Nakamura vs Stockfish, public match 8/23

[bhlangonijr]

man vs machine events are so rare now specially with top players that I wouldn't be critic of any format. Excellent news. Thanks for the initiative!

[Uri Blass]

I always like to see proof that these top engines are really playing stronger than humans. Engines are tuned to win from other engines. So there is a chance that all engines make the same kind of mistakes but their opponent engines are not able to make an advantage of these mistakes.

Well all you need to do for that proof is watch any human chess tournament. Even when all the players are super grandmasters there's still a bunch of blunders being made every game.

This is simply not correct and even players with rating below 2000 have draws with no blunder.

I am not sure if you can find blunders in most of the draws between karpov and kasparov.

Can you show me that most draws between them are games when one of them missed a win?

I do not claim that they played the best moves but non accurate move is not a blunder.

[Ajedrecista]

Hello:

Perhaps, with the f-pawn, he stands a chance to get a draw.

Analysis by Houdini up to depth 20 with 1 GB of hash:

Code: Select all

Without a7 pawn:

info multipv 1 depth 20 seldepth 41 score cp 53 time 47699 nodes 41128697 nps 862000 tbhits 0 hashfull 277 pv d2d4 g8f6 b1c3 d7d5 c1f4 f6h5 f4d2 g7g6 e2e3 b8c6 g1f3 f8g7 f1e2 h5f6 e1g1 e8g8 a2a3 c8f5 h2h3 d8d7 f3g5 e7e5
bestmove d2d4 ponder g8f6

------------------------

Without b7 pawn:

info multipv 1 depth 20 seldepth 49 score cp 95 time 40918 nodes 35440242 nps 866000 tbhits 0 hashfull 261 pv e2e4 c7c5 g1f3 e7e6 b1c3 b8c6 d2d4 c5d4 f3d4 d8b6 d4b3 g8f6 a2a3 b6b8 f2f4 e6e5 d1f3 b8b6 c3d5 f6d5 e4d5 c6d4 b3d4 b6d4 f4e5 d4e5 f1e2
bestmove e2e4 ponder c7c5

------------------------

Without c7 pawn:

info multipv 1 depth 20 seldepth 45 score cp 95 time 69324 nodes 61978071 nps 894000 tbhits 0 hashfull 433 pv e2e4 e7e5 g1f3 b8c6 f1b5 d8a5 b1c3 g8f6 e1g1 f6e4 d1e2 e4c3 d2c3 f7f6 f1d1 d7d6 b5c4 a5c7 c1e3 c8f5 c4d5 f8e7 h2h3 a8d8
bestmove e2e4 ponder e7e5

------------------------

Without d7 pawn:

info multipv 1 depth 20 seldepth 44 score cp 95 time 31554 nodes 28464690 nps 902000 tbhits 0 hashfull 218 pv g1f3 g8f6 b1c3 c7c5 e2e4 b8c6 f1b5 c8d7 d2d3 d8c7 e1g1 e7e6 d1d2 h7h6 b5c6 d7c6 d2f4 c7f4 c1f4 f6d7 f3e5 d7e5 f4e5 f7f6
bestmove g1f3 ponder g8f6

------------------------

Without e7 pawn:

info multipv 1 depth 20 seldepth 45 score cp 95 time 68251 nodes 63871954 nps 935000 tbhits 0 hashfull 470 pv e2e4 d7d5 e4d5 g8f6 f1b5 b8d7 g1f3 a7a6 b5d7 c8d7 e1g1 f6d5 f1e1 f8e7 c2c4 d5b6 d1e2 d7e6 d2d3 e8g8 b1c3 f8e8 c1e3 h7h6 e3d4 c7c5 d4e5 b6d7 c3d5 e6d5 c4d5
bestmove e2e4 ponder d7d5

------------------------

Without f7 pawn:

info multipv 1 depth 20 seldepth 43 score cp 160 time 42907 nodes 37694459 nps 878000 tbhits 0 hashfull 342 pv e2e4 e7e6 d2d4 b8c6 b1c3 f8b4 a2a3 b4c3 b2c3 d7d5 e4e5 g8e7 g1f3 e8g8 f1d3 h7h6 e1g1 b7b6 a3a4 c8b7 c1a3 d8d7 d1d2 a8d8
bestmove e2e4 ponder e7e6

------------------------

Without g7 pawn:

info multipv 1 depth 20 seldepth 47 score cp 109 time 76894 nodes 64213035 nps 835000 tbhits 0 hashfull 569 pv d2d4 g8f6 g1f3 c7c5 d4c5 d8a5 b1c3 b8c6 a2a3 f8g7 c1d2 a5c5 e2e3 c5b6 f1d3 d7d5 e1g1 h8g8 d1e2 e8f8 b2b4 c8d7 h2h3 a8c8
bestmove d2d4 ponder g8f6

------------------------

Without h7 pawn:

info multipv 1 depth 20 seldepth 43 score cp 76 time 50932 nodes 44333347 nps 870000 tbhits 0 hashfull 317 pv e2e4 c7c5 g1f3 b8c6 b1c3 g8f6 f1b5 e7e6 d2d3 c6d4 e4e5 a7a6 b5c4 b7b5 c4b3 b5b4 c3a4 d4b3 a2b3 f6g4 d3d4 d8c7
bestmove e2e4 ponder c7c5

It is curious to have the same eval (+0.95) in four cases: b7, c7, d7 or e7 pawns missing.

There is no doubt that Nakamura will have a very hard time in this match!

Regards from Spain.

Ajedrecista.

[Adam Hair]

I don't consider the material odd games as a real game of chess....

It dosen't indicate anything usefull related to the opponents as it's simply an odd variation of the classic game of chess....

A chess game with time odds would be more reliable....

Anyway,thanks for organizing this exciting match....
Dr.D

I would prefer to see a time odds match also.

[Robert Flesher]

I don't consider the material odd games as a real game of chess....
It dosen't indicate anything usefull related to the opponents as it's simply an odd variation of the classic game of chess....

A chess game with time odds would be more reliable....

Anyway,thanks for organizing this exciting match....
Dr.D

+1

I'd rather see a normal game and see him get crushed. Removal of a pawn renders the game a non theoretical encounter.

[bhlangonijr]

I would prefer to see a time odds match also.

+1

[Jhoravi]

I would prefer to see a time odds match also.

I don't like that idea because it limits the analytical strength of the main engine. The pawn handicap is ideal because despite the odds, its analytical power is at maximum and it's very interesting how it uses those resources to get around the material deficit.

[Vinvin]

Very nice events !
Thanks to organize it !

Will it be a video broadcast on Internet (even with a slow webcam) ?

[Vinvin]

And yep, I used 20cp contempt against Naroditsky and I plan to do the same against Nakamura.

With 1 pawn handicap you have to set it way higher (draw value around -100 cp). Otherwise, SF will be pleased to repeat moves in the opening.

[reflectionofpower]

I always like to see proof that these top engines are really playing stronger than humans.

Well all you need to do for that proof is watch any human chess tournament. Even when all the players are super grandmasters there's still a bunch of blunders being made every game. There are times you'd see them make game ending blunders and then the opponent responds with their own game ending blunder.

The programs do not get tired, space out or nervous. They see just about everything ALL the time. Humans do not have this ability so if a human is playing a strong chess program with fast hardware he is going to eventually make a mistake during a game and when he does the program will pounce on it immediately.