Page 1 of 4
The cat's out of the bag
Posted: Mon Jan 10, 2011 10:38 am
by Jeremy Bernstein
Thanks to Martin and TCEC (and Chessdom), Houdini is getting some wider press. Congratulations!
http://www.chessdom.com/news-2011/houdini-chess-engine
Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 11:00 am
by Ted Summers
Nice!
Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 12:56 pm
by Ted Summers
This story is not getting any better. Here is the current game.
Source:
http://www.tcec-chess.org/div_i.php

Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 1:25 pm
by BB+
Rybka as White played a Four Knight's Game with d4, and then after the Queens were traded, played too passively in my opinion (move 11, for instance -- though as noted elsewhere, they are still in book until move 13: All opening moves are randomly fetched from a PGN file which contains 200.810 different openings. They are all fixed to 12 moves / 24 plies.). As a human I wouldn't go for the RR+6 vs RBN+5 with Nxa7 as Rybka did, as two minors tend to win these, especially when White has redundant majors. White's passed a-pawn (and then the b-pawn) ended up just being weaknesses, and after they were lost, it's going to be hard for White to hold on. IvanHoe has played some weird openings in this tournament (well, it seems that these will be flipped in the second half of the tournament), but managed to win with the Pirc as Black against Stockfish in Round 4.
Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 2:09 pm
by Ted Summers
Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 3:09 pm
by kingliveson
Ted Summers wrote:The complete game:
This is because Rybka's contempt need to be set to zero.
Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 3:46 pm
by ppetrov
kingliveson wrote:This is because Rybka's contempt need to be set to zero.
Rybka 4 has no contempt setting (Rybka 3 had). The same effect could be achieved by tuning piece values, but out of the box R4 plays without 'contempt'.
Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 4:04 pm
by Robert Houdart
BB+ wrote:As a human I wouldn't go for the RR+6 vs RBN+5 with Nxa7 as Rybka did, as two minors tend to win these, especially when White has redundant majors. White's passed a-pawn (and then the b-pawn) ended up just being weaknesses, and after they were lost, it's going to be hard for White to hold on.
Rybka played poorly, as I explained in the chat at Chessbomb there were at least 3 doubtful decisions:
1) 16.Nxa7: Going into an inferior R+P v B+N ending. Rybka took more than 23 minutes for this decision, and produced a +0.24 eval which turned out to be completely wrong.
2) 21.a4 and 22.a5: Pushing the a-pawn too soon just created a weakness, as Black has an extra piece to attack it.
3) 26.b4: Pushing the b-pawn created a passed pawn for black on the c or d-file, just for free.
Houdini didn't have to do a lot, even the "little Stockfish" version that is running at Chessbomb for the in-game analysis would have won today.
Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 4:31 pm
by kingliveson
ppetrov wrote:kingliveson wrote:This is because Rybka's contempt need to be set to zero.
Rybka 4 has no contempt setting (Rybka 3 had). The same effect could be achieved by tuning piece values, but out of the box R4 plays without 'contempt'.
Inside joke...
Re: The cat's out of the bag
Posted: Mon Jan 10, 2011 4:34 pm
by kingliveson
From talkchess:
Albert Silver wrote:Fascinating, and an incredible blunder by ChessDom. It isn't to besmirch Houdini BTW, as that is not the problem. It is the fact they are promoting a small private computer chess tournament, with no official recognition, on their front page as the main article, after only three rounds, which is the sort of reporting reserved for major international chess tournaments.... It is amazingly bad judgment, and beyond stupid.
Of course, I'm happy for Martin, but this changes nothing.
I knew this was not going to please the Rybka guys...