Re: "thousand-qubit"
Quantum annealing is suppose to instantaneously consider all possible solutions at once, and then settle on the solution directly.
No, that is not what quantum annealing is "suppose" to do, or even what it is supposed to do, at least by people qualified to make a sensible supposition about it. Quantum annealing superimposes the equally-weighted candidate states, then gradually evolves them in parallel toward the global minimum, with quantum tunneling between states. It doesn't "settle on the solution directly".
QA isn't really very different from classical simulated annealing algorithms, which simulate heating the system to force it out of local minima. (If you're familiar how e.g. Newton's method will get stuck at a local minimum, you should see how adding this sort of noise to the optimization process can bump the system out of the local minimum and let it seek a lower one.) QA offers the possibility of just jumping right from one local minimum through a "wall" into a lower one, so you don't have to heat the system up enough to climb that wall. The Wikipedia entry has a diagram that demonstrates the idea visually.
So all QA does is try to speed up simulated annealing by taking some shortcuts during the optimization loop.
DWave machines are not universal quantum computers (if they're quantum computers at all), because of how their chips are designed - specifically because of the topology of the qubit interconnects, apparently. You can't implement Shor's algorithm on them. (Well, you can, but it'd just be a conventional simulation, not a quantum implementation.)
This is noted in the StackExchange discussion, and in Scott Aaronson's blog.
In the SE discussion, Joe Fitzsimmons says that in principle QA can be used for universal QC, and I assume he's right about that; but he also notes that the DWave implementation isn't sufficiently general for that. Meanwhile, various researchers have implemented SA algorithms on conventional machines that outperform DWave on their own demonstration problems - see Aaronson for details.
Biting the hand that feeds IT
