Same here with the links. Neat paper/result. Best of luck scaling your systems up!

]]>http://www.ar-tiste.com/feynman-on-honesty.html ]]>

*Would a QC researcher have said anything remotely like this?*

Of course! The last 8 words could totally vaguely fit into a lecture on Bell’s Inequality!

I’m amazed that this guy managed to convince Deutsch and Zeilinger that he had any understanding of what they were saying. (Assuming, of course, it was a two-way conversation…)

Note to self: If a journalist ever interviews me for a math/science article, don’t let them quote me until I’m damn sure they know what I’m talking about. Or maybe university departments should hire PR people to convey scientific advances in understandable but correct terms.

]]>Since this is in response to a lecture on skepticism of quantum computing, let me try putting it slightly differently: if we “sell” quantum computing as being “ten times as fast” as classical computing (whatever that may even mean), then it’s a much easier claim to disprove. We do ourselves a disservice, I think, when we oversimplify what quantum computing is. Frankly, I think that we can sell the idea that quantum computers let us solve the same problems using new kinds of techniques that are in some fundamental sense faster. Not as in “PowerPCs are faster than x86s,” but as in multiplication *a la russe* is faster than repeated addition. We can illustrate the point, I think, by pointing out that quantum computing allows us to take advantage of the structure of some problems in a very subtle sense.

——-

Stas asks: Where is the borderline between “imagination” and falsehood? Stas, you have asked a key question (IMHO)… one which is at the heart of my own interest in QIT.

One answer that will occur to *Shtetl Optimized* readers is the orthodox scientific response:

In the long run, ‘falsehood’ strategies do not work in mathematics, science, and engineering, because logic, experiment, and field trials provide a (public and trusted) check-and-balance.

It is then very natural to ask,

How can this level of trust be “teleported” to messier disciplines like politics, business, ownership, education, and even ethics?

The empirical answer is that one can encode math, science, and engineering into end-to-end models and simulations, and then use these simulations as trusted foundations upon which to build enterprises.

To appreciate the exponentiating power of this path, just listen to Intel’s fantastic “Chip Talk” podcasts! Intel’s cadenced technical pathfinding for the next generation of 32nm and 22nm hardware, as a platform for the next generation of teraflop computing, provides an outstanding (IMHO) example of technical imagination.

Of course, this path of technology-based federation has been trod ever since Robert Moray formed the Royal Society … which throughout its history has embraced the practical and federative aspects of science (as John Gribbon’s excellent history *The Fellowship* describes).

Nowadays the federative power of science and technology is expanding exponentially, thanks largely to the usual suspects: internet connectivity, global databases, Moore’s Law increases in computation power and storage.

In consequence, humanity’s opportunities for initiating federative enterprises are (arguably) greater *right now* than at any time since the late eighteenth century.

Many peoples’ interest in QIT (well, my interest at least) stems from recognition that the availability of practical algorithms for efficient quantum simulation is presently cadence-limiting for extending this federative activity to quantum systems. To an engineering mind, Scott’s lectures on quantum noise and quantum learnability suggest powerful new approaches for advancing in this direction.

So the good news is, quantum system engineers now are taking a passionate interest in each new theorem that the QIT community proves … and an even *more* passionate interest in the mathematical concepts and techniques that are used to prove these theorems.

Even more exciting—albeit somewhat disquieting too—is the inescapable reality that the fundamental theorems and mathematical tools of QIT are destined to fill a central role in 21st century politics, business, ownership, education, and ethics.

IMHO, QIT researchers should not be too upset about this … any more than Robert Boyle was upset that his gas laws were used to create design rules for steam engines.

]]>Chris: Ultimately what matters with technology is what it does, not what it is or how it works. This fact motivates the way technology is communicated to non-experts. While you care a lot about “subtle and awe-inspiring” issues, remember that a 10x speed-up on an important problem over today’s state of the art is a huge deal.

]]>I only skimmed the paper Geordie linked above, but it looks like it’s about solving an optimization problem given some features. It seems to me the key to solving image recognition is in the right choice of features, not in optimization methods. So can an AQC do that any better than vision researchers?

]]>