Komodo strikes first at the TCEC Superfinal

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.
rabbits23
Posts: 144
Joined: Fri Sep 12, 2014 2:57 am
Location: Randwick Australia

Re: Komodo strikes first at the TCEC Superfinal

Post by rabbits23 » Tue Nov 17, 2015 10:41 pm

Ted : You have no trouble with Stockfish 6 playing 6)... Be6 I gather? I
thought the move was dreadful because it locks up its position. Yes
eventually it got a draw but who knows what it could have done if it had
freed up the centre at the beginning.
Al

JJJ
Posts: 1286
Joined: Sat Apr 19, 2014 11:47 am

Re: Komodo strikes first at the TCEC Superfinal

Post by JJJ » Wed Nov 18, 2015 12:14 am

Yes, there will be unbalance opening at the end, maybe it will change a little.

User avatar
M ANSARI
Posts: 3408
Joined: Thu Mar 16, 2006 6:10 pm

Re: Komodo strikes first at the TCEC Superfinal

Post by M ANSARI » Wed Nov 18, 2015 5:31 am

Another interesting game ... game ... game 47. I think in this match Komodo looks much more seasoned than the SF ... but here in this position SF calmly plays NxB !!! Komodo goes crazy and evaluation jumps to almost +3.00. It really looked like SF was hit by a bug as it seems the plan is simple and for K to just push the king and take the black pawn by force and then simply march the passed pawn to promotion and game over. However things are not so simple and incredibly black somehow can draw this! This time it is SF that makes Komodo look like a patzer. However it also shows that humans really have no chance at all against engines anymore. The resulting draw is so deep that would be impossible for a human to calculate. The drawn position ends with black having 2 pawns vs. white having a pawn and queen !!!

[d]8/8/2b1k1p1/1p1n1p1p/5P1P/P1B3P1/2K1B3/8 b - - 0 96

Waschbaer
Posts: 68
Joined: Mon Dec 12, 2011 10:27 pm

Re: Komodo strikes first at the TCEC Superfinal

Post by Waschbaer » Wed Nov 18, 2015 8:12 am

In TCEC chat Mark explained this.
In Komodo the table base will not be asked for in every states of the running programm.
So Komodo switched between the evaluation of the endgame with and without checking the TB, the high eval was the one without asking the TB.

User avatar
AdminX
Posts: 5154
Joined: Mon Mar 13, 2006 1:34 pm
Location: Acworth, GA
Contact:

Re: Komodo strikes first at the TCEC Superfinal

Post by AdminX » Wed Nov 18, 2015 8:59 am

rabbits23 wrote:Ted : You have no trouble with Stockfish 6 playing 6)... Be6 I gather? I
thought the move was dreadful because it locks up its position. Yes
eventually it got a draw but who knows what it could have done if it had
freed up the centre at the beginning.
Al
[d]r1bqkb1r/1p2pppp/1np2n2/p7/P1p5/N4NP1/1PQPPPBP/R1B2RK1 b kq - 0 8

Having the benefit of now seeing how the game developed, I can safely say no. Had a move like 8. ... g6 been played, simply returning the pawn then I think we would have witnessed a less lively game. I know on the surface a move like 8. ... Be6 looks bad as it blocks the e-pawn. However, since white could not gain and advantage from say 9. Ng5, I see nothing wrong with it.
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers

whereagles
Posts: 561
Joined: Thu Nov 13, 2014 11:03 am

Re: Komodo strikes first at the TCEC Superfinal

Post by whereagles » Wed Nov 18, 2015 4:53 pm

Ansari: yeah SFs escape in game 47 was WAY beyond human (and reptile lol) realm.

rabbits23
Posts: 144
Joined: Fri Sep 12, 2014 2:57 am
Location: Randwick Australia

Re: Komodo strikes first at the TCEC Superfinal

Post by rabbits23 » Wed Nov 18, 2015 5:06 pm

Maybe Ted but I'm not convinced.But that's the beauty of chess I guess; there are no totally 'correct' answers. Well; none that I know of.
Allan

rabbits23
Posts: 144
Joined: Fri Sep 12, 2014 2:57 am
Location: Randwick Australia

Re: Komodo strikes first at the TCEC Superfinal

Post by rabbits23 » Wed Nov 18, 2015 10:46 pm

Apologies Ted. Of course there are totally correct moves or positions in
chess.I wasn't thinking deeply enough. King vs King is a draw. No debate
on that. It's the more complicated positions in chess that challenge the
mind to provide absolute evidence that I was referring to.Still I can't
prove my belief with move examples so I guess your argument is equally valid.
Allan

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

Re: Komodo strikes first at the TCEC Superfinal

Post by Peter Berger » Fri Nov 20, 2015 3:40 pm

It seems there is mostly one evalutation feature where both engines tend to disagree significantly: knight outposts.

Komodo loves them and evaluates them higher than Stockfish does. And this has worked very well so far.

User avatar
M ANSARI
Posts: 3408
Joined: Thu Mar 16, 2006 6:10 pm

Re: Komodo strikes first at the TCEC Superfinal

Post by M ANSARI » Sat Nov 21, 2015 4:22 pm

20 drawn games in a row and still going. I don't think I have ever seen that many games without a decisive result. I know that Komodo is ahead with so many draws I can't help but think that the engines are equal strength. However they seem to do it differently ... Komodo does it with better evaluation while SF does it by going through deep search or brute strength.

Post Reply