The Chess Mind

The pawn-odds match between the latest version of Rybka and the only version of Jaan Ehlvest* has concluded, and in some sense** the former won the match 5.5-2.5. The match score may seem especially depressing, given the material odds, but I think the match gave grounds for hope. After falling behind 3-0 and 4.5-.5, Ehlvest clearly figured out what he was doing, and finished the match with a win and two draws. Not bad - especially when you remember that Rybka wasn't going to get tired or depressed.

So I for one would like to see what would happen in a rematch. If Ehlvest can win such a match - and there's reason to think it's possible - that would provide good evidence that the gap between us and "them" isn't that big. Yet.

The remaining games are here.

*I'm hereby taking my stand against counterpart theory and person-stage views of personal identity.

**It's not clear to me that such a thing as Rybka actually exists per se (except, perhaps, as an abstract object), and if it does exist that it's the sort of thing that does something, and even if it (or a computer + program combo) does something, that the something in question is playing chess (as opposed to pushing electrons around in ways we find interesting). Listing all these qualifiers would make for tiresome writing and reading, so while they point to issues far more important than odds matches, I'll pretend for the sake of convenience that we can unproblematically speak of Rybka as a chess-playing entity.

The mini-match between Zappa Reykjavik and GM Erwin L'Ami has finished, and L'Ami managed a creditable 1-1 draw, holding his own in the second game without too much trouble. Congrats to the young Dutchman, even if he was the beneficiary of time odds (1:45 for him, :30 for Zappa).

Unfortunately, the other man-machine match, between Rybka and GM Jaan Ehlvest, is turning out disastrously for the carbon side. Rybka won all three games yesterday, and after a quick draw by repetition in round 4 (the "contempt factor" was set too low, I guess), Rybka won in round 5 as well.

Games from both matches, here.

The dog-and-pony show is on tour again, but this time with a twist. Man vs. machine matches have pretty much lost their value as a competitive spectacle; the interesting questions now are "How bad will it be?" and "Can humans win any games anymore?" Oh - and one other new question: how much material can the computer spot the human and still win?

That's the premise of the Rybka-Jaan Ehlvest match. Ehlvest is a strong Estonian/American GM once among the world's elite, while Rybka is currently the biggest, baddest bully on the computer chess block. How big & bad? Enough to give Ehlvest pawn odds - and win! To be fair, it was just one game, the computer had White, and the handicap was the miserable h-pawn. Giving that up was almost to its advantage, you might think. That might be overstating it, but in any case, it will give up different pawns in different games. Have a look here for some resources; as for the first game, see the link below.

The second match is more conventional, a two-game match, without odds, between Zappa Reyjavik (the successor to Zap!Chess) and Dutch GM Erwin L'Ami. And here we have good news to report: the human drew, with Black, and if anything missed chances for more! (Whether this bodes well for humanity, or can only be chalked up to L'Ami's strengths or Zappa's weaknesses remains to be seen.)

Here are the games.