Computer chess match on now

[Dan Scoones]

The leading chess-playing programs Houdini 3 and Stockfish 250413 are currently involved in a 48-game match at a time control of 150m+60s. Stockfish has just won game 25, making the score 3-2 for Houdini with 20 draws. The match may be followed live at this link: http://www.tcec-chess.net/live.php

[Dan Scoones]

Re: Computer chess match on now

After 28 games it is 4-4 with 20 draws.

[Mario Moran-Venegas]

Re: Computer chess match on now

Dan,
Is there a downloadable database (not single games) of engine games?

Mario Moran-Venegas

[Dan Scoones]

Re: Computer chess match on now

There is no obvious link on the site, but I believe the match organizers have a Facebook page.

[Mario Moran-Venegas]

Re: Computer chess match on now

Dan,
I meant any database of computer engine games, specially if rated 3100 or so.
I am thinking that the quality of the games might be better than those your species produces..

Mario

[Dan Scoones]

Re: Computer chess match on now

Try this link: http://www.top-5000.nl/dl/cct.zip

[Dan Scoones]

Re: Computer chess match on now

After 38 games the score is 5-4 in favour of Houdini, with 29 draws. Stockfish has the advantage in game 39 but Houdini is defending solidly.

[Jordan S. Berson]

Re: Computer chess match on now

The leading chess-playing programs Houdini 3 and Stockfish 250413 are currently involved in a 48-game match at a time control of 150m+60s. Stockfish has just won game 25, making the score 3-2 for Houdini with 20 draws. The match may be followed live at this link: http://www.tcec-chess.net/live.php

Dan,

Thank you for posting the link! I have been following this match closely for the past few days. Right now, Game 48 is underway, and Houdini leads with 5 wins to Stockfish's 4 wins... pretty close!!

One aspect of the match that I like is that every two-game mini-match features a different opening, and both computers must play the same variation (7-8 moves or so?). I don't know if that's usual, but it surely avoids repetition, which I'm sure is a major factor behind the decision.

The interesting part is how each computer does in certain openings:
1-2: D15 (Slav: 4.Nc3) - 2 draws
3-4: A04 (Reti) - 2 draws
5-6: E61 (KID: 3.Nc3) - Houdini wins with White / draw
7-8: B04 (Alekhine: Modern, Fianchetto, Keres Variation) - 2 draws
9-10: E10 (Neo-Indian: 3.Nf3) - Houdini wins with White / draw
11-12: C14 (French: Classical, Steinitz, 7...O-O 8.Nf3 c5) - 2 draws
13-14: A10 (English) - Houdini wins with White / draw
15-16: B22 (Sicilian: Alapin) - Draw / Stockfish wins with White
17-18: B12 (Caro-Kann: Advance, 4.Nf3 e6 5.Be2 c5 6.Be3) - 2 draws
19-20: C51 (Evans Gambit Accepted, 5.c3) - 2 draws
21-22: C25 (Vienna: 2...Bc5 3.Nf3) - 2 draws
23-24: B00 Nimzowitsch Defence: 2.d4 d5 3.exd5 Qxd5 4.Nf3 - 2 draws
25-26: A00 (Grob Gambit: e5) - Both computers win with Black
27-28: C02 (French: Advance, 5.Nf3 Qb6 6.a3 c4 7.Nbd2) - Stockfish wins with White and Black(!!)
29-30: B90 (Sicilian: Najdorf, 6.Be3) - 2 draws
31-32: A04 (Reti) - 2 draws
33-34: A01 (Nimzowitsch-Larsen Attack) - 2 draws
35-36: B18 (Caro-Kann: Classical, 6.h4 h6 7.Nh3) - 2 draws
37-38: E15 (Queen's Indian: Nimzowitsch, 5.b3 Bb7 6.Bg2) - Houdini wins with White / draw
39-40: E12 Queen's Indian: 5.Bg5 h6 6.Bh4 Be7) - 2 draws
41-42: E81 (King's Indian: Saemisch, 5...O-O 6.Be3 c5 7.Nge2 Nc6 8.d5) - 2 draws
43-44: C58 (Two Knights: Morphy Variation: 5...Na5) - 2 draws
45-46 B96 (Sicilian: Najdorf, Delayed Poisoned Pawn) - 2 draws
47-48 A22 (English: King's, 2.Nc3 Nf6 3.Nf3) - Draw / Houdini wins with Black

[Dan Scoones]

Re: Computer chess match on now

Houdini won game 48, making the final score Houdini 6, Stockfish 4, with 38 draws.

[Andy Shaw]

Re: Computer chess match on now

Houdini won game 48, making the final score Houdini 6, Stockfish 4, with 38 draws.

So I guess I know what I'm getting my mother for Mother's Day...

[Michael Yip]

Download match games

Is there an easy way to download the match games?

[Michael Yip]

Re:Game 9 of 48

http://reochessnews.blogspot.hu/2013...fish-game.html

I analyzed game 9.

Houdini 3 (3169) - Stockfish 250413 (3095)
Catalan Open[E04]
nTCEC - Superfinal - Season (9.1), 28.04.2013

[Paul Bonham]

Re: Computer chess match on now

Houdini won game 48, making the final score Houdini 6, Stockfish 4, with 38 draws.

We are probably 5 to 10 years away from the top computer engines no longer being able to beat each other at long time controls. Every chess game between them will be "ideal" in the sense that no winning initiative can be gained. Certainly not "ideal" in the sense of being entertaining.

As far as chess being a sport, computer engines are taking that possibility away. Once the top engines cannot defeat each other, can chess ever be called a sport, given that no sport is heading inevitably towards a predictable match result, and that is precisely what chess at this bleeding edge computer level is doing?

What is most interesting is the contrast between these dry, technical engine games and the exciting drama of the recent WC Candidates, in which errors were made by imperfect humans under extreme stress. Actually, though, the drama there wasn't so much in the games themselves (no great brilliancies that I heard mention of) but was in the determination of the final winner.

This does, however, open a philosophical debate for chess being a sport: if humans can never reach the level of computer engines and will always play imperfect chess, then chess between humans can be a sport if enough stress is introduced to force the humans to break down. The "sport" is in determining which humans will break down more than others. The entertainment value of the games isn't in the quality of the play, which is guaranteed to be imperfect, but in the awaiting of the point at which one or the other human makes the critical breakdown error.

So if we want chess to be a sport, we should not want humans to ever reach the levels of these engines. I wonder if Hikaru Nakamura and Zeljko Kitich are willing to engage in this philosophical debate?

Of course, the other way to make sure chess is a sport is to introduce one or more elements of luck or chaos. Once you do that, these engines are kaput. Which IMO would be a good thing, because I too was following these engine games when I could, and I was paying special attention to the commentary. What I noticed was that no one was analyzing the actual play, trying to find weaknesses or imperfections. No one cared about the actual play. Perfect or almost perfect engine chess is so dry and technical that no one really cares about the chess anymore. Chess as an intellectual pursuit has almost run its course. We are finally waking up to the fact that with perfect play, chess is not a draw, but a virtual infinite number of draws. All of them boring.

[Jordan S. Berson]

Re: Computer chess match on now

Houdini won game 48, making the final score Houdini 6, Stockfish 4, with 38 draws.

I wonder if Stockfish somehow knew it was in a must-win situation and overplayed the position...

[Tom O'Donnell]

Re: Computer chess match on now

... What is most interesting is the contrast between these dry, technical engine games and the exciting drama of the recent WC Candidates, in which errors were made by imperfect humans under extreme stress. Actually, though, the drama there wasn't so much in the games themselves (no great brilliancies that I heard mention of) but was in the determination of the final winner. ...

At the very least Aronian v Gelfand will go down as an amazing creative achievement: 26.Bh6+! and 30.g4! are a remarkable conception.

http://www.chessgames.com/perl/chessgame?gid=1712941