The Unbearable Heaviness of Computer Analysis

Having a solid opening repertoire and knowledge of typical plans is very important. But that doesn’t mean you need to memorize lots of computer analysis. – Nigel Davies

The subject of chess students lending undue weight to computer analysis of the openings recently came up in discussion with GM Davies and friends on Facebook.

Folks tend to forget that the computers are guessing just like we are. It’s axiomatic in the mathematical discipline of Game Theory that it’s not possible to find provably the best chess move without calculation to all terminal positions. Unless a computer’s brute force calculation is either exhaustive (not mathematically possible on current computing platforms in the opening) or intersects its Nalimov tablebase, the computer is operating strategically/heuristically (i.e., “guessing”) just as we do, though with a greater speed than we can muster and a flawless memory for the variations it has already calculated.

The other fact many learners don’t forget, but rather, can’t quite grasp, is that there are many paths through the forest to the inevitable draw. All that chess requires is that one stay on one of the paths to the draw and make sure one’s opponent steps off the path and onto the path to perdition instead of doing so one’s self.

Incidentally, some time ago WGM Natalia Pogonina remarked on Facebook that the Russian Chess Federation has access to supercomputer time for openings analysis, and that the more processing time given to any sound opening, the closer the computer heuristic evaluation of advantage approaches 0.0 (complete equality). Which seems obvious, but it’s worth noting.

The following game is presented without computer analysis of any kind. My opponent, a nice fellow and one of my “regular customers”, groused to the Tournament Director as he walked out the door, “That was no fun, he just rolled over me, I never had a chance!”

Jacques Delaguerre