Komodo loses the TCEC supermatch but wins the final game!

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

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
NATIONAL12
Posts: 305
Joined: Sat Jan 02, 2010 10:31 pm
Location: bristol,uk

Re: Komodo loses the TCEC supermatch but wins the final game

Post by NATIONAL12 » Sat May 31, 2014 9:11 pm

Yes I too noticed the Black Queen was totally immobilized all this time,i agree with your assessment of the game.
I thought Kings Gambit was a dead duck at this level and at best good for a draw.

Wonder how top engines would handle Sicilian Wing Gambit with big hardware and long time control.I had much success with it as both black and white but my latest AN just gets a draw.

This was the opening in which Highendman beat the Rybka forum team as you well know.

carldaman
Posts: 2004
Joined: Sat Jun 02, 2012 12:13 am

Re: Komodo loses the TCEC supermatch but wins the final game

Post by carldaman » Sat May 31, 2014 9:30 pm

M ANSARI wrote:Yes was a very impressive game and it was interesting to listen to all the chatter about which engine had the correct evaluation. It is hard to believe that the King's Gambit is playable at all these days on a human level, but really with the 2 top super engines it is almost unthinkable. With the King's Gambit accepted white sets himself up with a dangerously weak King that can easily backfire if one tiny mistake is made in development or tactical oversight. What we saw was an absolutely masterful play by Komodo that slowly and carefully took time to move the King to safety and then slowly started placing pieces on good squares and in one case gave a pawn up to give its pieces total coordination and an absolute strangle hold on the position. Sure the queen sac was pretty, but it was the entire setup and plan that really stands out. What really makes this game an eye opener is that Komodo evaluation understood the position while SF and Houdini were totally clueless until it was way too late. This could be a case that shows some problems in relying totally on lightening games to tune the evaluation. It is clear that both Houdini and SF value the queen much more than 3 pieces as it probably tests really well in fast time control games ... but looks like at LTC it can backfire. A beautiful game to see played that will go down as one of the best every played by an engine!
Great game for sure, but the last book move for Black, 8...Nge7, may well be inferior to the obscure 8...Bf6!?, pulling back the Bishop, guarding e5 and putting pressure on d4. I think Black should fare better after the sounder 8...Bf6, as White has to play the defensive 9.Be3 and only then Black should continue with 9...Nge7.

CL

NATIONAL12
Posts: 305
Joined: Sat Jan 02, 2010 10:31 pm
Location: bristol,uk

Re: Komodo loses the TCEC supermatch but wins the final game

Post by NATIONAL12 » Sat May 31, 2014 9:43 pm

You may well be right but the First 8 book moves were put up by Nelson from his stats to see how each engine could deal with position,SF drew and Komodo won.

carldaman
Posts: 2004
Joined: Sat Jun 02, 2012 12:13 am

Re: Komodo loses the TCEC supermatch but wins the final game

Post by carldaman » Sat May 31, 2014 10:21 pm

NATIONAL12 wrote:You may well be right but the First 8 book moves were put up by Nelson from his stats to see how each engine could deal with position,SF drew and Komodo won.
That's correct, but with a better move order there would have been no opportunity for 10. e5!!

NATIONAL12
Posts: 305
Joined: Sat Jan 02, 2010 10:31 pm
Location: bristol,uk

Re: Komodo loses the TCEC supermatch but wins the final game

Post by NATIONAL12 » Sat May 31, 2014 10:35 pm

I have made comments elsewhere about Nelson having to stick with 8 opening moves,on some openings it should be more and others less,i hope this will be adopted in next TCEC and give Nelson more freedom in his use of openings esp those used in current GM practice.
Also of course some neglected openings that may well be fine but need much more AN.

carldaman
Posts: 2004
Joined: Sat Jun 02, 2012 12:13 am

Re: Komodo loses the TCEC supermatch but wins the final game

Post by carldaman » Sat May 31, 2014 11:16 pm

NATIONAL12 wrote:I have made comments elsewhere about Nelson having to stick with 8 opening moves,on some openings it should be more and others less,i hope this will be adopted in next TCEC and give Nelson more freedom in his use of openings esp those used in current GM practice.
Also of course some neglected openings that may well be fine but need much more AN.
Very good, Paul. I've argued in favor of a variable length "book", with the exit positions being anywhere from 0-12 moves for both sides. For the SuperFinal, I'd also like to see 2 games reserved for the starting position (and thus no book).

User avatar
M ANSARI
Posts: 3467
Joined: Thu Mar 16, 2006 6:10 pm

Re: Komodo loses the TCEC supermatch but wins the final game

Post by M ANSARI » Sun Jun 01, 2014 6:52 am

I agree, the opening book should try to cover all aspects of chess, and it certainly would be interesting to have both engines start a game without book on both sides. I still like the lines used and I think the effort was to create drawn positions that needed proper play to stay drawn, while at the same time exposing any weaknesses an engine might have.

Post Reply