Its seems to me nowadays computers simply play better, i.e. they consistently choose better moves. Where does human factor come into play, how can a human help engine to beat another engine?

Or was Kasparov simply wrong?

Edit: here's the podcast, most of it is politics, chess part is near the very end https://www.samharris.org/podcast/item/the-putin-question

  • My guess is that a computer thinks logically, and just calculates the best move coming, but a human person would be able to "see" any "possibility" of a tactic being pulled off... – VortexYT Nov 9 '17 at 21:38
  • 1
    Positional understanding – Jimmy360 Nov 9 '17 at 22:56
  • I seem to remember matches human+computer vs human+computer, but did anybody actually try human + computer vs computer only? – user1583209 Nov 10 '17 at 0:47
  • @bof december 2016, I edited in the link – Sejanus Nov 10 '17 at 11:07
  • 1
    Some good answers. Let me add the following article, which illustrates one legal chess position in which the human mind is better than any computer. While such positions seem very contrived and artificial, although legal, their study might lead to an answer to your question. telegraph.co.uk/science/2017/03/14/… – Paul Burchett Nov 14 '17 at 20:23

Experienced human correspondence chess players with a strong chess engine definitely play better chess than just the engine itself.

If you think about it, the human player could always simply follow the engine moves without any thinking. However, you're not going to win the world correspondence title if you don't do your own analysis. You have an engine, but your opponent also have an engine! You just can't win a game unless you guide the engine.

A strong International Correspondence Chess Grandmaster know how to analyze a chess position with a chess engine. They can make a move on a board, and evaluate the new positions again with the engine. There is no way the engine by itself can beat this human+engine perfect combination - tactics + long-term strategic thinking.

  • Human can make some moves on the board, that simplifies the search process (searching at depth x is not equivalent to x+1)
  • Human understands drawn endgames. They can guide the engine not to choose a line that wins a pawn but draw 50 moves later in a rook ending
  • Human can try the engine lines, play some training games, test the new positions etc. Chess engines can't do that by itself!

This has been tested in 2014. A strong version of Stockfish (ELO 3200) was pitted against Nakamura (2800) with an early version of Rybka (3000) to help him. The Rybkamura team lost.


So evidence is against Kasparov's claim.

  • 4
    The match shows that Nakamura + Rybka was weaker than Stockfish, but how does it prove that Nakamura + Rybka was not stronger than unaided Rybka? – bof Nov 11 '17 at 12:09
  • 1
    @bof, no it doesn't. That's why I used "evidence", not "proof" – jf328 Nov 11 '17 at 21:24

The general thought is that computers still have problems with calculating long-term plans and positional considerations correctly.

There are many examples of modern computers getting "confused" in closed positions where long-term planning is worth more than brute force thinking. Many have a contempt factor which forces them to make a poor move rather than accept a draw against a weaker opponent. Nakamura has exploited this algorithm in many famous games.

For positional considerations, the reason for this weakness is the rarity of possibilities. Except for some lines in the Ruy Lopez, Alekhine, and the Caro-Kann, capturing toward the center is the norm. There are exceptions where capturing away from the center is better, mostly for attacking purposes. Programming in the exceptions would cause an increase in the size of the program and slow down the speed. However the computer does search far enough ahead the this is becoming less of a concern.

  • 4
    Are you missing a "not" in your first sentence? – D M Nov 9 '17 at 23:14
  • Corrected first sentence. – Fred Knight Nov 10 '17 at 19:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.