Dan’s Biz Bookshelf: ‘Building a Story Brand 2.0’
It’s Only Common Sense: Why Should They Take Your Call?
But the other point is that there’s yet another classical algorithm on the stage, which is Selby’s algorithm, which I think was first announced on my blog. It’s a local-search algorithm, but it’s one that is able to figure out that the qubits are organized into these clusters. What the Google paper finds is that Selby’s algorithm, which runs on a classical computer, totally outperforms the D-Wave machine on all the instances they tested.
If I know that these eight qubits form a cluster, and I should be thinking of them as one giant variable, then I just find the best setting of that variable, and I’m done. There are only 256 — 2 to the 8th — cases to check. That you can do pretty quickly.
If the clusters were 800 bits, then you wouldn’t be able to do this. On the other hand, building 800 qubits that are all talking to each other is a super-hard engineering problem. And even if you did [build those qubit clusters], it’s not at all clear that quantum annealing would be able to tunnel over that.
Remember, quantum annealing does best when there’s a tall, thin potential barrier. When you make the potential barrier wider, as would happen if you had 800-qubit clusters, then quantum annealing would have trouble as well.
We’re finally seeing clearly the logical consequences of the design decisions that D-Wave made 10, 15 years ago, which were, “Go for as many qubits as possible as quickly as possible. Don’t really worry about their lifetime, about their coherence. Don’t worry about error correction. Don’t worry about solving something where we’re confident theoretically that there’s a quantum speedup.”
I think of them as taking the dirty approach, and most of the others are trying to take the clean approach. Of course, it is possible that the dirty approach will get somewhere before the clean approach does. There are many precedents for that in the history of technology, where the dirty approach wins. But it hasn’t won yet.
