Comment by SAI_Peregrinus
2 months ago
The main issue I have with Kurzweil's predictions is that they assume exponential growth, instead of sigmoid. Exponential growth doesn't really happen in physical processes, eventually they run out of inputs & level off. The hard part of predictions is figuring out where they level off.
Fair enough. I find he mostly tries to stick to the exponential bits and avoid things that are about to level out. Like he focuses on compute per dollar that can run a lot and not on Moore's transistors per unit area which is already getting close to atomic limits.
Where I find him to go off the rails a bit is anything that's not an exponential extrapolation, like I've never really seen the nanobots thing.