Congratulations to Komodo team for winning TCEC

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
User avatar
Leto
Posts: 2028
Joined: Thu May 04, 2006 1:40 am
Location: Dune

Re: Congratulations to Komodo team for winning TCEC

Post by Leto » Mon Nov 30, 2015 1:28 pm

Congratulations to the Komodo team and thanks to all those involved for making this an excited final to watch.

Peter Berger
Posts: 390
Joined: Thu Mar 09, 2006 1:56 pm

Re: Congratulations to Komodo team for winning TCEC

Post by Peter Berger » Mon Nov 30, 2015 9:53 pm

M ANSARI wrote:Only few games remaining but even if Komodo loses all remaining games it will still win TCEC. Congrats to all that worked on Komodo and a really impressive showing. Komodo has shown superiority in evaluation accuracy, and to a chess player that is one of the most useful tools for analysis ... especially when there is no clear tactical advantage in a position. Komodo is also no tactical slouch and was able to match SF monstrous tactical ability in most cases. I still feel that SF was suffering some bugs and that probably hurt its play, but I still feel that even without bugs Komodo would have won. It just seemed that Komodo was the better chess engine and had the more accurate evaluations. Sure it did suffer some defeats due to amazing tactical ability of SF deep search, but those were very rare instances.
My own assessment of the match is completely different.
With a match result of 9-2 89= Komodo 9.3 is 24 ELO points ahead ( actually even statistically significant just a tiny little bit stronger than Stockfish021115).
I purchased Komodo from Chessbase recently for like 70 bucks ?! I got version 9.1 - judging from Larry Kaufman's posts Komodo reaching version number 9.3 has improved by about 30 points since then. The GUI deterioated compaired to my last buy ( Deep Junior Yokohama) as it lost the ability to display the ECO code of the opening played.
Also Stockfish has improved a little recently .It is reasonable to assume that latest Stockfish is just a tiny little bit stronger than the version competing in the tournament.
Most of the match featured moves I cannot judge. There were a few instances were Stockfish played a move that can be judged to be kind of silly even by a player like me who is *very*far away from GM strength. Major differences between both programs seem to be about evalution of pawn structure and bishop vs. knights mostly. This suggests a few changes for Stockfish that could be tried. Someone will try them IMHO.
I don't think commercial programs will be able to compete in the long run. Stockfish can be grateful for Komodo's existance as there is an obvious goal to achieve - surpassing Komodo. This doesn't work as nicely the other way round.
Peter

User avatar
Leto
Posts: 2028
Joined: Thu May 04, 2006 1:40 am
Location: Dune

Re: Congratulations to Komodo team for winning TCEC

Post by Leto » Mon Nov 30, 2015 10:27 pm

Peter Berger wrote:
M ANSARI wrote:Only few games remaining but even if Komodo loses all remaining games it will still win TCEC. Congrats to all that worked on Komodo and a really impressive showing. Komodo has shown superiority in evaluation accuracy, and to a chess player that is one of the most useful tools for analysis ... especially when there is no clear tactical advantage in a position. Komodo is also no tactical slouch and was able to match SF monstrous tactical ability in most cases. I still feel that SF was suffering some bugs and that probably hurt its play, but I still feel that even without bugs Komodo would have won. It just seemed that Komodo was the better chess engine and had the more accurate evaluations. Sure it did suffer some defeats due to amazing tactical ability of SF deep search, but those were very rare instances.
My own assessment of the match is completely different.
With a match result of 9-2 89= Komodo 9.3 is 24 ELO points ahead ( actually even statistically significant just a tiny little bit stronger than Stockfish021115).
I purchased Komodo from Chessbase recently for like 70 bucks ?! I got version 9.1 - judging from Larry Kaufman's posts Komodo reaching version number 9.3 has improved by about 30 points since then. The GUI deterioated compaired to my last buy ( Deep Junior Yokohama) as it lost the ability to display the ECO code of the opening played.
Also Stockfish has improved a little recently .It is reasonable to assume that latest Stockfish is just a tiny little bit stronger than the version competing in the tournament.
Most of the match featured moves I cannot judge. There were a few instances were Stockfish played a move that can be judged to be kind of silly even by a player like me who is *very*far away from GM strength. Major differences between both programs seem to be about evalution of pawn structure and bishop vs. knights mostly. This suggests a few changes for Stockfish that could be tried. Someone will try them IMHO.
I don't think commercial programs will be able to compete in the long run. Stockfish can be grateful for Komodo's existance as there is an obvious goal to achieve - surpassing Komodo. This doesn't work as nicely the other way round.
Peter
I'm guessing you have the ChessProgram 14 GUI. It does still display the eco codes. Go to File, Options, Game tab, and make sure there's a checkmark on Show Openings. When an opening is reached on the board it will display the eco code and the name of the opening/variation on the status bar below the board.

Henk
Posts: 5826
Joined: Mon May 27, 2013 8:31 am

Re: Congratulations to Komodo team for winning TCEC

Post by Henk » Mon Nov 30, 2015 11:22 pm

I did not like the openings they played.

APassionForCriminalJustic
Posts: 415
Joined: Sat May 24, 2014 7:16 am

Re: Congratulations to Komodo team for winning TCEC

Post by APassionForCriminalJustic » Tue Dec 01, 2015 3:37 am

Peter Berger wrote:
M ANSARI wrote:Only few games remaining but even if Komodo loses all remaining games it will still win TCEC. Congrats to all that worked on Komodo and a really impressive showing. Komodo has shown superiority in evaluation accuracy, and to a chess player that is one of the most useful tools for analysis ... especially when there is no clear tactical advantage in a position. Komodo is also no tactical slouch and was able to match SF monstrous tactical ability in most cases. I still feel that SF was suffering some bugs and that probably hurt its play, but I still feel that even without bugs Komodo would have won. It just seemed that Komodo was the better chess engine and had the more accurate evaluations. Sure it did suffer some defeats due to amazing tactical ability of SF deep search, but those were very rare instances.
My own assessment of the match is completely different.
With a match result of 9-2 89= Komodo 9.3 is 24 ELO points ahead ( actually even statistically significant just a tiny little bit stronger than Stockfish021115).
I purchased Komodo from Chessbase recently for like 70 bucks ?! I got version 9.1 - judging from Larry Kaufman's posts Komodo reaching version number 9.3 has improved by about 30 points since then. The GUI deterioated compaired to my last buy ( Deep Junior Yokohama) as it lost the ability to display the ECO code of the opening played.
Also Stockfish has improved a little recently .It is reasonable to assume that latest Stockfish is just a tiny little bit stronger than the version competing in the tournament.
Most of the match featured moves I cannot judge. There were a few instances were Stockfish played a move that can be judged to be kind of silly even by a player like me who is *very*far away from GM strength. Major differences between both programs seem to be about evalution of pawn structure and bishop vs. knights mostly. This suggests a few changes for Stockfish that could be tried. Someone will try them IMHO.
I don't think commercial programs will be able to compete in the long run. Stockfish can be grateful for Komodo’s existance as there is an obvious goal to achieve - surpassing Komodo. This doesn't work as nicely the other way round.
Peter
It doesn't matter how you try to write this man, a 9-2 result is a massive difference. Forget about statistical significance or 24 ELO, Stockfish's play was rather poor - and there were plenty of positions where it sat there in the negative throughout the game. Unlike some, I don't personally fall for distorted perspectives. You can believe what you want. There is nothing close about 9-2. Stockfish got spanked; under TCEC conditions Stockfish is the inferior foe.

You know how many times I've heard that argument that commercial engines won't keep up in the long run?! Two old men right now are beating Stockfish in the face - undoubtedly an entire community. That is good enough for me. There will always be someone, somebody to fight. But I do think that just having one engine being number one makes computer chess rather dull. We don't need another Rybka era or Houdini era. Therefore, having both Stockfish and Komodo fighting it out is exactly what we all need.

Ferdy
Posts: 4111
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Congratulations to Komodo team for winning TCEC

Post by Ferdy » Tue Dec 01, 2015 9:25 am

Reviewing round 3 game, Komodo-Stockfish.
[d]r2r2k1/pp2bpp1/1qn1bn1p/3p4/7B/2NQPN2/PP2BPPP/R2R2K1 w - - 6 14
How white will continue?
Komodo played Nb5 intending to block the isolated pawn at d5 by Nbd4 later. This is a well know method of dealing an isolated, blocked it then attack it. Sf keeps its queen active by placing it at b6 attacking the weak b2.

Then some moves later, Komodo has just played h3.
[d]2rr2k1/pp1bbpp1/1qn2n1p/3p4/3N3B/3QPN1P/PP2BPP1/R2R2K1 b - - 0 16
Why? what is the idea? If you follow the game one neat idea that I learned from this is black will play g5 breaking the pin, so white will play Bg3, but black will follow up it with Ne4 attacking the bishop in g3 but now, white has Bh2 keeping the bishop pair, so the trick was to play h3 :)

And here is a position that is difficult to assess, with common black isolated pawn at d5, but white has doubled isolated pawn! in b file but has a good knight outpost at d4. There are open file and half-open files, there is two bishop from both sides and each side will attempt to get a 2-bishop advantage. How the engines played from this position?
[d]2r1r1k1/pp1bbpp1/5n1p/3p4/3N3B/1P2P2P/1P2BPP1/R2R2K1 b - - 0 19
The result is a draw but there are a lot of lessons to learn from this game.

Post Reply