I'm Scott Aaronson, quantum computing/computational complexity researcher. AMA

Yes, when you look at the history of AI, a "quantum winter" is an obvious worry---i.e., a situation where the hype about QC becomes so unmoored from the reality as finally to cause the popular narrative to switch its polarity. Picture a pointy-haired boss who's now pouring money into QC, for no better reason than that it's new and faster and it's the future and people are talking about it and it could speed up his company's data mining by trying all possible answers in parallel. If something spooked that boss, one could imagine him joining a stampede for the exits with no greater understanding, canceling good research along with bad. If it happened to AI, then why not to us?

The fear of a quantum winter is one reason---if any reasons were needed besides the truth!---why I've spent so much effort on my blog over the past decade trying to counter irresponsible QC hype. Like, if anyone ever comes to me and says "you lied to me! it turns out that scalable QCs are really hard to build, and a lot more basic science needs to be done, and even if you did manage to build them, as far as anyone knows they'd only give you exponential speedups for a few special problems, not for most of the stuff my company cares about," I'll have a pretty enormous record that I can point to when I reply, "I WAS SCREAMING THAT AT THE TOP OF MY LUNGS AND YOU DIDN'T WANT TO HEAR IT!" For all the good it will do. :-)

Then again, maybe the worry is overblown. Some people claim that we've now passed the point where there will never again be an AI winter, any more than there will be an "electricity winter." The train just has too much momentum. Likewise, so long as the pressure continues to get more and more computing power and stave off the end of Moore's Law, it could be that QC will continue to entice people, regardless of the naysayers and regardless of how hard the engineering problems turn out to be.

I honestly don't know. I feel like I have a hard enough time understanding and communicating the truth about where this field stands in the present---and sometimes, trying to advance it by a small increment---without also prognosticating its future. :-) The latter involves all sorts of questions of economics, politics, and psychology that I have no special expertise about.

That's a tough one! So far, I've only seen ONE game meant to teach quantum mechanics that I thought actually worked, in terms of being (1) actually about QM rather than some vague analogy, and (2) fun to play. It's this one:

http://quantumgame.io/

Notably, this game doesn't even try to teach about entanglement (which, no surprise, is hard to keep track of in your head!). Tt deals only with a single photon passing through a network of beamsplitters and phaseshifters: a situation that has one foot in quantum physics and one foot in classical physics (the macroscopic state of a laser beam obeys exactly the same math). But the puzzles are really clever!

If you're just trying to create an educational game, why bother using the Quantum Experience? Won't it just introduce enormous errors and delays (~30 seconds per run when I tried it), complicating and obfuscating whatever you're trying to teach? Why not just make some puzzles that force people to reason about small, idealized quantum systems, along the lines of the successful example above?

Thanks; that made my day!!!

"cue-bit." I.e., the name of the letter Q, followed by "bit."

Qubits lose their quantum superposition states as they interact with their environment---effectively getting "measured" by their surroundings. This is called decoherence, and is the central engineering obstacle to building useful QCs.

The longer a qubit lasts while maintaining its quantum coherence, and (especially) the more operations you can do on it while keeping it coherent, the better that qubit's quality.

In practice, we will never be able to build qubits of perfect quality (i.e., ones that maintain their coherence forever except when we deliberately measure them). In the 1990s, some people thought this would be a fatal obstacle to scaling up QCs. But then two closely-related discoveries, called quantum error-correction and quantum fault-tolerance, dramatically changed the picture. These discoveries showed that, even if your physical qubits fall short of perfection, as long as they have a high enough quality, you can glom a bunch of them together into a single "logical qubit"---that is, a qubit that lives in the collective state of multiple physical qubits, and that can still be recovered even if any small number of those physical qubits lose their coherence. Furthermore, one can do an arbitrarily long quantum computation on these encoded (logical) qubits, continuously monitoring to see which of the physical qubits have suffered errors and correcting those errors (but not monitoring in a way that would collapse the logical qubits!). In this way, one can in principle build a reliable quantum computer out of unreliable parts---a generalization of John von Neumann's famous discovery from the 1950s, which showed that the same was true of classical computation.