America the nonexistent

A commenter on a previous post writes:

A lot of great discoveries came from non-scientific losers. E=MCC. Airplanes. America. Someone discovered how to make an airplane by playing with a box. Physics is mostly theoretical. America, I guess, is the most scientific discovery. They applied the scientific method to determine its existence, but they used no control group, and no placebo. For that, America’s existence is not yet proven. There seem to be other ways of establishing truth than just the scientific method. Scientists are contemporary soothsayers. They should use every means possible of proving a fact.

Despite its insightfulness and coherence, the above argument raises some immediate questions:

  1. What does it have to do with anything I said?
  2. E=MCC?
  3. What would mean to use a placebo or control group to test America’s existence? Would it mean sending a ship in a different direction, and checking that it didn’t also reach America? Would it mean verifying that America can’t be reached from Europe by foot — since if it could, then it wouldn’t be America, but rather part of Eurasia?
  4. Has England’s existence been scientifically proven? What about France’s?
  5. Where do so many people get the cockamamie idea that there’s such a thing as a “scientific method” — that science is not just really, really, really careful thinking? (I blame the school system.)

31 Responses to “America the nonexistent”

  1. Bram Cohen Says:

    People are, in fact, taught in school that ‘the scientific method’ consists of double-blind tests, and that’s the core of everything scientific. In actual fact, double-blind tests are only used rarely, and only apply to certain fields, and even then are only done when you’re so desperate for a controlled result that you resort to black box testing.

    Your comment about blaming the school system was probably a joke, but it is, in fact, true.

    Science is not, as you state, ‘thinking carefully’. That is a necessary but not sufficient condition for science. To do science you must reason based on the available evidence.

  2. Scott Says:

    Your comment about blaming the school system was probably a joke, but it is, in fact, true.

    Yes — like most of my jokes.

    Science is not, as you state, ‘thinking carefully’. That is a necessary but not sufficient condition for science. To do science you must reason based on the available evidence.

    Anyone who ignores available evidence is not thinking carefully.

  3. Greg Kuperberg Says:

    E=MCC?

    See here .

  4. Aaron Says:

    Apologies for the lengthy comment!

    Coincidentally enough, I’ve recently been thinking about this question myself. I mean really, has anyone every scientifically proven that the scientific method even exists?

    I tend to agree that science education in primary school is to blame for almost everyone having no clue what-so-ever as to what science is. Actually, come to think of it, I’m not sure most scientists even know what science is! I certainly don’t… which is why I’ve been reading up on it.

    I think most non-scientists think of science as being observations followed by explanations, i.e., effects and their causes, and this is exactly how “science” is pitched to kids in grade school. They think that being scientific means jotting down observations or actions in a notebook (“Birds all have similar body shapes.”) and then drawing a ‘careful’ conclusion about the cause of those observations (“There must be an archetypical ‘bird’ shape in God’s mind.”) But, this is a highly unsatisfying explanation of what science does, because it says nothing about how evidence is used to support or criticize hypotheses, fails miserably when there are statistics or errors involved, and provides no clues about how to sort among alternative hypotheses. Plus, it lets things like creationism in the door, and that idea ruins the party every time.

    I’ve heard most scientists say that science is about falsifying hypotheses (an idea due to Karl Popper), but this is unsatisfying too, since it’s devilishly hard to actually falsify anything completely, particularly when statistics are involved. Certainly, some of what scientists do is falsify things, but I don’t think we spend that much time doing it. I mean, just how much time do you really spend trying to prove negative results?

    Some people might say that what distinguishes science from non-science is puzzle solving (an idea due to Thomas Kuhn), i.e., in order to do science, you actually have to solve problems (prove theorems, derive solutions, build machines). And, while this too is certainly a necessary part of science, it’s still an unsatisfying explanation of why we have things like electron microscopes or flu vaccines. If puzzle solving was all there was, then we’d be like those crazy inventors with a garage full of junk that we piece together into various contraptions. That is, puzzle solving provides no direction to the accrual of knowledge that comes out of solving puzzles.

    So what distinguishes science from things like astrology or politics? Apparently, it’s the ability to learn from our previous errors in such a way as to make those errors happen less often in the future. For less empirical fields, like theoretical physics, computer science and mathematics, which don’t deal with real data and thus real errors, I would argue that the ‘errors’ here are the gaps in the theories, the cases and circumstances left out of the theorems we’ve proved. In all cases, it’s the filling-in of these gaps, of increasing the level of abstraction, and the explanatory or predictive power of our models, that is what scientists do. And, tautologically, science is what scientists do, otherwise, it would be science, and they wouldn’t be scientists.

    If this stuff sounds at all interesting to you, I highly recommend B.K. Jennings’ recent arxiv posting titled “On the Nature of Science“, which is short and to-the-point, if light on technical details. On the other hand, D. Mayo’s “Error and the Growth of Experimental Knowledge” is thorough and deep, describing why Popper and Kuhn’s ideas are an insufficient explanation of how scientists actually do science, why you can’t be even a little bit of a Bayesian, and how to properly use errors and statistics in science.

  5. Scott Says:

    So what distinguishes science from things like astrology or politics? Apparently, it’s the ability to learn from our previous errors in such a way as to make those errors happen less often in the future … In all cases, it’s the filling-in of these gaps, of increasing the level of abstraction, and the explanatory or predictive power of our models, that is what scientists do. And, tautologically, science is what scientists do, otherwise, it would be science, and they wouldn’t be scientists.

    I took a course on philosophy on science as an undergrad, and that’s more sensible than almost everything they told us there. At the risk of starting another flame war, most of what I read seemed to doom itself from the start in several ways:

    (1) Obsessing over the same few examples (GR, QM, the heliocentric theory)

    (2) Completely ignoring computational complexity and Occam’s Razor (which immediately banish the “grue problem” among others)

    (3) Failing to account for human stupidity (it can take decades of painstaking experimental work to learn a tautology)

    (4) Different standards for science and everyday life (questioning the “theory-ladenness” of atoms but not of the boogers in your nose)

  6. Greg Kuperberg Says:

    has anyone every scientifically proven that the scientific method even exists?

    Every year, many grade schools conduct double-blind trials to see if the so-called “scientific method” exists. They call these experiments “science fairs”. They never quite confirm their theory of the scientific method.

  7. Dave Bacon Says:

    Actually when I read that comment I feel kind of sad: “Paraphrenia” is late onset schizophrenia. Although I do like the idea of writing all of my equations in this longhand. A=PiRR. I’m a bit confused about how I would write e^(i pi )=-1 as I’m not quite sure how to write an imaginary pi number of “e”s.

  8. Robin Blume-Kohout Says:

    So what distinguishes science from things like astrology or politics? Apparently, it’s the ability to learn from our previous errors in such a way as to make those errors happen less often in the future.

    I think there’s a sense in which science could be described as evolutionary astrology. (There, now I’ve outraged absolutely everyone!) By which I mean:
    (a) the goal is to predict the future,
    (b) predictions compete; the bad ones get scrapped.

    So if astrology was actually required to work, it would turn into science. The phrase “…to make those errors happen less often in the future,” is an excellent one — what is an error but an incorrect prediction, weighted by the importance of the decision made on its basis?

    There seem to be two levels of selection going on. On one level, we select the theories and metatheories (GR and creationism are theories; Popper’s falsifiability criterion and “have faith” are metatheories) that most successfully predict the future and let us avoid costly errors. On another level, organisms that consistently choose stupidly get naturally selected out.

    Note to Scott: Bram’s admonition to “reason based on the available evidence” is not quite so easily absorbed into careful thinking. One might fail to reason based on available evidence without ignoring available evidence… if there exists no available evidence. I might think very carefully about the angelic capacity of pinheads, without ever ignoring evidence, and still not be doing science.

  9. Scott Says:

    I might think very carefully about the angelic capacity of pinheads, without ever ignoring evidence, and still not be doing science.

    And all this time, I thought I was doing science…

  10. Scott Says:

    On another level, organisms that consistently choose stupidly get naturally selected out.

    Robin, a possible hole in your theory: in the age of proteomics and recombinant DNA, not only does creationism show no signs of being weeded out, but those who believe in it are outreproducing those who don’t.

  11. Anonymous Says:

    So, what’s up with Occam’s razor?

    I mean, if theories A and B give exactly the same predictions, then who cares which one we choose. I agree, that if I have to use one, I’ll pick the one *I* find simplest for the task at hand. But if you use the other one because you find it simpler, who am I to say you say you picked the wrong one? We made the same predictions, afterall (by assumption).

    In mathematics, these things come up all the time (different bases for vector spaces, representatives for cosets, sign conventions, binary vs. decimal, model structures on categories,…) and everyone correctly understands it’s different strokes for different folks. (Or, rather, people fight tooth and nail, but they always feel silly the next morning.) Yet, if you say there’s no reason not to take the Earth as the origin of your coordinate system for calculating the orbits of planets, people think you’re ignorant of 400 years of science. Similarly, if someone took my favorite theory and added “Because God wills it,” to the beginning of every sentence, and insisted that all computations occur in base 1, I’d have no objection, though I might smile when they’re not looking.

    Why shouldn’t we completely forget about Occam’s razor?

  12. Scott Says:

    Why shouldn’t we completely forget about Occam’s razor?

    That’s an excellent question — and indeed, it was only really answered in the last few decades, in a series of results by Vapnik and Chervonenkis, Valiant, Blumer et al., and others.

    The short answer is that there are vastly fewer hypotheses with short descriptions than with long descriptions. And that means that, if you find a short hypothesis that explains the data you’ve seen, then you can say with reasonable confidence that that hypothesis will also predict most future data drawn from the same ensemble. The same is not true for long hypotheses, intuitively because of the problem of overfitting.

    Of course I just gave you a plausibility argument; what computational learning theory does is turn it into theorems. If you’re interested in this subject, here‘s one place to start; for more see the book by Kearns and Vazirani.

  13. Scott Says:

    Or, to address your question more directly: to see the advantage of a short hypothesis over a long one, you need stop thinking about them in isolation, and think instead about the sets of possible alternative hypotheses from which they were chosen.

    Here’s an example: if a sequence of bits starts 0000000, it’s a good bet that its continuation will be 00000000000000. But if the sequence starts 0101110, it’s not a good bet that it’s continuation will be 10011010110011. But what could possibly be the difference between these cases? If each bit is random, then aren’t the strings

    000000000000000000000

    and

    010111010011010110011

    equally unlikely?

    See if you can answer this one yourself. :)

  14. Aaron Says:

    Robin Blume-Kohout said…

    I think there’s a sense in which science could be described as evolutionary astrology. (There, now I’ve outraged absolutely everyone!) By which I mean:
    (a) the goal is to predict the future,
    (b) predictions compete; the bad ones get scrapped.

    Actually, astrology is a funny thing, because it really does seem like a science. It has its own impenetrable jargon, it makes predictions that seem to be right pretty often, its practioners engage in something that looks like peer review… to really see why it’s not, you have to think a little harder. That is, the jargon just obscures the fact that the predictions are only “correct” in retrospect, that the ‘peer review’ is really just the ruthless criticism of another school of thoughts’ assumptions (hmm… sounds like certain other disciplines, no?), and it has absolutely no capacity to learn from the predictions it gets wrong. I’m pretty sure that today, you could show that the gravitational pull of the Rocky Mountains actually has a greater influence over your atoms when you’re just born than the position of Jupiter, but that would be use a science to critique astrology, and that’s not allowed. But otherwise, yeah, real science is just like evolution applied to astrology…

  15. Aaron Says:

    Scott said…

    I took a course on philosophy on science as an undergrad, and that’s more sensible than almost everything they told us there.

    That’s kind of amusing actually. Each year, my undergraduate advisor pushed me to take some philosophy courses, and each year, I opted for physics courses instead. I’m kind of glad I ignored his advise all four years, now. Philosophy seems a lot more interesting in the context of history and in the context of answering some real questions, than it does as its own outright subject. But physics… now there’s a useful thing to know!

  16. Aaron Says:

    Greg, no one else has mentioned how clever your comment was, and I can’t let it go by without saying so… sadly, I’m sure the cleverness of it would be lost on many a science fair organizer…

  17. aram harrow Says:

    I took history of science classes, rather that philosophy of science classes, and would recommend them. Understanding science in its social, polical, cultural and economic context is one of the best ways to refute the picture of strict loyalty to “the scientific method.”

    For example, The Leviathon and the Air-Pump talks about the victory of Boyle’s experimental approach to science over Hobbes’ theoretical (or “arm-chair”) method. As a hint, I’ll mention that it was NOT because of experiment’s greater predictive power or carefulness of thought.

  18. Nagesh Adluru Says:

    Hey Scott! How did you get so much true insight into science? Is it because you do lot of science in “your own” way which happens to be true or you “somehow realized” it is the truth and then took it on!

    This is important in the sense that may be your way of realization can help build better system!

  19. anonymous2 Says:

    A recommendation for Scott, based on some of his previous posts:
    http://en.wikipedia.org/wiki/The_Two_Cultures

  20. Robin Blume-Kohout Says:

    Scott said,

    Robin, a possible hole in your theory: in the age of proteomics and recombinant DNA, not only does creationism show no signs of being weeded out, but those who believe in it are outreproducing those who don’t.

    We do seem to be living in a world that’s post-evolutionary, as well as post-modern, post-industrial, and frequently post-rational, don’t we?

    Seriously, though, not only does evolution take O(10^4) years to substantially change humans, but we’ve probably managed to muck up the process entirely by our control of the environment. Jokes aside, I seriously doubt that we’re selecting for anything at all right now — too much mobility in the gene pool, over 10 generations or so.

    My point w/re to evolution is that science is a variant of a very general activity: predictive modeling of our environment. This habit of ours evolved — like throwing rocks — for a reason: predicting the future helped our ancestors to survive.

    This can explain some things about why we act in the way that we do. Critters that didn’t bother to take any data on whether Object X was a yummy carrot or a hungry bear… got removed from the gene pool. On the other hand, critters that kept taking lots and lots of data before jumping to conclusions… right, them too.

    So I like to think that an evolutionary perspective on science explains why I jump to conclusions so often!

  21. Greg Kuperberg Says:

    Scott: Your example with binary strings is similar to a standard “paradox” of probability. Suppose that you are dealt 13 cards from a shuffled deck and they happen to all be spades. Then we should be suspicious because that outcome is said to be extremely unlikely, even though it is actually just as likely as any other outcome.

    My answer to the paradox is that we should be suspicious for the opposite reason: 13 spades is actually far more likely than most of the (52 choose 13) outcomes. That is because the deck is not strictly shuffled, but rather only allegedly shuffled, with some probability on the order of 9999 out of 10000.

    Occam’s Razor often amounts to a Bayesian estimate of probabilities.

  22. Robin Blume-Kohout Says:

    Aaron wrote,

    Actually, astrology is a funny thing, because it really does seem like a science. It has its own impenetrable jargon, it makes predictions that seem to be right pretty often, its practioners engage in something that looks like peer review…

    You could turn that around, too. Sometimes, certain aspects of science seem a lot like astrology. They appear to be scientific, but it turns out to be cargo cult. Sometimes it takes quite a long time for these bits to get weeded out. I can’t wait to look back when I’m 70 and see which bits of 2006 dogma turn out to be utterly absurd!

    There’s true, good, platonic-ideal Science… and then there’s science, which (as you pointed out) is all the stuff that scientists do. Some of the latter is actually astrology.

    I think you can tell the difference by asking:
    1) Does it really make predictions? You know, sharp, falsifiable things — not “You’ll find your heart’s desire.”
    2) Do predictors that consistently get it wrong get ejected from the meme pool?

    The horoscope in my local paper fails both of these criteria. Then again, so does string theory…

  23. Bram Cohen Says:

    Scott, I have to disagree with you about whether one can ignore evidence and still be thinking ‘carefully’. Many creationists reason with obsessive-compulsive scrupulosity, taking great care not to think a thought which might anger god. They’re thinking ‘carefully’, but not correctly or accurately.

  24. Anonymous Says:

    000000000000000000000

    So your shortest hypothesis is that your (or the intelligent designer’s) random number generator is out of order? Reminds me of the physicist’s and the computer scientist’s proof that all odd numbers are prime:

    3 OK, 5 OK, 7 OK, 9 inaccurate measurement, 11 OK, …

    3 OK, 3 OK, 3OK, 3 OK, …

    –ff

  25. Paraphrene Says:

    This post has been removed by the author.

  26. Paraphrene Says:

    “Actually when I read that comment I feel kind of sad: ‘Paraphrenia’ is late onset schizophrenia.”
    –Previous Poster

    “Paraphrenia” is an archaic term used before we really knew what the hell schizophrenia was. There are some disputes over what it means, and perhaps the Previous Poster should read further than the first Google result. Some think it’s late-onset schizophrenia. Some think it’s mild schizophrenia. Some think it’s the proper diagnosis for the eccentric genius. Maybe it’s when you hear voices, but they tell you you’re doing a good job. Technically, you can get away with hallucinations without being “abnormal.”

  27. Paraphrene Says:

    “Seriously, though, not only does evolution take O(10^4) years to substantially change humans, but we’ve probably managed to muck up the process entirely by our control of the environment.”
    –Robin

    I blame the filth of modern times on the burning of the Library of Alexandria.

    We’re getting a dry-spell of profound thought. Without those monumental documents embedded into our history, we’re left with a society that values Hollywood and McDonald’s.

    Our work is cut out for us: We have to rediscover the forgotten wisdom of the ancient world.

  28. Scott Says:

    We’re getting a dry-spell of profound thought. Without those monumental documents embedded into our history, we’re left with a society that values Hollywood and McDonald’s. Our work is cut out for us: We have to rediscover the forgotten wisdom of the ancient world.

    LOL! What makes you think that if we did somehow recover the Library of Alexandria, the McDonald’s eaters would care? Plato, Virgil, Euclid, and Archimedes are right there in Barnes & Noble, whenever they have a spare hour…

  29. Paraphrene Says:

    I went to a church, once. I wrote a paranoid story about it, but that’s a different thing. Rock music was the main attraction. It was all propaganda and the music was a weak imitation of popular style, but similar enough to the original to inspire the same gushing reaction from the crowd.

    I want to define their intention as accurately as my health will permit, as the topic makes me queazy, and so a whispering phrase has gathered into a sentence from my most neglected of mental recesses: “Make the bible come alive!”

    That’s why evangelical Christians have run amok in America. If not for the excitement of The Church…
    Haven’t you any concept of inspiration? You honestly believe in free will? The masses are attracted to bright colors. Right now, loud music and eternal bliss are more popular than ancient wisdom.

    Find a great philosopher with a lot of charisma, inflate his public image, and the McDonald’s eaters will care.

  30. Dave Bacon Says:

    [sarcasm mode on] Oh that makes everything better. Next time I have a friend come down with late onset schizophrenia, archaicly known as paraphrenia, I’ll feel real happy knowing that the word also might mean that my friend is really just an ecclectic genius. [sarcasm mode off]

  31. Paraphrene Says:

    [sincerity mode on] You must already have a schizophrenic friend, so you needn’t wait until “next time.” Apparently an extended vocabulary has only exacerbated your problem. My sympathies. [sincerity mode off]