> THough, the current s curve is in processors. We are still at the
> beginnning of the cost/performance curve with respect to
> multiprocessor/neural net computers. I think once we start to hit the
> wall with processors, we'll start to see more and more affordable
> multiprocessor PCs, which ought to last another decade at most before
> another wall is hit, at which point more exotic technologies will start
> to become available, as soon as the neural computers are cheap enough to
> make widespread use in engineering applications. This will bootstrap the
> affordability of the R&D to develop the new exotic technologies.
I don't think processors are the most important development. The combined
power of the global computer network triples every 10-12 months (I forget
who's law this is). What we need is more networked computers per person, and
more persons (until the advent of real AI.)
--Wax