the dead-end that is String Theory which never made it out of metaphysics, a few quid spent on thinking about superdeterminism might be considered reasonable in terms of funding priorities.
I agree, of course, that it is no biggy to consider we may simply be observing an instantiation of complex probability, but we still don’t know as a physical fact that “wave collapse” is just the separation of superpositioned states into components, and no hidden variables as an explanation of spooky action is dependent on independence of observer and observed. It is in the nature of physicists to follow these scents.
I think also that the idea of separate quantum branches that may or may not have “fire breathed” into them is confusing model with observation. Mathematically separating out the components of a superposition is easy and attractive, but nature has simply not shown us these other branches so MWI isn’t scientific or physical currently. I would say “fire breathed in” = physical. We can’t discuss epistemic v ontic for no observation. On the other hand, describing MWI as an extravagant theory, as was mentioned, is a non-criticism. One wouldn’t think the idea extravagant if one were talking only about a few particles and a few states, so the impression of extravagance may partly be a function of actual physically observed abundance. As long as the Mathematical credit card isn’t maxed out, and we can count the objects being proposed, no probs.
Your basic point is, I believe, that if I and others knew as much about the quantum as we do about teapots, we would see superdeterminism as not even worth considering. But, to save time and effort, I will do a thought experiment – I know everything about the quantum. I still don’t object to some theorists taking a punt on superdeterminism. Any analysis of real puzzles is fine in my book.
I don’t see what the brouhaha is about the quantum. It looks like any theory in Physics.
Thanks for the explanation anyway. Hope you are up and about soon. Should Everettian tendencies be reported as a possible side-effect of the jab? 😉

229. Andrei Says:

Scott,

“superdeterminism proposes a type of cosmic conspiracy”

It does not, at least not necessarily.

“…(limiting what experiments we choose to do to those that just happen to have certain answers)”

Every physical theory imposes constraints on what someone can do. We cannot defy gravity and fly into the sky like Jesus, or walk on water. We cannot violate momentum or energy conservation. We cannot pass through walls, etc. Why are you not complaining about those “conspiracies”? What makes you think that your ability to press a specific button at a specific time is any different? It might be the case that it is physically impossible to make a certain measurement at a certain time simply because it would imply a violation of a conservation principle.

“I.e., superdeterminism would allow for an arbitrarily magical-looking world, so that the real mystery to be explained would be why the observed world is not arbitrarily magical.”

Such superdeterministic theories would be falsified and quickly rejected. You can also imagine a quantum theory with different constants, fields, etc that would appear magical. This is not a reason to reject the concept of a quantum theory.

Your fallacy is to assume that a superdeterministic theory MUST involve absurd consequences. This is an unjustified assumption on your part. It’s easy to prove you wrong. Here is the proof:

1. If a superdeterministic theory can reproduce quantum mechanics (if it cannot it is falsified so nobody would bother with discussing its consequences) then it will describe a world exactly like the one described by QM. A superdeterministic interpretation of QM will, by definition, make the same predictions as QM. So, there is no chance you can get an “arbitrarily magical-looking world”.

230. ARaybould Says:

Tim Converse #221:

My apologies for being unclear, but I had two points in mind.

The first one was of Occam’s razor as a unifying principle between your question of, on the one hand, the existence of things beyond our light cone, and on the other, the reality of Everettian branches.  My second point was the third alternative you considered: i.e. that a split-the-differences view (or what Scott once called elsewhere “Many Worlds minus the Many Worlds”) seems to bring on more complications than fully embracing many-worlds. If the other branches are real in some senses but not in others, now we have postulated at least two different forms of reality (which is hardly a small matter!), and perhaps have implied that when a branch occurs, some sort of choice is made as to which reality each branch exemplifies.

To take an issue I raised in an earlier post (#172), suppose one thinks that other branches than the one we find ourselves on are physical in some sense (and if they are not, is this even a many-worlds view?) but not supportive of consciousness, and let’s also suppose you are an observer who has just completed a Schrödinger‘s cat experiment. Presumably, then, the self on either branch considers the one on the other branch as lacking consciousness, while they have retained theirs. After the world has gone through a few more branches, the chance that anyone is conscious seems to become vanishingly small!

I agree with your point that Everett’s view is parsimonious in postulates, and indeed, I have seen the claim that, because it does away with the ad-hoc assumption of wave-function collapse, it is preferable, on the basis of Occam’s razor, over the Copenhagen interpretation. As what we see as metaphysical possibilities are subjective, however (constrained by what we know), I don’t think that should be taken as the last word.

231. Richard Cleve Says:

Scott #215:

You don’t say it explicitly, but isn’t your argument against superdetermism just an invocation of Occam’s Razor? That the framework of QM where the Bell inequality is violated (and all the weirdness of not having local hidden variables) is incredibly more simple than a competing explanation in the framework of superdeterminism?

232. Richard Cleve Says:

P.S. Oh, I see now that ARaybould, in the last paragraph of comment #230, also characterizes your argument with Occam’s Razor.

233. Scott Says:

asdf #225: As I said several times, if an objective collapse mechanism were ever found, that would completely change every aspect of this discussion! But while people should continue to test QM at larger and larger scales and higher and higher precision, I don’t expect such a collapse mechanism ever to be found, and neither do almost any of the physicists.

So then that leaves a debate among numerous camps (Everettians, Copenhagenists, QBists, Bohmians, shut-up-and-calculators, Zen Masters…) none of whom believe there’s a dynamical physical process of “collapse.”

To me, none of this really feels like physics degenerating into postmodernism. Part of it (e.g., decoherence theory, the existence or nonexistence of an objective collapse mechanism) feels like just straightforward physics, and part of it (e.g., the reality or unreality of the other branches) feels like physics bumping up against the mind/body problem and other metaphysical enormities that have been with us at least since the ancient Greeks.

234. Scott Says:

O. S. Dawg #226: The hardness of SVP (let’s say, under the l2 norm) depends a lot on the approximation ratio. If you just want a ~√n-approximation, then Aharonov and Regev showed that the problem is in SZK and in NP∩coNP, and thus almost certainly not NP-hard. On the other hand, if you want a constant-factor approximation, then Micciancio showed that the problem is NP-hard under randomized reductions. This means that, if there were a polynomial-time algorithm for it, then NP⊆BPP, which would be basically as world-changing as P=NP.

235. Scott Says:

Steven Evans #228 and Andrei #229: So then let me be more explicit. The original papers by Gerard ‘t Hooft on “superdeterminism” were shockingly blase about the absurd implications I mentioned — implications that would mean you could explain basically anything (telepathy, superluminal signaling, etc.) via similar devices, and that physics would be over — and (to their credit) were also clear enough that there was no possible other way to interpret them. None of the other papers I saw about “superdeterminism” showed any inkling of appreciating the enormity of the problem. And none of them contained what I saw as the slightest hint of a promising idea to balance the absurdity.

By the usual standards I apply to anything else, this would be more than enough reason for me to ignore the topic thereafter.

The only difference with “superdeterminism” is that people keep bringing it up! They say: but you can’t prove that someone in the future won’t find a superdeterministic theory that perfectly explains QM, while prohibiting all magical super-quantum effects, without our needing to stick in those desired results from the outset—even if you’ve seen no hint of such a theory in the past 20 years. You can’t prove that there’s no diamond buried underneath this mountain of shit—even if you searched and didn’t find one, maybe someone will stick a diamond there later.

All this is obviously true, and obviously beside the point. So why don’t more people see this?

Part of it might be the name. People who reason verbally think if you deny “superdeterminism,” it must be because you believe in a mystical free will. In reality, of course, the opposite of “superdeterminism” isn’t free will; the opposite of “superdeterminism” is physics where you actually have to explain and calculate things rather than saying “well, maybe some unknown conspiracy in the initial conditions of the universe messes with our measuring devices and brains in order to always cause it to look this way.”

Or maybe it’s just that some of the leading advocates of superdeterminism (Gerard ‘t Hooft, Sabine Hossenfelder) are clearly smart. Surely there must be more to it, then, surely they must see something inaccessible to everyone else’s scientifically informed common sense?

I can only call things like I see them. My experience has been that arbitrarily smart people can become arbitrarily attached to arbitrarily bad ideas and arbitrarily unwilling to admit when they’ve hit dead ends. FWIW, though, there are vastly more smart people in the anti-superdeterminist camp! It’s just that most of that camp ignores the subject, spends its time on other things, and doesn’t have blogs where people keep bringing superdeterminism up. 😀

236. Vampyricon Says:

Scott #84

Hopefully the following will clarify my position. My feeling is that philosophical analysis has probably already done nearly all it can for the interpretation of QM. I’m not optimistic that any new interpretation, or any new argument for an existing interpretation, is going to come along and make everyone slap their foreheads and exclaim “it’s all so clear now! why didn’t any of us think of that earlier??”

I think this argument generalizes too much. One could also be pessimistic about any new information coming out about the Great Pyramids’ builder that would make everyone slap their foreheads and exclaim the same thing. It could stay as “probably Khafre, but maybe Khufu” for the rest of eternity (example shamelessly pulled from an SSC post; might not reflect the reality of the situation), but there’s still a fact of the matter about who ordered them built. Historical analysis has done all that it could, but that doesn’t mean that a zen anti-answer would be the appropriate response.

237. Scott Says:

Richard Cleve #231: In a case like superdeterminism, even mentioning Occam’s Razor feels like overkill! Occam’s Barely-Perceptible Gust of Wind would suffice to topple such a ridiculous edifice. 🙂

238. Scott Says:

Vampyricon #236: The difference is that, with the pyramids, even if we judge new empirical data unlikely, we understand exactly what such data could look like—say, some papyrus unearthed in the Nile Delta saying “I, Khufu, order this pyramid to be built…” With interpretations of QM, by contrast, can you give a single hypothetical example of what the new data bearing on the question would be? Crucially, anything that changes the rules of QM (like an objective collapse mechanism) is disallowed, since then we’re no longer talking about interpretations but about rival physical theories.

239. Vampyricon Says:

Scott #190,

Apologies for the double-post (merge them into one if you wish), but it seems the “chain of things I can talk to” criterion of reality seems to deny the reality of people in the past, e.g. the ancient Greeks you mentioned in the comment. It also seems to imply that reality is not reciprocal. our far-future descendents are real for the ancient Greeks but the ancient Greeks are not real for our far-future descendants.

And finally, a question about decoherence: I am under the impression that one still needs a projection postulate under decoherence. Doesn’t that invalidate the claim that it solves the emergence of classicality?

Wyrd Smythe #219, I think the answer would be that spacetime is (expected to be) something with a quantum state, so asking where the Schroedinger equation exists (or the appropriate quantum laws of evolution) would be like asking whether an atom is solid or liquid. Is that Tegmarkian? I don’t see how it is necessary. The Tegmarkian Platonic multiverse would be the claim that all consistent mathematics exist (in some manner?). But the Everettian can say that the Schroedinger equation and the quantum state of the universe are all that exists.

240. Scott Says:

Vampyricon #239: I can obviously talk to people who talked to people who (etc.) talked to the ancient Greeks. Just like I can obviously talk to people who will talk to people who will (etc.) talk to my remote descendants. The ancient Greeks can even directly talk to me, and I can even directly talk to my remote descendants, although in those cases (alas) the communication is only one-way. But clearly we’re linked to the entire rest of the human race by numerous chains of causality.

As for decoherence, it clearly solves the problem of getting you to a superposition of quasi-classical worlds, in every one of which it looks like your quantum state was measured and collapsed according to the standard rules of QM, and where the amplitude on each world is the square root of the probability that you should expect to find yourself in that world. That’s no mean feat!

Decoherence does, however, leave the obvious questions of what “breathes fire” into one of the worlds—chosen with probability equal to the squared absolute value of its amplitude—and whether anything breathes fire into the other worlds as well. This is the part that I posited is basically the quantum-aware version of the mind-body problem, mysterious in much the same way the original was.

241. ARaybould Says:

Richard Cleve #232
Just to be clear, I had not been thinking about superdeterminism – I had not even heard of it until yesterday!

242. Richard Cleve Says:

ARoubould #241: Yes, sorry I misread your comment (you were comparing with Copanhagen not superdeterminism).

243. Steven Evans Says:

Scott Says:
Comment #235 March 9th, 2021 at 9:25 am

Fair enough. I accept that you have concrete, technical reasons for your position on superdeterminism. I won’t mention it again unless I have equally concrete reasons, which you predict won’t happen.

244. Vampyricon Says:

Scott #238, #240,

Thanks. I think I understand your position better now, at least. As for the Greeks, that was rather embarrassing, as I didn’t even think of the case where Y could have talked to Z before X talked to Y.

I do see the similarities between the “other worlds” problem and the mind-body problem, though I can’t see how they’re the same problem. Would you submit to a zen anti-solution to the mind-body problem as well? 😉

245. Scott Says:

Vampyricon #244: After 20+ years in quantum information science, I’ve “merely” reached the level of Zen where I can see how the measurement problem decomposes into a part that’s straightforwardly handled by decoherence theory and a part that’s of comparable enormity to the mind-body problem, and where I see how Everett, Copenhagen, and Bohm are merely different crutches that could be better or worse at leading people to that realization. If there’s a level of Zen where you actually see the solution to the mind-body problem or its quantum analogue, then I certainly haven’t reached it yet! 😀

246. fred Says:

1Zer0 #222

Technically one could raise a baby in an environment that’s made of pure white/gray/black (granted, it’s cruel).
Would that still stimulate all the cells in the retina? But the brain would maybe grow a visual cortex that’s not ready to interpret colors.
Once the kid is an adult, what would happen if one day you started introducing some red?.. and after some sufficient time, you introduce green and blue. Would they be surprised when introducing more intermediate colors (yellow, orange, …) or red/green/blue is enough (matching the three types of cones)?

A more ethical experiment would be to yourself wear permanently orange glasses (filtering blue)… after a few years, what would it be like to suddenly take off the glasses? Would the blue be instantly perceivable or would it take a while for the brain to learn to re-interpret it uniquely?
People have done this type of experiment by wearing glasses that have mirrors that reverse the image coming to the eyes, apparently after a while for them it just feels like the image is no longer upside down… interesting stuff!

When it comes to “loudness” or “brightness”, our brain is always looking for changes (either from immediately prior stimulus, or from memories for more complex differences), uninterested in steady signals.
And sudden wide shifts in brightness, loudness, or temperature all produce the same internal alert, and we really react to the scale of that surprise.
You can see that in babies, where a new experience is creating a strong surprise, and they’re not quite sure whether it’s a good thing or a bad thing, and they either laugh or cry for no reason.
Or in adults, where there’s a range that goes from a shift in attention (for small changes) to having a jump scare (created by a loud sound, a sudden bright light, or a hand on our shoulder when we didn’t expect it). So that’s kind of a generalized “loudness”.
Fundamentally all content of consciousness can probably be broken down in more basic bricks of sensation on a one dimension scale that has dual extremes, like good vs bad, from the time we were simple organism with very basic sense of dividing stimuli into an action to either move toward it or move away from it.

247. fred Says:

Scott #245

“If there’s a level of Zen where you actually see the solution to the mind-body problem or its quantum analogue, then I certainly haven’t reached it yet!”

There’s actually no such thing as the “mind-body problem”, i.e. it’s not a problem, it’s a solution.

https://www.amazon.com/Healing-Back-Pain-Mind-Body-Connection/dp/0446557684

Joking aside, this book is about a Zen type realization too. How years of back pain can be relieved by just reading the words that are in the book. It worked for me!

248. fred Says:

Scott #240

“Decoherence does, however, leave the obvious questions of what “breathes fire” into one of the worlds—chosen with probability equal to the squared absolute value of its amplitude—and whether anything breathes fire into the other worlds as well.”

What you’re asking is the reason for an apparent break of symmetry.
Why would one branch matter more than the others?

Schrodinger wondered about a similar but even more basic/direct question, to which we can all relate:

His thought experiment:
Two people, A and B (their brains can be as close atomically as we want), a dark room, and a garden with a tree:

In situation 1), A is in the dark room, while B is outside looking at the tree.
In situation 2), B is in the dark room, while A is outside looking at the tree.

From the perspective of the universe “wave function”, situation 1) and 2) are as equivalent, and interchangeable as we want.
Yet, we know that there exist actual things we can call “point of views” (what we call A and B’s subjective experience) where 1) and 2) are totally non-interchangeable.
That’s a clear apparent massive break in symmetry.
If A and B are the only conscious entities out there, what makes it that you are A and not B?
Similar to your question “why am *I* in this branch and not the other?”.

249. fred Says:

Scott #235

“Part of it might be the name. People who reason verbally think if you deny “superdeterminism,” it must be because you believe in a mystical free will. “

Part of it may also be this (I’m paraphrasing):
when Bell was asked “but if there’s no such thing as a truly independent choice?”, he said “ah, then, indeed, what I showed doesn’t hold”.
He didn’t say “bah! even if there isn’t independent choice, you’d have to resort to crazy conspiracy theories to explain the results!”… if it’s so obvious to you, Scott, it’s strange that Bell dismissed it the same way?

250. fred Says:

Here’s the quote (from the wiki on Superdeterminism):

“In the 1980s, John Bell discussed superdeterminism in a BBC interview:[3][4]

“There is a way to escape the inference of superluminal speeds and spooky action at a distance. But it involves absolute determinism in the universe, the complete absence of free will. Suppose the world is super-deterministic, with not just inanimate nature running on behind-the-scenes clockwork, but with our behavior, including our belief that we are free to choose to do one experiment rather than another, absolutely predetermined, including the “decision” by the experimenter to carry out one set of measurements rather than another, the difficulty disappears. There is no need for a faster than light signal to tell particle A what measurement has been carried out on particle B, because the universe, including particle A, already “knows” what that measurement, and its outcome, will be.”

If it’s so easy to dismiss “super-determinism”, why didn’t Bell simply bring up that this would *also* have to rely on a crazy conspiracies to work?

251. fred Says:

Another transcript of Bell’s own words, from “The Ghost in the Atom: A Discussion of the Mysteries of Quantum Physics”, page 47:

Q: I was going to ask whether it is still possible to maintain, in the light of experimental experience, the idea of a deterministic universe?

You know, one of the ways of understanding this business is to say that the world is super-deterministic. That no only is inanimate nature deterministic, but we, the experimenters who imagine we can choose to do one experiment rather than another, are also determined. If so, the difficulty which this experimental result creates disappears.

Q: Free will is an illusion – that gets us out of the crisis, does it?

That’s correct. In the analysis it assumed that free will is genuine, and as a result of that one finds that the intervention of the experimenter at one point has to have consequences at a remote point, in a way that influences restricted by the finite velocity of light would not permit. If the experimenter is not free to make this intervention, if that also is determined in advance, the difficulty disappears.

252. Scott Says:

fred #249-251: Because Bell probably never imagined that anyone would actually take it seriously—so he could just explain it as an amusing loophole in his theorem.

253. Tim Converse Says:

ARaybould #230: Now that I better understand you, I completely agree, especially your first large paragraph (that a split-the-differences approach seems as though it introduces new complexities not found in either Everettian or Copenhagen interpretations).

Scott #240: I agree 100% that we can talk to the ancient Greeks (or rather, they can talk to us) and that they’re linked to us by multiple chains of causality, which is why we all have no problem agreeing that they existed.

In that sense, they have a very different relationship to us than do (purely hypothesized) other branches of the wavefunction, which cannot send us any information even if they do exist in any useful sense of the word ‘exist’.

But how is an ancient Greek person similar to a galaxy that is outside our lightcone again? I think you had some transitive chain of communication in mind (like Greek manuscripts recopied by Irish monks and eventually making their way to our Kindles in English translation), but I don’t see it in the faraway galaxy case. I may just be missing a cosmological point, or possibly you meant some looser sense of transitive linkage that (unlike in the case of ancient Greeks) doesn’t permit information transfer in either direction(?).

254. Scott Says:

Tom Converse #253: Yes, it’s a cosmological point. If I could set off now to meet the Zorkians, and the Zorkians could set off right now to meet the Zaffleborgs, then even if I can’t set out to meet the Zaffleborgs because they’re outside my horizon, it still seems like I should ascribe reality to the Zaffleborgs, because of the conditional causal chain connecting me to them.

255. Steven Evans Says:

Scott, Having read through the thread, regarding MWI, aren’t we simply forced to admit the branches are not known to be physical? All of the states in the superposition were seen to be having a physical effect pre-“collapse”, but post-“collapse” only one of them is seen to continue having a physical effect. There’s no debate, is there?
And why would one consider supposed extravagance of MWI as an issue? MWI neatly contains the essential idea of all states continuing to have a physical effect post-“collapse”, which can’t be checked physically anyway, so there isn’t much point pondering how much epistemic fat MWI contains, is there?

256. Scott Says:

Steven Evans #255: Well, that’s the whole question, isn’t it? As we’ve been discussing at length in this thread, there are cases of something that can no longer have a “physical effect” on us—for example, a galaxy that’s receded past our cosmological horizon—that most people would nevertheless regard as physical, because it was once part of our observable universe. There are other cases—for example, a world where the South won the Civil War, or whatever—that if not for QM, only a thoroughgoing modal realist would be tempted to regard as physical. So the question is, are other Everett branches more like the former or like the latter?

257. Bruce Smith Says:

Scott #254:

Your example has connections (chain links) of two types: “A could travel and meet B” and “B at time 2 is connected to B at time 1”. Note that you connected them in that order, even though the first type of link moves into the future and the second type moves into the past.

Another commenter has given a chain using only the second type of connection, forming a chain connecting one person B in two Everett branches which diverged at time 2: “B at time 3 in branch X is connected to B at time 1”, and “B at time 3 in branch Y is connected to B at time 1”. In this case the overall chain’s links moved into the past, then future, respectively. I think this is no worse than your galaxy example’s 3 links, which moved into future, past, future, respectively.

These pictures seem equivalent to me, as well as to some other commenters. I don’t yet understand how this kind of chain convinces you to treat galaxies-at-times, which are too far away for you-now to interact with, as real, but “yourself in another Everett branch” as less clearly real. To me the multi-link chains involved are entirely analogous. (Both of them involve “links to past” joined to “links to future” in one path, making the overall path non-causal even though each link is causal.)

I guess I can see one difference — in your example, each local short-enough part of the overall path doesn’t involve anything weird, whereas in the Everett example, there is a “moment of splitting of Everett branches” such that any portion of the overall path (which connects B3X to B3Y), however short, which includes it, has something weird in it. Is *that* the real difference?

258. MiKeP Says:

Hi Scott, not gonna risk moderation. Just, I’ve read you on why complex numbers in QM. So, in case you’re interested, Charlie Wood writes in Quanta of a preprint still in peer review (but is on arXiv) that a 3-way Bell test can distinguish R- from C-based QM. An A-B-C extended EPR pair test they propose looks possible, tho not done yet. If true, AIUI, the explanation of entanglement correlations via past interaction (ie hidden variable) would be toasted in a dramatic new way: A-C results correlate, but there are no A-C pairs.

259. Scott Says:

MiKeP #258: Did it occur to you to check whether I’d already blogged about that exact paper, when it came out? I thought it was really nice! But it’s not about “hidden variables” at all (in the usual sense of classical objects)—rather, it’s about ruling out the possibility of a “hidden real QM” that locally simulates the complex QM that we see.

260. MiKeP Says:

Scott #259: Now that you point it out, I do remember reading your blurb. You didn’t mention my wrinkle, tho. Which the paper doesn’t “talk about” but in my admittedly incomplete quick browse, seemed to be an implication. The part whereby Alice’s & Charlie’s data don’t have a common past history but their connection thru a 3-way entanglement creates statistics suggesting shared info. This info hides in the phases in the complex model but is no less seemingly acausal than the physical info sharing required by the reals-only model. Thus, my question is more relevant in this context, of this post, than it had seemed when you alluded to the same paper a month back. As you’ve read it, I suppose it’s not too much asking you to swat me down properly, please.

261. gentzen Says:

fred #250: “In the 1980s, John Bell discussed superdeterminism in a BBC interview: There is a way to escape the inference of superluminal speeds and spooky action at a distance. But it involves absolute determinism in the universe, the complete absence of free will.”

This “absolute determinism” part seems to be related to why making “Many Worlds minus the Many Worlds” work … risks to end up with a superdeterminist theory/interpretation: Without the “many worlds” everything is completely determinstic. So both free will and randomness have a hard time to emerge in the theory. They somehow have to get back in via the initial conditions.

One reason why I brought up “Many Worlds minus the Many Worlds” (in addition to the reason that Scott Aaronson coined that term) is that Sidney Coleman denied the need for many worlds, at least according to Peter Woit:

While some might take this and claim Coleman as an Everettian, note that there’s zero mention anywhere of many-worlds. Likely he found that an empty idea that explains nothing, so not worth mentioning.

However, I should not deny that I mentioned superdeterminism here, and that I reviewed superdeterministic proposals favorably in the past, for example Jarek Duda’s “maximum entropy random walk” proposal in August 2017.

Even with respect to Sabine Hossenfelder’s proposed experimental tests of superdeterminism, I am at least undecided. My email to her and Tim Palmer also contained the following paragraph:
“The superdeterminism of Neumaier’s argument might first seem different from the superdeterminism defended in your paper. However it is in a sence similar to what you propose in section “5.3 Future-bounded Path Integrals,” just simpler. And I had similar thought as those expressed in section “6 Experimental Test” when I first read Neumaier conjectured explanation for how randomness emerges. I thought about the difference between the randomness of a six-sided dice, and the randomness of shuffling cards, and wondered about whether the shuffled cards would always be sufficiently random.”

I also think that keeping the term “superdeterminism” with all its negative connotations (and opportunities for straw man arguments) was a good thing. Otherwise it would have been too easy to just ignore the enormous consequences of accepting superdeterministic explanations as “physical explanation”, instead of merely as “valid” consistent mathematical models (or approximations).

262. Steven Evans Says:

Scott Says:
Comment #256 March 9th, 2021 at 10:51 pm

“So the question is, are other Everett branches more like the former (galaxies disappeared over the horizon) or like the latter (alternate histories)?”

We simply don’t know whether or not the other Everett branches are physical. The empirical evidence tells us the other Everett branches at least represent blueprints for alternate physical realities that we could have experienced, but we don’t know if they represent continuing physicality as our only test of physicality is observation. All analogies and metaphysical considerations flounder at the shores of this simple fact, don’t they?

263. fred Says:

Tim Converse #253

https://i.imgur.com/pl7LjNF.png

264. fred Says:

“possibly even a … photon or a hydrogen atom, so you can see quantum probability in action and be enlightened.”

If I’m not mistaken (*), one of the simplest ways to see quantum probability in action is to look at light shining on a piece of glass: a proportion x is transmitted through the glass, and a proportion 1-x is reflected back. And that proportion x is a quantum probability.

(*) Feynman had some nice talks on this

265. Scott Says:

MiKeP #260: Ah—if that’s your question, the idea of “entanglement swapping” is much older than this particular paper (it’s even been used in experiments). I hadn’t thought about the implications (if any) of entanglement swapping for local hidden-variable theories, which are of course already ruled out by the Bell inequality with simple bipartite entanglement. If you didn’t know about the Bell inequality, and only knew about entanglement swapping … the trouble is, couldn’t you just imagine the hidden variables leaping from one particle to the next, in order to explain the correlations you see? That seems no crazier than what the local hidden-variable people are committed to anyway! 🙂

266. 1Zer0 Says:

Fred #246,

That experiment is an interesting approach. I believe the mind would immediately recognize an old but not experienced in a long time or completely new sensation (Like for the people having new Qualia – once the new Qualia is there, it’s recognized right away.) Aren’t there these Enchroma glasses for colorblind people to make them experience colors for the first time and they actually see it right away? I thought I read something like this a few years back.

Maybe any mind state or sensation is a linear combination of basic qualia categories. Just like a hamel basis spawns a vector space or an axiomatic system spawns a theory, the basic qualia categories spawns our internal experience. Like define for example Visuals, Sound, Feelings, Smell and Flavor to be the foundational categories. Then every possible experience including the experience of “self” is a linear combination of those basis qualia. x = a* Visuals + b* Sound + c* Feelings + d* Smell + e* Flavor, where a,b,c,d,e are the coefficients determining the strength of the experience. I know it’s horrible because I just multiply a qualia objects with a real number, thus completely incompatible with the theory of fields but I don’t have a better way to express what I mean with “mind state”. Thoughts should be encoded as visual or sound information (the inner voice when thinking).

267. ARaybould Says:

Scott #256: The possibility of a Southern victory in the Civil War was at least a logical possibility (i.e. one not inherently contradictory or definitely ruled out by nonredactable facts) up to some point (when that point was can be argued over indefinitely, but surely some uncertainty in timing is moot here?) If the branches of a many-worlds view are taken to be merely logical possibilities, then is there any difference between this view and the Copenhagen interpretation? If not, then for many-worlds to be an alternative to it, do the branches not need to be more than just logical possibilities? – but in what way?

I take your point that if all branches are ‘1st-class real’ then there are many worlds in which the South did win the Civil War, and the question of what is really real would be a non-issue.

268. Kenneth W. Regan Says:

Dick Lipton and I take the same attitude in (the second preface to) the second edition of our QC textbook, calling it “simple realism.” An example I’ve used in talks is to ask, what object does a classical probability distribution on n items define? Answer: the simplex in n-space. Well, if you look at the simplex for higher n it actually gets surprisingly spiky, and anyway, its properties were not really understood until the 1950s. Whereas the object defined by quantum probability is the sphere, and if you asked Plato to imagine its perfection in any dimension, he probably could have.

269. Max Madera Says:

Steven Evans #262: That is how I see it. As long as QM is unitary and there is no “mechanism” for creating worlds splitting matter (à la amoeba universe that some early MWIists envisioned), branches are only there to wonder what would it have looked like if I had been in the parallel universe and other meditations. As soon as they split and become a different branch they say bye-bye to my physicality.

270. fred Says:

To me, the fact that we don’t see any other intelligent life in the universe is a clue that all branches are physical, and we just happen to be in a very unlikely one – if only one branch were to be realized, the most likely one, we wouldn’t exist.

Remember this as you will find yourself seemingly outliving all your friends and relatives, surviving countless accidents and deadly diseases, in a smaller and smaller proportion of all the branches (just like the current “you” has outlived the other “you”s that had the bad luck of dying at a younger age in many of the other branches).

271. Scott Says:

Bruce Smith #257: On reading your lovely and insightful comment, my first thought was—

“Shoot, I’ve been hoist by my own petard! I proposed a criterion for the ‘reality’ of unobservable entities based on transitivity—can I bump something that can bump something that can … etc. bump the thing in question. I used this criterion to explain why I regard galaxies outside my cosmological horizon as ‘real,’ but feel no similar certainty about the other Everett branches. And yet I somehow didn’t even notice that by that very same criterion, I have memories of my past self, and my past self can send messages to my doppelganger over in the other Everett branch, and therefore I ought to regard the doppelganger as real too! Time to pack up and go home. You’ve convinced me: by my own criterion, Everett was right, full stop, no need for this wishy-washy Zen stuff.”

But then I thought about it some more, and I realized that this seems to prove too much. Like, even if the world hadn’t been quantum, wouldn’t precisely the same argument have forced us into David Lewis’ (or Max Tegmark’s) modal realism, where all the possible worlds exist? After all, I have memories of my childhood, my doppelganger in some (non-quantum) hypothetical universe where I became a Silicon Valley startup person rather than a complexity theorist has memories of the same childhood, ergo there’s a bidirectional causal chain connecting me to the doppelganger, and the doppelganger must be real!

In both the Everett case and the modal realism case, I’d say that the suspicious step is the one forward in time—from your past self to the contemporary doppelganger in the alternate version of reality. Who’s to say that the doppelganger even exists? Isn’t that the very question at issue? In the case of galaxies outside our cosmological horizon, the Copernican principle itself seems to force us to admit that someone just at the edge of our cosmological horizon—someone who we could talk to—could see those galaxies if they wanted to, and therefore that the galaxies must “exist.” But with MWI and with modal realism, there seems to be no analogue of the Copernican principle to invoke: for us and for our past selves alike, for all we know maybe only a single future path will have the fire of conscious experience breathed into it, and we have merely epistemic uncertainty about which path it will be.

272. J Says:

Kenneth W Regan #268,
Would you mind explaining that answer in layman’s terms?

273. Tim Converse Says:

Scott #257:

(At this point even I might be tempted to conclude that we’ve beaten this topic to death. But not quite.)

Right, if you mentally go backward in time and then forward on a different branch you have proven nothing about the existence of that other branch. You must first demonstrate or assume that the other branch exists – which of course is the very question at issue.

(That was actually kind of the point when i introduced it in comment #201. Not to demonstrate that many worlds exist – to the contrary, to highlight that a chain of hypothetical causal links (varying in direction) running from A to B .. to Z might not always work as a way to infer the existence of Z.)

Similarly, if you could set off now to visit the Zorkians, and at the same time the Zorkians could set off now to visit the Zafflebergers, then you have a nice set of conditional causal links …. if the Zafflebergers exist – which of course was the very question at issue.

Now you’re invoking a Copernican principle to say that if we can see Zork from here, then from Zork the Zorkians can also see Zaffleberg (which must therefore exist). And now you have established three actually-existing locations which are connected by transitive conditional causal links, from which you can derive faith that … Zaffleberg exists, which we just assumed in the previous sentence.

I am fine with assuming that Zaffleberg exists because the Copernican principle. In fact, that is the reason why I think that such galaxies do exist.

This is equivalent to saying that my preferred cosmological theory (which is among other things Copernican) is preferred even though it forces me to stipulate the existence of entities which I unfortunately cannot observe. It has too many nice features to throw it over just because it brings some unobservability baggage along with it.

Similarly, hard-core Everettians have a preferred theory (aka the Schrodinger equation) which they prefer due to its simplicity and determinism as compared to the Copenhagen interpretation. Unfortunately this preference brings along with it an incredibly vast array of unobservable entities. So be it.

No one is saying that if you believe in Copernicanism and distant galaxies then that means that you must believe in Everettianism and many-worlds. Feel free to choose yes from Column A and no from Column B. All we are pointing out is a structural similarity in these two cases with regard to preferred theories and consequent non-observables.

274. Steven Evans Says:

Max Madera Comment #269

“there is no “mechanism” for creating worlds splitting matter”

The Everett branches may be physical. We just don’t know. It’s not a scientific question currently as there is no known observation that could be carried out to resolve it. The appeal of MWI is that w/o such an idea we have to say that all the components of a superposition that weren’t “collapsed to” during the “measurement”, components that were having honest physical effects pre-“collapse”, are suddenly apparently having no physical effect and by a completely arbitrary “choice” as all outcomes were equally possible. And there’s no mechanism for this apparent ceasing of physical effects either, if you want to get into mechanisms. It’s quantum no-mechanics.

275. MiKeP Says:

Tim Converse #273: So I can choose Copernhagen?

276. fred Says:

“Who’s to say that the doppelganger even exists?”

But if you were an electron going through two slits in a way that noone else can tell which way you went, it seems that all your doppelgangers would have to be real?

277. MiKeP Says:

Scott #265: In a local hidden variables theory, all interactions even if they involve hidden variables still need to be local. Swapping or changing local hidden info at B cannot reach out nonlocally & spookily change the course of local hidden info at C. In that sense, once an EPR pair is “far enough apart”, as EPR would have said, or as we say, spacelike separated, their hidden variables are “baked in”. BTW, technically it should be permissible for the hidden info to evolve independently or interact locally. That complicates matters so we normally elide these possibilities.

FWIW, EPR (the paper) had specifically in mind conserved classical dynamical variables where we now read between the lines & substitute “entanglement”. I like to simply think of angular momentum as the hidden variable. Angular momentum transfer can be straightforward or slippery. I guess transfer via tidal drag might be the slippery kind. In any case, a binary interaction at B with an unentangled object could create a 3-way entanglement with a partner at C, but from the partner at C’s PoV, the partner at C’s own angular momentum partner at B is a more complicated thing now. Locally, the “partner at C” itself remains utterly unaffected. The goings-on at B & C are “out of the picture” to one another. So this is one way, my way, of thinking of a classical (local) equivalent to entanglement via “hidden variables”.

In short, there’s still no local way to transfer angular momentum between B & C that are spacelike separated. Anyway, it seems to me that any such funny business with entanglement magic, regardless what name we give it, wouldn’t be includable in any system satisfying the premises of Bell/CHSH Thms (in my circle, what we call a “Bell model”). QM denies the premises & permits the impossible by not giving a mechanism. Where the classical mechanic considers the Bell model as forced by physics.

Overall, what I think this reals QM v complex QM example teaches us is that there really is a spookiness going on in entanglement. It needs someplace to hide that we can consider in some way to be nonphysical. The reals-only case deprives us of this someplace, so we have to add it back. Since implicitly we aren’t allowing the theory itself to be deformed, under the rules we’re playing by, the only way we can add things in explicitly must take the form of physical content. In the reals theory, our choice is thus between “manifestly spooky” & “no way to do that”. Have I got that right? Or did I mess things up?

278. Andrei Says:

Scott,

“The original papers by Gerard ‘t Hooft on “superdeterminism” were shockingly blase about the absurd implications I mentioned — implications that would mean you could explain basically anything (telepathy, superluminal signaling, etc.) via similar devices, and that physics would be over — and (to their credit) were also clear enough that there was no possible other way to interpret them.”

Can you please point me to such a paper? I’ve never seen ‘t Hooft admitting this.

His last paper:

“Explicit construction of Local Hidden Variables for any quantum theory up to any desired accuracy”

https://arxiv.org/pdf/2103.04335.pdf

is nothing along those lines.

I do not have the required expertise to evaluate this paper, but, assuming for the sake of the argument that ‘t Hooft’s claims are true we can deduce that nothing like telepathy or superluminal signaling would be possible. This is because the model is constructed to reproduce the predictions of QM.

The intent of superdeterminism is not to replace QM with a theory that makes different predictions, but to complete QM with local variables along the lines suggested by EPR. Such a theory would, by design, have the same predictions as QM, so everything that QM forbids would also be impossible in superdeterminism. QM does not allow for superluminal signaling so neither will its hypothetical superdeterministic completion. And if someone builds a superdeterminstic theory that allows for superluminal signaling it would be dead at birth, already falsified, so there is no reason to expect physics to be over.

279. Max Madera Says:

Steven Evans #274: I don’t think I am in disagreement with you.

As I see it, in standard QM we amplify to “1” the amplitude of the branch that we experience (once we experience it, obviously) and to “0” all other branches: the “+” becomes a logical “or” for macroscopic objects or worlds. There is no mechanism (no physical process) for such collapse.

In MWI, if I am not mistaken (too bold a claim, probably), after the splitting we amplify all branches to “1” although we only experience one: the “+” becomes a sort of logical “and”. (In fact, in the expositions of MWI that I have seen, they always start with a single-branch universe and never put any amplitudes multiplying the branches.) I understand that within each branch there is life (or other possible lives or histories) as we experience it. Within a branch we do not have to care about the amplitude.

However, the process by which, at the macro-world (after no interference is possible) we elevate + to “or” or “and” (or we make the amplitudes 0 or 1) is FAPP a non-physical question, as nothing in the Schrödinger equation allows you to change the amplitudes taking into account your experience or the set of all possible experiences.

280. Man to research Says:

I am 43 and is it too old to start a phd in cs?

281. Scott Says:

Man to research #280: Not necessarily! But please, no more off-topic questions.

282. Max Madera Says:

When I mean that the latter process of changing the amplitudes is not physical, I mean that it is an additional postulate not encoded in the Schrödinger equation. In standard QM (or in Qbism) it is the observer who does that. In MW it is god or the Universe. But updating the amplitudes or probability distributions by giving a branch/world the certificate of existence or parallel existence is something that is made by hand. It is not in Schrödinger’s.

I guess that one can always say that in MWI the amplitudes are always there: you do not need to change them and hence you can discount one postulate. But I always felt that this is a little bit like cheating. Once the worlds split apart the amplitudes are useless. Just like in standar QM, in MWI the amplitudes are a device to calculate probabilities with complex numbers (following all possible interferences). At the time I claim that all branches are equally real, the amplitudes mean nothing. And this is why, as far as I have seen, in MWI everybody starts with an initial state that is exactly the same as in Copenhagen: the single branch that you experience.

283. Max Madera Says:

Steven Evans #274 (cont.):

When I mean that the latter process of changing the amplitudes is not physical, I mean that it is an additional postulate not encoded in the Schrödinger equation. In standard QM (or in Qbism) it is the observer who does that. In MW it is god or the Universe. But updating the amplitudes or probability distributions by giving a branch/world the certificate of existence or parallel existence is something that is made by hand. It is not in Schrödinger’s.

I guess that one can always say that in MWI the amplitudes are always there: you do not need to change them and hence you can discount one postulate. But I always felt that this is a little bit like cheating. Once the worlds split apart the amplitudes are useless. Just like in standar QM, in MWI the amplitudes are a device to calculate probabilities with complex numbers (following all possible interferences). At the time I claim that all branches are equally real, the amplitudes mean nothing. And this is why, as far as I have seen, in MWI everybody starts with an initial state that is exactly the same as in Copenhagen: the single branch that you experience.

284. asdf Says:

Meh, I’m still holding out for objective collapse. It would mean there is no multiverse branch where Trump is still president. Just think, you could get the Nobel physics prize and the Peace prize for the exact same paper!

285. Steven Evans Says:

Max Madera Comment #279

We could be closing in on our host’s threshold for bike-shedding, but I think this represents a response to your general thrust, and I agree we’re not nec. disagreeing in the main.

Based on current empirical evidence, the physical effect of the components of the wave function is there in the observed interference pre-“collapse”, but post-“collapse” that physical effect appears to have disappeared for all but one component. Where did that physical effect go? That’s the mystery.
You seem to be stating amplitudes are meaningless in MWI. But, presumably, it is intended that in the child branches after a separation event (a “collapse” from our pov), the previously superposed components are now not superposed so correctly have amplitude 1. Other objects may well still be in superpositions in the various branches and that will be reflected in physical observations.

The amplitudes are always giving the right answer for physical observations in the 1 branch we know of anyway. The mystery is the change event called “collapse” and in particular the apparent disappearance of physical effect (the ontic to epistemic ratio of the wave function is irrelevant to this point).

286. Raoul Ohio Says:

Scott,

Any thoughts on the recent chatter about “causaloid framework”, “quantum switch”, etc.? Here is a brief summary: https://www.quantamagazine.org/quantum-mischief-rewrites-the-laws-of-cause-and-effect-20210311

287. Geza Says:

Andrei #278: the criticism is not that a given superdeterministic model *implies* telepathy or superluminal signalling but that the superdeterminism explanation would equally well allow you to explain superluminal signalling even in a Minkowski world or the fact that we never observe telepathy even in a world governed by laws that allow for it…

To say (like the superdeterminism loophole to the Bell inequality does) “whenever somebody performs a Bell measurements the initial conditions ensure that the results violate the Bell inequality even though the conditions of Bell’s theorem (local hidden variables) are satisfied” does not seem different from saying “no, of course the laws of nature don’t allow telepathy, but the initial conditions of the deterministic universe are such that whenever you make these conjuring motions, burn a piece of cloth, and look in a glass ball you will perceive what the person last wearing the cloth is thinking right now”. (or, reversely: “sure, the laws of nature allow superluminal signalling, but the initial conditions are such that whenever you set out to do so, attempt will fail”).

I view superdeterminism as the postulate of a cosmic conspiracy against experimental science, since it says that the initial conditions are chosen such that *all* measurement outcomes ever possible are not showing what a “free” observer (undetermined by the initial conditions) would see. It’s logically allowed, but will only ever convince those who like the predetermined outcomes. Once one allows this for Bell experiments, what forbids its use to explain any other discrepancy between my favourite theory and experimental results?

288. Orno Says:

There are anthropic edge cases where the predictions made by different interpretations differ (I believe). Subscribing to this zen interpretation, what would you expect to happen if you were put into a box with a certain famous cat?

289. Scott Says:

Orno #288: I’m not sure which “anthropic edge cases” you’re talking about. If I’m actually inside the box with the cat, then I expect to die with probability 1/2. I’m not such an utter idiot as to believe that I can jump off cliffs, get into boxes with probabilistic poisons, etc., and then magically condition on being in an Everett branch where I happen to stay alive.

290. fred Says:

An new interesting article about QM and causality.

https://www.quantamagazine.org/quantum-mischief-rewrites-the-laws-of-cause-and-effect-20210311/

291. fred Says:

Scott #289

” I’m not such an utter idiot as to believe that I can jump off cliffs, get into boxes with probabilistic poisons, etc., and then magically condition on being in an Everett branch where I happen to stay alive.”

But we’re constantly put into such boxes whether we like it or not.

In a way, you *will* find out whether there are conscious Scott A. in all the branches, because we’re all constantly aging and creeping towards death, and probability of death reaches 1.0 eventually (no-one’s ever lived to 200 afaik).

You’re alive now, but don’t you think that there are branches (“real” or not) where Scott A. was born, but has already died?
Can’t we estimate this probability by looking at probability of death by age in the general population? Fom your point of view, the sample of people alive isn’t subjective, while, from your point of view, your probability of being currently alive is always 100% (it’s like a personal anthropic principle).

For example, 30% of the population doesn’t make it to 65. So, if you’re 65 at the moment, you could roughly say that you’re no longer alive in 30% of the various branches (children of the branch where you were born). So, finding yourself alive at 65, you can say you’re lucky the one “real” branch you find yourself in is part of the 2/3 branches where you’re theoretically alive.

If you ever find yourself at 110 having survived multiple cancers, you’ll know that either you’ve been very very lucky to be conscious in the one branch where you happened to avoid death compared to the rest of the population, or more likely, there is simply a conscious Scott in every possible branch out there.
You just have to wait and see.

292. Peter Byrne Says:

Scott,

Your essay is the most sensible account of the “mysteries” of quantum interpretation I have ever read. And, mercifully brief. Thank you.

Peter Byrne
Everett biographer

293. fred Says:

I have to add that, eventually, there will be an age after which being alive will probably mean constant utter misery.

So, if being conscious in every branch of the MWI is a thing, then it’s as much of a curse as a blessing.

294. fred Says:

Geza #287

“Once one allows this for Bell experiments, what forbids its use to explain any other discrepancy between my favourite theory and experimental results?”

I’m pretty sure that physicists who find superdeterminism interesting are well aware of this, but they are looking for a particular causal mechanism that would lead to QM observations.
Maybe it could also be possible to prove that superdeterminism can’t work, no matter what you try… if the idea is that stupid, maybe it should be feasible to prove it.

All we know for sure is that, according to current physics, there’s really no such thing as an independent choices either. But maybe you could also prove that this is wrong… that there is a way for Alice to make a choice that doesn’t depend on the prior history she shares with Bob.

295. Scott Says:

Peter Byrne #292: WOW, that means a lot to me!

296. Scott Says:

fred #291: In the Everettian picture, of course there are branches where I’m already dead, branches where I was never born, branches where I live for a million years, etc. But that wasn’t the question. The question was whether it matters decision-theoretically. E.g., should I take insane risks with my life, under the assumption that I’ll always “anthropically” myself in a branch where I happened to survive? My own answer is an unequivocal “no,” regardless of whether the other branches have the same reality as this one. If you disagree … well, you’re welcome to live your life as you see fit! 😀

297. Scott Says:

fred #290: Thanks! I enjoyed that article, as I almost always enjoy Natalie Wolchover’s work at Quanta. I clearly remember being at Perimeter Institute and talking with Lucien Hardy back when he was developing his causaloid framework 15 years ago. I really hope those ideas manage to link up with the rest of quantum gravity research.

That said, I confess that the papers I’ve seen so far on quantum information with indefinite causal structure haven’t really resonated with me. It always felt like “merely” a fancy, higher-level formalism for talking about perfectly ordinary quantum-mechanical situations—a formalism that could always be “compiled down” or “unraveled,” whenever desired, to statements in the usual language of quantum circuits with fixed causal order. So then, I kept asking myself, why not compile it down and be done with it? 🙂

This is probably related to my lifelong inability to understand type theory, category theory, lambda calculus, or functional programming languages—an inability that must surely be traceable to a missing lobe of my brain, or something, given the number of brilliant people who evangelize for those things, but which stubbornly remains nonetheless. Quantum circuits with indefinite causal order seem to me like neither more nor less than a kind of “quantum functional programming”—one where higher-level constructs, like unitary transformations, are treated as if they were lower-level constructs like states that can be manipulated in superposition. (And yet, I keep wanting to scream, they’re not! 🙂 )

298. mjgeddes Says:

#Scott 297

Well, they seem to be saying that the notion of ‘causality’ is emergent, which is certainly reasonable (Sean Carroll thinks that causality is emergent as well I believe). It seems that the notion of an ‘ arrow of time’ is more like a continuum, it can be more sharply defined (macroscopic world), or less sharp or even absent (microscopic world). Indeed, I suspect that the ‘sharpness’ of the arrow of time, is in some sense exactly equivalent to ‘the measure of existence’.

As to category theory, well, to a dumb-arse like me, all of abstract math is more or less equally incomprehensible any way, so I don’t see why category theory should have a different status to say algebra or calculus 😀 You can’t really have an arbitrary cut-off point where you doubt the utility of some field of math just because you’ve reached a level of abstraction where you’re not comfortable.

As far I can make out, the “3 magic words” for understanding all of pure math psychologically are: Symmetry, Connectedness and Mapping. Algebra is all about “Symmetry”, Geometry&Analysis&Topology is the “Connectedness” and mathematical logic (including set and category theory) is the “Mapping”.

In category theory, you have objects and arrows between them. The arrows are the “Mapping”, which includes an identity arrow and associative composition. That’s it. It does seem to me psychologically that the notion of “Mapping” (Category Theory) is quite clear and intuitive, and also distinct from “Connectedness” (Geometry/Topology) and “Symmetry” (Algebra).

299. Wyrd Smythe Says:

FWIW, I do believe we will eventually determine a mechanism for the Heisenberg cut, and then this vexing question will be solved. I suspect it will be some combination of decoherence and system size.

In Everett’s paper, he writes: “Alternative 2: To limit the applicability of quantum mechanics by asserting that the quantum mechanical description fails when applied to observers, or to measuring. apparatus, or more generally to systems approaching macroscopic size.”

He discards the idea saying, in part: “For what n might a group of n particles be construed as forming a measuring device so that the quantum description fails?”

But I doubt it’s as simple as finding a single n, although I think that may be a big part of it. I do think it will boil down to some interaction with the environment. I am definitely struck by how difficult it is to maintain a superposed quantum system. Such systems do get larger, but are still small and still require carefully maintained conditions explicitly to avoid interaction with the environment.

300. Bruce Smith Says:

Scott #271:

But then I thought about it some more, and I realized that this seems to prove too much….

FWIW, that was actually the conclusion I’d hoped you would draw, both about the Everett-related chain and the far-galaxy-related chain. The fact that “having a causal connection” is not transitive (when the chain elements can reverse direction) makes it problematic to draw any conclusions from that chain, it seems to me.

I agree with you that the two cases are different. But it doesn’t seem so easy to me to explain exactly why. They have in common that according to our model of physics, there can be no event in the future cone of both chain-end-events. They differ in why that’s true. But in both of them, the “existence story” requires a leap of faith. When you reach the Zorkians, they say “we used to be able to see the Zafflebergs (and we presume they got the signal we sent them way back then), but now we can’t see them, and if we sent a signal now, it could never reach them”. You and the Zorkians agree the model in which the Zafflebergs still exist is simpler, but you can’t verify it by experiment. (Maybe there can be similar situations involving a succession of communicating observers falling into a huge black hole.)

Or when you imagine your past self, and he imagines several variants of your future self, he can’t tell which one will become real — as far as he can tell, it could be any of them. If you decide to rely only on his knowledge, not your own, the other variants of you-now are as real as the real one. But of course, it would be strange not to use your actual knowledge about that!

The big difference to me seems to be in the “Everett split itself” (and I agree this really has nothing to do with QM — we could just as well be talking about “classical randomness”, or even “classical sense data which we didn’t know until we observed it”, in a deterministic world). The two possible outcomes of whatever local process “split” are, more or less by definition, incompatible with being “two events in a single world”, even though they are physically at the same place. (This doesn’t mean they can’t “both really exist” — only that they can’t “both really exist in the same world”. And the model in which they “both really exist” is again in some ways the simplest one.)

All that said, I do end up with essentially the same conclusion you do, about both cases. I’m just apparently less confident that I can justify that conclusion. Maybe it’s basically up to you (as a choice among equally valid points of view), as I think Tim Converse #273 is saying.

I do think “really exists” is likely a meaningless question about hypothetical universes, though. Saying something like “when I was young, it was possible I’d end up in Silicon Valley, but as it turned out, I didn’t”, might be perfectly accurate, and it points out that “exists as a future possibility” has a different meaning than “exists now” or “existed in the past”, and these meanings are relative to a point of view, which includes a place and a time (and a choice among possible worlds, if you want to believe in those, and probably an incomplete set of data about your world which you take as defining it). So the causal chain “for B3X, B1 existed in the past”, plus “for B1, B3Y is a future possibility”, when combined, only gives you “for B3X, B3Y is another way things might have turned out (but maybe didn’t)”. I think we can think more clearly by never trying to simplify those relations to just “something exists”.

301. Raoul Ohio Says:

Scott #297:

HaHa! I could never grok category theory, lambda calculus, or functional programming languages either (never tried type theory). As a grad student I made a brief attempt to read Russell and Whitehead — WTF? I read somewhere that a couple hundred pages into the interior of R&W, there is a footnote stating that a recent theorem implies that 1 + 1 = 2.

On the other hand, (I have it on good authority that) there are major advantages in functional programming languages. The good news is that the useful parts have now been incorporated into Java (and perhaps other full service languages). Thus you can take advantage of pragmatic FP constructs, without having to deal with anyone that might go off and start yammering about Lisp.

302. Andrei Says:

Geza,

“the criticism is not that a given superdeterministic model *implies* telepathy or superluminal signalling but that the superdeterminism explanation would equally well allow you to explain superluminal signalling even in a Minkowski world or the fact that we never observe telepathy even in a world governed by laws that allow for it…”

Superdeterminism is a generic concept like “field theory”. Is it possible for a field theory to predict telepathy? Yes. Does it mean that all field theories are non-scientific? Not really. General relativity or electrodynamics are field theories and are universally accepted as science.

The problem with your criticism is that it assumes that a superdeterminstic theory can at the same time:

1. Reproduce QM’s predictions (so that is not immediately falsified) and
2. Predict telepathy.

Now, if you accept that QM itself does not predict telepathy (if it does then your criticism is unfounded anyway), it follows that the superdeterministic theory that reproduces QM would not predict telepathy either. If it does, it would not reproduce QM, right? So, the assumption is wrong, and your criticism unfounded.

“To say (like the superdeterminism loophole to the Bell inequality does) “whenever somebody performs a Bell measurements the initial conditions ensure that the results violate the Bell inequality even though the conditions of Bell’s theorem (local hidden variables) are satisfied” does not seem different from saying “no, of course the laws of nature don’t allow telepathy, but the initial conditions of the deterministic universe are such that whenever you make these conjuring motions, burn a piece of cloth, and look in a glass ball you will perceive what the person last wearing the cloth is thinking right now”. (or, reversely: “sure, the laws of nature allow superluminal signalling, but the initial conditions are such that whenever you set out to do so, attempt will fail”).

Superdeterminism does not say anything about the “initial conditions of the deterministic universe”. ‘t Hooft’s model does not involve them at all.

There are some initial conditions that are relevant, but they involve the experimental setup, not the whole universe. But this is not different from general relativity or electromagnetism. Superdeterminism explains Bell correlations in the same way general relativity explains planetary orbits (correlations between the motions of massive objects). In general relativity the gravitational field at point A depends on the global mass distribution (including distant places as B and C). So, A is not independent of B and C, yet the theory is local.

In classical electromagnetism the electric field at A is not independent of the distant charges at B or C, but is a superposition of the fields produced by all charges. So, again, A is not independent of B and C, yet the theory is local.

Superdeterminism is the same, in fact ‘t Hooft model is a field theory, albeit a discrete one.

Once you impose the very constraining condition that the model reproduces QM you can be sure that no telepathy will be predicted. Science would be just fine.

“I view superdeterminism as the postulate of a cosmic conspiracy against experimental science, since it says that the initial conditions are chosen such that *all* measurement outcomes ever possible are not showing what a “free” observer (undetermined by the initial conditions) would see.”

Your view is wrong. The so-called superdetermistic “conspiracy” is no different from the other laws of physics. The experimenter cannot choose to perform a “forbidden” measurement because there exist no initial state that can possible evolve towards such a choice. It’s exactly the same type of conspiracy that precludes you from violating energy conservation. No matter what initial state you choose the energy will be conserved. Your choices are constrained by the laws of physics.

This is very important. Superdeterminism does not impose a finely tuned initial state. You can have any initial state you want (‘t Hooft really insists on that) with the only condition that the initial state satisfies some conditions, exactly like general relativity or electromagnetism. The “impossible” measurements are not impossible because we reject those initial states in an ad-hoc way, but because those initial states do not satisfy some physical constraints.

“It’s logically allowed, but will only ever convince those who like the predetermined outcomes. Once one allows this for Bell experiments, what forbids its use to explain any other discrepancy between my favourite theory and experimental results?”

It is the condition that the superdeterministic theory reproduces QM that forbids anything that is not compatible with QM. Simple.

303. Orno Says:

Scott #296: I don’t think it’s just taking insane risks where it matters – I certainly wasn’t thinking of whether it’s “safe” to run off a cliff because of anthropics and Everett when I posted my comment. But consider this: under Copenhagen, there’s an infinitesimal chance* that you would observe living to 200, which certainly makes those scenarios not worth thinking about much. Whereas under Everett, this and all associated weirdness is arguably guaranteed to happen. I’m curious: do you believe this really has no bearing whatsoever on the decisions you should make or on your views of the world? Because I’d really like to feel like you’re correct, but intuitively I don’t.

*Of course, we live in a world of advancing technology, cryonics and whatever, making this an existing possibility. But it would have been true for someone who happened to learn of MWI in 1000 BC.

304. Scott Says:

Orno #303: Yes, I really believe that belief or disbelief in MWI should have no bearing whatsoever on what risks you’re willing to take in your life. Or to put it differently: whatever about MWI causes you to think it would bear on that, that’s precisely the part you need to discard in order to continue along the Zen path. 😀

305. Scott Says:

Bruce #300: If I say that, e.g. my great-grandparents who I never met had a real physical existence, and that I know this among other reasons because I can talk to my parents who talked to them and who can relate stories about them, then don’t I seem forced to say the same about the Zaffleborgs, who are separated from me in space in just the same way my great-grandparents are in time?

306. Orno Says:

Scott #304: It’s not just taking risks, it’s also decisions like having children or saving for retirement that might be affected. Though arguably those are decisions about risks as well, and most likely you’re right about the Zen path anyways.

307. Scott Says:

Orno #306: The truth is that I live my entire life consumed by crippling regrets about everything that could have been but wasn’t—all the things I said but shouldn’t have, or didn’t say but should have, all the opportunities that I passed up for happiness or for saving the world or for writing or discovering things of lasting importance. But I’m not sure that that has anything to do with my level of belief in the many-worlds interpretation!

308. asdf Says:

Raoul Ohio #301, see if this does anything for you:

https://en.wikibooks.org/wiki/Haskell/Category_theory

It demystified Haskell types for me.

309. asdf Says:

Since we’re talking about stuff like superluminal signalling and getting rid of causality, can I ask if anyone has made sense of this preprint?

https://arxiv.org/abs/1910.02780

The authors claim to derive some parts of QM directly from special relativity, by taking into account superluminal solutions to the Lorentz transformation. They do some handwaving about what the paths of superluminal particles might look like, then do something that looks like a Feynman integral over those paths and QM-like equations fall out. I haven’t tried to read it carefully yet and am not sure that I can. I do have the impression that one of the authors (Arthur Ekert) is a legitimate QM and QC guy. I don’t know anything about the other author.

310. Tim Converse Says:

Scott #305:

No. There is a fundamental difference between your great grandparents and Zaffleborg.

There are multiple causal chains that lead from your great grandparents to you, including that your parents talked to your great grandparents and later told you stories about them.

There is no causal chain that leads from Zaffleborg (which we posit is outside our light cone) to you.

That is the difference.

So your great grandparents are *not* separated from you in time in the same way that Zaffelborg is separated from us in space.

You want to argue transitively that a “could visit them” link exists between us and Zork, and a “could visit them” link exists between Zork and Zaffleberg, but the lightcone limits cuts this chain in a way that permits no information to get to us about the existence of Zaffleberg, and vice versa. So if we posit that Zorkians might visit Zaffleberg, we are doing so without any way to know that Zaffleberg exists. And if we assume that they *could* visit Zaffleberg, then we have assumed our conclusion (Zaffleberg’s existence).

I suspect that this kind of faith in translating links from our position to links from the Zorkians’ position is ultimately equivalent to a Copernican assumption (which is the very thing that is at issue). Sure, if we assume Copernicanism first, then (unobservable) galaxies like Zaffleberg exist.

311. Vampyricon Says:

Scott #238,

Having thought about it for a few days, I think I know what new data would look like. QBism is instrumentalism, and GRW is a new theory, so I will ignore them.

New data for MWI would be an uncontroversial derivation of the Born probabilities from first principles. Of course, “uncontroversial” is doing a lot of work here, but if it is probability-interpretation-free and applicable to all cases (or maybe even sufficiently many), I think it would count as “uncontroversial”.

New data for hidden variable theories would be a fully relativistic formulation which can be extended to QFT while maintaining the intuitiveness of the pilot wave picture: something similar to particles riding on waves, but I personally am skeptical of the possibility.

312. Steven Evans Says:

Andrei Says: Comment #302

And now the “S” word doesn’t seem to be almost certainly nonsense again.

What is your argument against Andrei’s point, Scott?

It’s essentially the point I made earlier in the thread but with some technical meat added.

313. PA Says:

I find that all the interpretations are a bit too vague. For example, (point 5 of the FAQ) one can consider the observer as a quantum system that gets entangled with the system to be observed. However, two electrons are always entangled (identical Fermions). So maybe one needs to use a precise measure of entanglement and define a threshold. The system evolves according to Shroedinger’s equation up to the point where the entanglement is below that threshold. Observation occurs above the threshold. At this point, it should be possible to make precise predictions regarding experiment-observer systems (and we would have a new constant of physics: the observer-entanglement-threshold)

However, this would not be a regular QM, but a sort of spontaneous collapse kind of theory.

314. Morten Andersen Says:

Hi thanks for an enlightening post as always! However, one question regarding #4. As I understand it, it answers the Born question with an argument that the Born rule couldn’t reasonably have looked differently. But I have an added question: “What are the exact physical circumstances under which the Born rule should be ‘invoked’+” or something that touches more on why there is a Born rule at all, rather than why it has a specific formulation? I’m not even started on the Zen path and to me it seems it is in a sense the Born rule that breathes life into the branches, cf. the later question. So there is a Born-related complex (BRC) of questions: a) why is there a Born rule, b) what is the usage criterion (i.e. what counts as measurement), c) what effect do measurements have on the universe? It seems to me A+C are purely interpretational (e.g. c is bassically a “many-worlds-or-not” distinguisher) but the answer to b) seems more likely to have physical consequences because one can in principle measure if a super-position is preserved or not or if a Born-like collapse has taken place.

315. Scott Says:

Steven Evans #312: My argument is simply that the superdeterministic theory that does all these wonderful things — e.g., naturally reproduce QM while not reproducing superluminal signaling or telepathy — is a fantasy. I personally see no reason why anything like that should exist; at any rate no one has constructed anything like it.

But it’s more than that: I don’t accept the framing that this is a “promising research program that just needs more time to succeed.” I’ve seen nothing — nothing — of the slightest scientific interest ever come out of the superdeterminism program. And I don’t see why anything would, given that

(1) its original motivation was a terrible one (basically, people who didn’t understand or want to accept Bell inequality violation, and then after the latter was finally explained to them, responded by searching for some arbitrarily exotic way in which they might still be right), and

(2) the “mechanism” they decided on is effectively indistinguishable from magic — it’s just that you arbitrarily declare that this magic is only for violating the Bell inequality, and not for any of the more interesting things that magic would seem able to do once you introduced it into the universe.

I’m closing this thread tonight, honestly because I’m tired of superdeterminism discussion.

316. Bruce Smith Says:

Scott #305: I agree with Tim Converse #310 — the two situations are simply not analogous, once you notice the direction of causality along each connection. To the extent that they are each asking for a “Copernican assumption”, it’s a different one in the two cases.

I agree with you that the “Copernican assumption” being asked for in the Zafflebergs-exist-now scenario is a reasonable one. I also share it.

But I think it’s important to understand why the existence in both cases is a moot point in a *moral* sense. Nothing you could do now (or in the future) could influence the Zafflebergs, since no signal from you now can reach them. So you don’t need to know whether they exist, to decide what you want to do now. Likewise, nothing you can do now can influence the “you that might have been” or the world he resides in, so it also doesn’t matter to you whether that person or world exists or not. It’s a “non-physical question” — a matter of terminology, or of how you want to think about reality, rather than about reality itself.

==

As for the people confused by MWI (or its classical equivalent) into thinking they don’t need to pay attention to branches of reality in which they die (if they are really confused that way, and not just pretending to be, in some sense) — I think their error is related to a much more common error these days, of thinking “what I should do” is fundamentally determined by “what I would experience if I did that” rather than by “how it would affect the world if I did that”. Most of your effect is not on you personally (even while you’re alive, not to mention after you die), so these are pretty different in practice.

If for some reason you were only concerned with “effect on your personal experience”, it might make sense to pay attention only to futures in which you still exist — in fact, only to that near-term portion of them in which you do. So you could evaluate your actions using a weighted average over only those branches of the “multiverse”.

But if you are more sane than that, you should use a weighted average over all possible futures (weighted by their probability), including the ones you don’t exist in. In fact, to the extent possible, you should perform this averaging in the far future, even if that means you don’t exist in any branches by that time. (And nothing about this is incompatible with believing those other branches “really exist” even after you move into one of them and can no longer influence the others.) (If you are not a solipsist, you can derive evidence for this from the fact that other beings we observe appear to have evolved to decide things based on this method.)

In other words, I fully agree with your comment #304.

317. mjgeddes Says:

To conclude, I’d say that understanding QM is very hard 😉

Bohm and MWI try to establish an objective picture , but don’t quite get there (MWI is better than Bohm though, which conflicts with relativity theory). Copenhagen (and it’s more modern incarnation QBIsm) emphasize the role of observation and the subjective character of the wave-function, but are ultimately equally unsatisfying. What should we conclude?

I’m not convinced the best interpretation has been found yet. A better understanding of the nature of space and time could very well change the picture, in particular, I favour some new conception of space-time , quite possibly a ‘growing block universe’ model along the lines of what I talked about above.

In the mean-time, I’d have to go with a multiverse picture (that’s MWI), but it’s just not entirely satisfying because it doesn’t match up to reality of our conscious perceptions. Bohm/MWI ignores the subjective character of reality, but Cophenhagen/QBIsm misses the objective character. Some deeper picture could well reconcile the two.

To make the nature of reality understandable to us, as conscious observers, math alone just doesn’t really cut it. We rely on *psychological categories*, some of which cannot be precisely defined (they are ‘open-ended’). This lack of precision is not necessarily a weakness. Pure math can achieve perfect precision, but the cost is that it ceases to refer to concrete reality. So I think psychological categories are more powerful than mathematical ones.

Psychological categories always come as triples, examples:

Math = {Symmetry, Connectedness, Mapping}
Comp Sci = {Causality, Complexity, Compositionality}
Physics = {Mass-Energy, Space, Time}

Two psyche categories in a triple can always be considered to be analogous to ‘objects’ , and the other one is analogous to a ‘morphism’. So the nature of the psyche categories seem to mirror the nature of math categories (from category theory), while at the same time being more general and not precisely defined. We also see a recursive nature to the psyche concepts.

In conclusion, to us as observers, any real explanation of reality needs to connect a precise mathematical model with the imprecise psyche categories of mind.