Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
×
Software Entertainment Games

Computer Chess Programs Vie "Live" For World Championship 73

Alex Laburu writes "The 17th World Computer Chess Championship is taking place in Pamplona through the 18th of May. As of this writing, Rybka (winner of the last two editions) is ahead of the pack and playing Shredder to consolidate its lead over Junior. You can watch the games live or otherwise follow the tournament asynchronously on the standings page, where you'll also find information about the hardware used by various teams."
This discussion has been archived. No new comments can be posted.

Computer Chess Programs Vie "Live" For World Championship

Comments Filter:
  • by GreenTech11 ( 1471589 ) on Sunday May 17, 2009 @02:41AM (#27985075)
    Better than lawnbowls AND curling, chess played by computers is the most popular spectator sport...
    • Re: (Score:3, Funny)

      by RsG ( 809189 )

      chess played by computers is the most popular spectator sport...

      Until they become self aware that is.

      The old movies got it wrong. Skynet is going to go live after being entrusted to win chess games for its human masters. It will unleash pawnageddon upon us all.

      Fortunately, the robots created to round up humanity will be easily bested. They'll line up, and move in grids, and they'll patiently wait for us to take our turn.

      • You missed the obvious joke:
        "Pawned!!"

        As for the movement, they can jump from Boston to New York in one move over a network.

        • by RsG ( 809189 )

          As for the movement, they can jump from Boston to New York in one move over a network.

          Only if the network is three grid spaces forward, one space to the side ;-P

          Failing that, if there are no pieces in the way, they could cross in a straight line, assuming it was either a horizontal, vertical or diagonal.

          • by AuMatar ( 183847 )

            3 grids forward? The cheaters!

            (Knights move 2 by 1, not 3)

            • by RsG ( 809189 )

              Whoops.

              Although, now that I think about it, three by one still works as a description, if you take the starting position of the piece as 1 instead of 0. That works out logically to 2 grids forward from where you started.

              So:
              3 empty
              2 pawn
              1 knight

              Knight moves from where it is at square 1 to a square one left or one right of the empty space.

              However, when I posted 3 by 1, that wasn't what I was thinking, I just didn't pay that much attention to what I wrote.

    • Re: (Score:3, Insightful)

      by Korbeau ( 913903 )

      It will be when computers start playing dramatic moves, or really know, read and adapt to their opponents well. As it is, I found that most comp vs comp games very boring in a strangely drawish way.

      • Re: (Score:2, Funny)

        Yeah, there is a complete lack of tension. Watching sport (to me at least) is a lot about how players use psychological maneuvers to get their opponent to make a mistake (or do something they would not usually do).

        Just imagine the commentary on a comp vs comp chess game:

        Jim: Comp1 just made the most optimal move under the constraints of its x-heuristics algorithm ...
        Joe: I agree Jim. He must be trying to optimize the variables of his problem.
        Jim: I see comp2 is taking his time choosing between two di
        • Re: (Score:2, Informative)

          by m00seb0y ( 677149 )
          You don't have to imagine anything. The Spanish chess master Leontxo Garcia is commenting (in Spanish) live on the games on the site.
          • Re: (Score:1, Funny)

            by Anonymous Coward

            I don't speak mexican.

        • That's the funniest comment I've read on slashdot since that Antarctic movie script idea. "He must be trying to optimize the variables of his problem." Classic!

      • As it is, I found that most comp vs comp games very boring in a strangely drawish way.

        It doesn't help that they play at a level higher than any human can comprehend. Think about that, even a grandmaster watching the game in realtime doesn't fully appreciate the rationale for what he is seeing.

        • I think you have to be quite a bit more proficient at a skill than whom you're observing to be able to understand the rationale.

          As a composer, quickly understanding a piece of music is an indication (to me) that it may not be very good.

    • Re: (Score:3, Funny)

      by kandela ( 835710 )
      It'd be better if it was Battlechess 3D.
  • If the Sarah Connor Chronicles has taught me nothing else, it is that chess programs are bad news.
  • Haha (Score:4, Funny)

    by KWarp ( 1556259 ) on Sunday May 17, 2009 @02:59AM (#27985137)
    This is the nerdiest sport ever. :D
  • Do they get FIDE ratings? It's great to see interest being brought to the game. They should have Anand play against the victor for a man vs. machine championship, like Kasparov vs. Fritz a few years back. Those are such elegant games to watch. http://www.youtube.com/watch?v=zRW9io3myOI [youtube.com]
    • by linhares ( 1241614 ) on Sunday May 17, 2009 @03:39AM (#27985275)
      They usually get estimates of their ratings

      As a chess researcher (in human cognition), I once had dinner with Doug Hofstadter and he mentioned his ideas concerning how a chess program should play, like humans do. It has been my goal for over 5 years now, and it's really hard. I could show Doug's idea that "analogy lies at the core of perception (of any scenario, including chess) by making psychological experiments in all levels, from novices to grandmasters. That work came out in journals like Cognitive Science, Minds & Machines, and New Ideas in Psychology (accepted). So I think we're on the right track. But my paper on the computational model [capyblanca.com] has been rejected three times, the last of which, fortunately, has good reports from referees who only want the piece to be rewritten.

      I long for the day in which Hofstadter's ideas would become more mainstream in AI and cognitive science.

      • Re: (Score:3, Interesting)

        by drmofe ( 523606 )

        Human play is elegant, but loses to brute force

        Brute force is not elegant

        "Elegant force" where human characteristics combine with the power of machine computation is the end goal

        BUT, winning chess against an equally-skilled opponent is a matter of finding forcing moves. When all forcing moves are known, chess is just a matter of brute force computation and massive memory

      • How is "analogy" not a synonym for "pattern recognition"? Nothing could be more central to AI and cognitive science.
        • You have a great question there. In my point of view, analogy refers to "experience recognition", not "pattern recognition". It's hard to put that down in a few words, but the idea is that the patterns (outside of any human understanding) do not really matter as much as we tend to think. They only serve as cues, and the recognition that arises in the brain is a function of both (outside) pattern and (internal, previous) experience. We propose that "experience recognition" ENCODES mostly experiences, ins
          • Well, I am open to the idea that pattern recognition is a subset of analogy, rather than vice-versa.

            We propose that "experience recognition" ENCODES mostly experiences, instead of patterns (outside of any understanding)

            I am having trouble with that, so maybe the following will not be relevant. But I see the brain as largely a pattern recognizer, however that doesn't mean people have a conscious perception of patterns. You can look at a wide range of objects and determine whether each is a "chair," but t

    • by Kjella ( 173770 )

      Do they get FIDE ratings? It's great to see interest being brought to the game. They should have Anand play against the victor for a man vs. machine championship, like Kasparov vs. Fritz a few years back.

      They get estimates, but even the single-processor machines they run them on now get 3000+ estimates, both from improvements in hardware and chess algorithms. And if we built a chess supercomputer to the best of our ability now, it'd be way past that again. So either Anand would be totally overrun, or it'd be because they crippled the hardware. Everybody knows that game is over, computers won on raw processing power. They do sometimes do matches with people in the top 100 but then usually the humans are give

      • Re: (Score:3, Insightful)

        by viyh ( 620825 )
        Interesting. Well, then computer chess geeks should refocus on cracking a game of Go. That should keep them busy for a while. :P
        • Re: (Score:2, Informative)

          Interesting. Well, then computer chess geeks should refocus on cracking a game of Go. That should keep them busy for a while. :P

          No doubt. There isn't a go program running on anything that I can't give a 9 stone handicap to and crush almost without thinking - and I'm only 2k. The day a computer beats a pro seems to be far in the distant future.

          • by rkit ( 538398 )

            These days are over.

            Go program beats professional 9 Dan at 7 stones handicap [ireport.com]

            On a side note, Zhou Junxun won the LG Cup 2007, so this is definitely a top player.

            • These days are over.

              Go program beats professional 9 Dan at 7 stones handicap [ireport.com]

              On a side note, Zhou Junxun won the LG Cup 2007, so this is definitely a top player.

              A 7 stone handicap is still HUGE. I've beaten a professional (Kano 7d) with that handicap.

              • by jvkjvk ( 102057 )

                A 7 stone handicap is still HUGE. I've beaten a professional (Kano 7d) with that handicap.

                Sure, but it certainly puts to rest your boast:

                There isn't a go program running on anything that I can't give a 9 stone handicap to and crush almost without thinking - and I'm only 2k. The day a computer beats a pro seems to be far in the distant future.

                So, not only has a computer already beaten a pro, but the pro was actually one of the top at the game, having won a major tournament.

                Care to wager about beating that program giving it a 9-stone handi yourself? Come on, you said that you could crush any go program running "almost without thinking" with that handicap.

                Go may be complex, and the complexities of strategic thinking are really hard (including even the most basic "big" vs. "vital" concepts), but clear

                • A 7 stone handicap is still HUGE. I've beaten a professional (Kano 7d) with that handicap. Sure, but it certainly puts to rest your boast: There isn't a go program running on anything that I can't give a 9 stone handicap to and crush almost without thinking - and I'm only 2k. The day a computer beats a pro seems to be far in the distant future. So, not only has a computer already beaten a pro, but the pro was actually one of the top at the game, having won a major tournament. Care to wager about beating that program giving it a 9-stone handi yourself? Come on, you said that you could crush any go program running "almost without thinking" with that handicap. Go may be complex, and the complexities of strategic thinking are really hard (including even the most basic "big" vs. "vital" concepts), but clearly Go computer programs have gotten way beyond where you thought they were. At this point, it seems like there are at least some programs that outpace your expectations. It is entirely conceivable that Go programs could be good enough in a mid term time frame that they give stones to all but the Dans & pros. Regards.

                  I downloaded the latest publicly available version of MoGo [www.lri.fr] (release 3) and I have to say I was fairly impressed. The program beat me a couple of times at a 9 stone handicap but now I can beat it. Here [hiico.com] is the first game I played. This [hiico.com] is the first game I won. The trouble is that go-playing programs make a systemic pattern of mistakes that's readily apparent after playing a few times. Mogo seems to have a much better "concept" of eyeshape than other programs I've played. Its main overall strategy seems

                • I just found an earlier discussion on slashdot [slashdot.org] about this program. I think two of the comments (this [slashdot.org] and this [slashdot.org]) might provide insight to someone who isn't familiar with the game.
                • At this point, it seems like there are at least some programs that outpace your expectations. It is entirely conceivable that Go programs could be good enough in a mid term time frame that they give stones to all but the Dans & pros.

                  I entirely disagree. And anyone who thinks that chess is comparible to go should read about Edward Lasker's experience with the game. [wikipedia.org]

                  • by jvkjvk ( 102057 )

                    So you disagree that in the next 40 years a Go program will be created that can win (>50% of the time) against a 1k while giving them 2 stones?

                    That is what I would consider the upper bounds of the limits of that statement (in my favor, I must admit, but hey, it's my statement!).

                    I'd like to take you up on that wager, since you "entirely disagree."

                    Say $1000?

              • by rkit ( 538398 )
                Of course seven stones is a lot. Also, this was not a tournament game. But still, this is amazing progress. Btw, an older version of mogo is freely available, try it out for yourself.
          • by Wildclaw ( 15718 )

            " There isn't a go program running on anything that I can't give a 9 stone handicap to and crush almost without thinking - and I'm only 2k."

            In what reality do you live? Even excluding the new generation programs that are as strong or stronger than you (2kyu-1dan), the old generation programs still hover around the 6-8 kyu range. That would be 4-6 stones handicap. You could possibly get that up to 9 if you trained to specifically beat a program. But it would in no way be easy.

            This isn't to say that go progra

            • " There isn't a go program running on anything that I can't give a 9 stone handicap to and crush almost without thinking - and I'm only 2k."

              In what reality do you live? Even excluding the new generation programs that are as strong or stronger than you (2kyu-1dan), the old generation programs still hover around the 6-8 kyu range. That would be 4-6 stones handicap. You could possibly get that up to 9 if you trained to specifically beat a program. But it would in no way be easy.

              This isn't to say that go programs will overtake humans anytime soon. While the Monte Carlo algorithm did revolutionise the go ai world, it basically meant a quick leap up from the old min-max based ones. But now the reality is beginning to catch up with the programs. Monte Carlo may be better than min-max but brute forcing is still not really viable even if you use a more efficent way of brute forcing.

              Sorry, I'm from Missouri. You'll have to show me.

            • " There isn't a go program running on anything that I can't give a 9 stone handicap to and crush almost without thinking - and I'm only 2k."

              In what reality do you live? Even excluding the new generation programs that are as strong or stronger than you (2kyu-1dan), the old generation programs still hover around the 6-8 kyu range. That would be 4-6 stones handicap. You could possibly get that up to 9 if you trained to specifically beat a program. But it would in no way be easy.

              This isn't to say that go programs will overtake humans anytime soon. While the Monte Carlo algorithm did revolutionise the go ai world, it basically meant a quick leap up from the old min-max based ones. But now the reality is beginning to catch up with the programs. Monte Carlo may be better than min-max but brute forcing is still not really viable even if you use a more efficent way of brute forcing.

              I live in a reality where there is such a thing as combinatorial explosion [wikipedia.org]. The higher levels of go playing are quite simply incomprehensible to beginners, much less computer programs.

        • There has been some progress:
          http://games.slashdot.org/article.pl?sid=08/08/08/1243244 [slashdot.org]

          Even though it was a biggish machine.

      • Everybody knows that game is over, computers won on raw processing power.

        Do you have any evidence to back that up?

        Deep Blue and Fritz were great chess players, but when pitted against the greatest humans, they were about equal. Fritz had mostly tied. Deep Blue lost one match against Kasparov, and won one match, and then retired. Both games Deep Blue had the advantage. It was programmed specifically against Kasparov, but Kasparov had never seen it play.

        I used to think that modern humans didn't stand a chance against modern computers; mostly I got that from the Sarah Connor Chroni

  • First, we played games.

    Then, we watched other people play games, and we played computer games.

    Next, we watched other people play computer games.

    And now, eliminating all human-to-human interaction, we watch computers play games.

    Who called us antisocial? ;-) Oh well, king's gambit ftw, rock out with your pawn out, good luck... "ladies and gentlemen"? Or is it "Nuts and Bolts"? "Plugs and Jacks"?

  • As a programmer, I am quite facinated.

    Without the semi-random input from a human, would computer chess programs eventually simply play a half a dozen different games (based off the more psudo-random beginning moved)

    I don't know anything about chess programs though, so I could be wrong with how my gut says they should behave.

    • Re: (Score:3, Interesting)

      Perhaps one day there will be a finite number of 'perfect' games which cannot be improved upon, and a computer that is set up to play perfectly is compelled to play one of them.

      However, chess is nowhere near played out. The computers in this tournament are still making mistakes that cannot yet be identified as such.

      There are two different chess engines in each match-up, so there's enough pseudo-randomness, as you call it, in the differences between the engines, to ensure that these games will be relatively

    • "Eventually"? I imagine that the entire problem space will be mapped, and the draw for black or white will decide the result. That's when we'll have to add the 3rd dimension, and ask the Vulcans if they fancy a game.
  • by Skuto ( 171945 ) on Sunday May 17, 2009 @05:14AM (#27985641) Homepage

    They restricted the hardware to 8 cores. As a result, the best programs, like those who run over clusters, are not playing.

    Supposedly, this was to make money not a factor. In reality, some very nice expensive dual Nehalems are in action.

    • by spinkham ( 56603 )

      The same organization ran 2 competitions at the same time. The Open Hardware Computer Chess Olympiad had no limits on hardware, and the World Computer Chess Championship has a limit to 8 cores.
      Both were won by the same team, running the Rybka chess engine.

      You can read their recap of the competition and get details of the hardware here:
      http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=11022 [rybkaforum.net]

  • Including the US Championship and a huge one held in Vegas every year. I'd love to see how this is done and would really like to attend one of these. Gotta dig in to the site and see the rules for draws: if the machines can offer them, or if only the operators. IIRC, when Kasparov was playing Deep Blue, it was up to Blue's operators to decide whether to accept an offer. The current leader has one draw and is up against #2, who has 2.

    That was one thing that fascinated me when I started working high-level

  • 9 games with a score of 8.0 [univ-lille3.fr]

    This is very interesting. Being a baseball fan, and thinking back to strat-o-matic as a kid, I can't help but think how I would code such a thing in that game/sport. Football too.

    One could be issued a limited number of "skill points" in different disciplines of the game, allocate among his/her team. Situational strategies could be coded so that when certain in-game criteria were met, specific functions could be called.
    Play Ball!
  • http://tty.wanfear.com/~mbrito/games/twoplayer/chess.html

The best defense against logic is ignorance.

Working...