0

IBM's "Deep Blue" computer can beat any human being, including a former world champion such as Gary Kasparov. Thus, it follows that Deep Blue must give a handicap to a human player in order for a human to have a chance of winning.

In games like golf, handicaps are relative to "par," that is, the number of strokes a "scratch" golfer would require to play a hole or a field. By definition, most players require a positive handicap relative to "par;" a few can get by with a negative handicap. In "Deep Blue's" case, all the handicaps would be positive.

Has Deep Blue, in fact, given such handicaps to humans (a pawn, piece, or more)? If so, what were the results? Can a handicapping system be constructed around how large a handicap a human player would need to beat "Deep Blue?"

1
  • I still doubt Deep Blue can beat any human being. Had it not been for Kasparov's psychological lapse in Game 6, Kasparov could have at least drawn the match with Deep Blue. Note that Kasparov won a game against Deep Blue in the match that he lost. I think someone like Carlsen has a great chance of beating the 1997 Deep Blue in a match.
    – Wes
    Nov 18 '14 at 23:26
2

I'm going to answer this question from a literal interpretation - dfan's answer gives some nice background that may be what you're looking for but doesn't answer the question directly.

There are three things to note about Kasparov - Deep Blue specifically:

  • The match was actually really close. There have been accusations that IBM actually cheated by not just using an engine but also GM assistance during the match, and even discounting this you could probably argue that Kasparov would have come out on top if it weren't for the distractions surrounding the games rather than the games themselves. As a result it seems fair to assume that Deep Blue was at beast around the same strength as Kasparov was, and so pawn odds would probably be unnecessary as it would give the human too much of an advantage.
  • Deep Blue was dismantled almost immediately after the match, so the answer to whether it's possible to play an odds game now against it is no. Similarly, nobody has played against it since the second match.
  • It's very likely that at some point the developers did play odds games against the engine while developing it - GM Christiansen was involved with the development but it's very common for software/hardware developers to test things out as they make them and they wouldn't have been able to test out the machine's full functionality without odds because they likely weren't world-class GMs themselves.
2

Note that IBM's chess computer was called "Deep Blue", not "Big Blue". It was dismantled a long time ago and current chess engines are significantly stronger than it anyway.

There have been some handicap matches against computer chess engines, although I'm not aware of any very recent ones.

In 2007 Rybka played a match against GM Jaan Ehlvest with pawn odds, winning 5.5-2.5.

In 2008 Rybka played a match against GM Dzindichashvili with pawn and move odds, winning 2.5-1.5.

Also in 2008 Rybka played a match against GM Milov with two regular games (winning 1.5-0.5), two pawn and move odds games (losing 1.5-0.5) and four games with exchange odds (losing 2.5-1.5).

I'm not aware of more recent handicap matches. At this point I think that the handicap required for a GM to have an equal chance against the strongest chess engines would be large enough that the GM would not find such a match appealing.

3
  • OK, fixed the question to say "Deep Blue". Thanks for the heads up and answer. But even if a GM wouldn't play the engines at a handicap, maybe "ordinary" players would be willing to, and pay a fee for the privilege.
    – Tom Au
    Nov 18 '14 at 17:33
  • 2
    In August 2014, Stockfish played two games against GM Hikaru Nakamura with pawn odds and won 1.5-0.5. chess.com/news/stockfish-outlasts-nakamura-3634
    – JiK
    Nov 18 '14 at 19:27
  • @JiK Thanks. From the article, looks like the odds were "pawn and move" (he played White).
    – dfan
    Nov 18 '14 at 19:40

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.