Real, foolproof computer elo?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

S.Taylor
Posts: 8514
Joined: Thu Mar 09, 2006 3:25 am
Location: Jerusalem Israel

Real, foolproof computer elo?

Post by S.Taylor »

Computer level of play, is that if any human player can make certain middlegame moves which are truly correct but that the computer can not immitate, then that is already a plus for the human.
If the reason why the computer does not make that move is because it is not the best, then, the human did not demonstrate any superiority, so the the computer is better than him for sure.

Whatevere human finds moves otb which are best and computer cannot immitate, then the computer has not yet reached that humans level completely.

Still, we would like to see that the computer can see in the follow-up analysis what IS correct and best, and why a certain human move was not.
User avatar
Ovyron
Posts: 4556
Joined: Tue Jul 03, 2007 4:30 am

Re: Real, foolproof computer elo?

Post by Ovyron »

The game has only three results in the end, lose, draw and win. That means that as long as several moves don't change the outcome of the game, all these moves are equally good. It turns out positions in where there's only one move that wins that the engine misses and the human finds are rare, so rare that the human doesn't get to play them against the computer in a normal game.

For winning a chess game, all you need is your opponent blundering, you notice it, and take advantage of it, that's what computers are really good at, while humans are good at blundering. Computers find moves which are best that humans cannot imitate, more often, that's why computers have surpassed the human level.