14

In competitive correspondence chess, can humans win at all against computers? Do humans contribute at all to the game?

Provided that it is allowed by the rules, I fail to see how it is not just testing who has the best computer or the best computer program. In this scenario, the players would just be an interface, or not even that, since the computer can also send a move through the internet to the opponent.

7

I must disagree completely with the above answer. In the correspondence tournaments that I have participated in using a computer was considered cheating and if you were the tournament winner you had to do an analysis of one of your winning games, chosen by the tournament directors, to prove that you knew what you were talking about.

Second the computers main advantage against human players is their speed. A computer calculates thousand upon thousands of instructions per second and is able to analyse thousand of possible moves on each of its turns. However the computer has no true intelligence, which a human player does. The computer program Hydra was built to completely dominate the chess world and it did in fact beat every other chess program in the the world and has never been beaten by a human over-the-board. However it was beaten by a correspondence player in two games and drew their third game. I have been unable to find if the player consulted a computer or not, but as was pointed out in the discussions of the games on www.chessgames.com the player, correspondence chess International Grandmaster Arno Nickel, wasn't simply following another program since Hydra could beat all other programs as the time.

To say that unaided humans have no chance at all as though it were an obvious given is simply wrong. Take away the time constraints, have a motivated skilled correspondence player and there is a strong chance that the player will still win.

Hydra ran on a specially designed system that spread the processing required out over a network and is the only chess engine that would have a chance against the correspondence world champions if they were not constrained by time controls.

  • 1
    I define "competitive correspondence chess" as that played under the ICCF banner, because that's the only federation that's part of FIDE, and they allow computers. I'll try to dig up some sources for my comment when I have some time. – RemcoGerlich Apr 16 '13 at 13:53
  • The correspondence chess tournament that I was part of was through the USCF. My understanding in that tournament was that computers were not allowed. I believe that the USCF is part of FIDE. It has been a few years though, maybe they have changed things. – Edward Goodson Apr 16 '13 at 16:23
  • I was able to verify that the ICCF dose not ban the use of chess engines in correspondence play. iccf.com/Message.aspx?message=225 – Edward Goodson Apr 16 '13 at 16:36
  • 1
    "The computer has no true intelligence" the submarine isn't really swimming – Evan Harper May 24 '13 at 22:33
5

In competitive correspondence chess everybody uses the strongest computer they can get, but there are still consistent differences in strength between different players.

Computers are strong, but they're nowhere near perfect. They're extremely good in positions where calculation is the primary factor but not that good in endgames and positional play. Good correspondence players know which computer lines to trust and how to direct the computer to the critical lines.

Unassisted humans have no chance at all, of course.

  • Does that mean the "super GMs" (above 2700) are not human? ;) – Travis J Apr 12 '13 at 2:14
  • 3
    A super GM playing unassisted correspondence chess against a good computer using correspondence player has no chance. – RemcoGerlich Apr 12 '13 at 7:21
5

As others have noted, in international correspondence play, conducted under the auspices of ICCF, computers are allowed. Correspondence chess has been an interesting battleground for this debate because it has been stated above, the primary strength of the computer is brute force calculation and speed. When Botvinnik was first attempting to design a chess computer program, he was hoping to develop a true AI. Instead, modern computers win because they simply can calculate many more possibilities and pathways than a human could in a given amount of time. A human might miss the best continuation simply because he or she doesn't have adequate time to analyze it. Correspondence chess solves this issue because the human has considerably more time to analyze, whereas the computer doesn't really benefit from being able to examine the same position for several days. Interestingly, I should point out that there is quite a lot of evidence supporting the argument that humans contribute greatly to overall performance.

First, back in 2005, Arno Nickel, one of the strongest correspondence chess players in the world, played a four-game match with Hydra, which at the time was arguably the strongest chess-playing supercomputer in the world. Nickel was allowed to use a PC and chess software that he had bought off the shelf, similar to what he would use in an ICCF tournament. In a straight-up match between Hydra and the basic PC software, Hydra would have won every game. In the actual match, the match ended early because Nickel crushed Hydra 2.5-0.5 (two wins and one draw for Nickel). In other words, the combination of man and weak machine was too much for the best super-computer on the planet.

Second, there is a variant of chess called Centaur chess. Centaur chess involves a human-computer team. Humans make the move, taking into account analysis by the computer. At first glance, one would expect that this is simply a matter of who has the best computer software (and many of the players have developed their own chess programs). However, Centaur tournaments allow participation not just by human-computer teams, but by stand-alone computers without human assistance, and usually about half the participants are such stand-alone computers. To my knowledge, every one of these tournaments has been won by a centaur, i.e. the human-computer teams always beat the computers.

So, while it is true that the average human player would have little chance against an equal human playing with a computer, it is also true that computers have little chance playing against an equal computer playing with a human. All of which suggests that humans still contribute to the game, even when computers are involved.

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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