Kurzweil usually makes predictions about things that have strong trends over time. His favorites are the trends that increase exponentially, because most people extrapolate the early, slow growth linearly and come to a completely wrong conclusion. If you watch his TED talk, he explains his technique. http://www.ted.com/talks/lang/eng/ray_kurzweil_on_how_techno...

Uh, yeah, except the funny thing about exponential curves is that at small scales they look a hell of a lot like sigmoid curves.

It also makes it easy to verify that he was right.

Kurzweil's oddly-specific dates have usually been correct.