nTCEC - Season 2 Superfinal - Stockfish vs Komodo now live

[Martin Thoresen]

Stage 4 finished last night and 2 engines qualified for the Superfinal.

The total number of games for the Superfinal will be 48 (black/white of the same opening).

The 24 openings are created by Nelson Hernandez from the rybkaforum.
You can read his explanation here:
https://www.facebook.com/tcec.chess/pos ... 9149266285

List of participants, double round robin:
3117 - Stockfish 191113 (dev)*
3098 - Komodo 1142 (dev)*

*Indicates update since Stage 4.

All information and rules:
http://www.tcec-chess.net/viewtopic.php?f=13&t=20

Live games + schedule and results:
http://tcec.chessdom.com/live.php

Finished games in the Superfinal will appear here:
http://tcec.chessdom.com/superfinal.php

Hope to see you all there.

Thanks,
Martin

[shrapnel]

I'm posting this right now on 23rd November....... others should note !
It's a TERRIBLE mistake of the Stockfish team to use this particular Development Version (Stockfish 191113).
In my LTC Tests, it is able to only draw with Houdini 3 at Contempt 1 ( forget about Contempt 0 ).
In short, this development version is very weak and Komodo should relatively easily win the Superfinal.... provided of course the Komodo version isn't also a regression !
SF would have been better off sticking with the previous winning version.
Mark my words !

[Uri Blass]

I'm posting this right now on 23rd November....... others should note !
It's a TERRIBLE mistake of the Stockfish team to use this particular Development Version (Stockfish 191113).
In my LTC Tests, it is able to only draw with Houdini 3 at Contempt 1 ( forget about Contempt 0 ).
In short, this development version is very weak and Komodo should relatively easily win the Superfinal.... provided of course the Komodo version isn't also a regression !
SF would have been better off sticking with the previous winning version.
Mark my words !

I think that it is terrible mistake of you to use a very small number of games at long time control(you certainly could not play more than 50 games at long time control in the last 4 days) and get the conclusion that stockfish 191113 can only draw with houdini3 at contempt=1.

[shrapnel]

I think that it is terrible mistake of you to use a very small number of games at long time control(you certainly could not play more than 50 games at long time control in the last 4 days) and get the conclusion that stockfish 191113 can only draw with houdini3 at contempt=1.

Touche' !
Yes, true that I've played only a small number of games....but there is such a thing as "gut feeling" and my instinct tells me that while this SF version isn't exactly weak as such , it lacks the 'magic' and inspiration of some of the earlier versions.
Score is tied 2-2 at the moment and of course I could be wrong !

[S.Taylor]

I'm posting this right now on 23rd November....... others should note !
It's a TERRIBLE mistake of the Stockfish team to use this particular Development Version (Stockfish 191113).
In my LTC Tests, it is able to only draw with Houdini 3 at Contempt 1 ( forget about Contempt 0 ).
In short, this development version is very weak and Komodo should relatively easily win the Superfinal.... provided of course the Komodo version isn't also a regression !
SF would have been better off sticking with the previous winning version.
Mark my words !

I agree and i just cannot understand if it reverted back to the previous version, because the version in stage 4 seemed better than anything. It both won the most games plus didn't lose ANY.
If the version being used now was the same but with a few improvements, then i would understand.

Indeed it has already lost 3 games.Is that because the komodo is so much greater than the one in stage 4?

[Martin Thoresen]

Standings after 16 games:

Code: Select all

N Engine           Pts    SB Stockfish 1 Komodo 1142 

1 Stockfish 191113 8.0 64.00             + 4 = 8 - 4 
2 Komodo 1142      8.0 64.00 + 4 = 8 - 4             

Live games + schedule and results:
http://tcec.chessdom.com/live.php

Finished games:
http://tcec.chessdom.com/superfinal.php

[syzygy]

I agree and i just cannot understand if it reverted back to the previous version, because the version in stage 4 seemed better than anything. It both won the most games plus didn't lose ANY.

You really should take a statistics class some day.

[S.Taylor]

I agree and i just cannot understand if it reverted back to the previous version, because the version in stage 4 seemed better than anything. It both won the most games plus didn't lose ANY.

You really should take a statistics class some day.

I don't deny statistics.
But i always want to see more than just small differences.
And i don't really enjoy seeing an engine which i want to have as a tool, lose games, which suggests it still doesn't see things, so i still can't trust it.


(and consequently, I look for other tests for it to pass, besides just statistics).

[IGarcia]

Fantastic tournament, great web pages with games info and chat.
We have seen very good games and the final was, at the end, very close to draw.

Thanks Martin for this new Season.
Congratulations for Komodo team and all other programmers involved (Mathias GUI, SF team, et all)

Ignacio.

[Modern Times]

I predicted during the qualifying rounds that Komodo would win, and indeed it did !" Not often that my predictions are right.... Congratulations to the Komodo team and the late Don Dailey, well deserved victory. It was very close and could have gone either way, but on the day it was Komodo that prevailed.