alt Hacker NewsI wonder if this is like when Thinking Machines Corporation hired Richard Feynman mostly to get a little clout. They didn't have a lot for him to do, and he thought they were a bunch of naive kids with an idea that would never work in the real world but had a fat pocketbook. In the end he was able to mathematically prove that they could safely do a hardware optimization, but they didn't believe him and tried to overprovision instead, but late in the design they ran short of transistors and had to trust his math. It worked, but the machine was still mostly useless even when it did work properly.
Replies
the full article as a lot more context, including the fact that he reached out to Hong to join her startup rather than vice versa. he seems to be excited about the problem they're attacking and has a good idea of what he can do with a "founding mathematician" title, coming up with problems that will guide their research.
The commercial problem with Thinking Machines was that they had to ship a programmer with every one sold because almost no one knew how to program a SIMD machine in Occam