Note that this is a 1997 paper, seemingly written before Kasparov's rematch and overall loss to Deep Blue. I imagine there have been more ideas that have become standard in the 20 years since, though machine learning still seems to not be used much.

Blatant self promotion for anyone interested in this subject, I wrote up a series of blog posts about the evolution of Chess and Go AIs with some more detail and pretty pictures: http://www.andreykurenkov.com/writing/a-brief-history-of-gam...

Coincidentally the latest Phrack has a tight little chess playing AI written in just 256 bytes: http://phrack.org/issues/69/4.html#article