The Superfinal

[shrapnel]

but still those last 6 games somehow reminded us again of the influence of books.

When you have a score more than half a pawn, say 60-70cps right after the opening, this means the game has already finished.

My thoughts exactly !
Even in the game which Stockfish won, which Ansari posted, I was surprised to see sub-optimal Book moves in the first 8 moves itself, compared to strong Books like HIARCS 14i or l33t.
I know from my own experience in online games that coming out of the Book with such a great advantage practically guarantees a win for the Engine with the big advantage, so I agree with you completely.
Don't know what Book Thoresen is using, but its certainly not in the same class as his Hardware !!

[Lyudmil Tsvetkov]

coming out of the Book with such a great advantage practically guarantees a win for the Engine with the big advantage

Perfect.

That is the reasoning of experience.
You might have suboptimal books in lower-rated tourneys, but not when SF and Komodo are playing on 16 cores and 2 hours per move. Even though the current TCEC book is much improved to older books used, there is still a lot to wish for if you want perfect playing conditions.

[Dr.Wael Deeb]

but still those last 6 games somehow reminded us again of the influence of books.

When you have a score more than half a pawn, say 60-70cps right after the opening, this means the game has already finished.

My thoughts exactly !
Even in the game which Stockfish won, which Ansari posted, I was surprised to see sub-optimal Book moves in the first 8 moves itself, compared to strong Books like HIARCS 14i or l33t.
I know from my own experience in online games that coming out of the Book with such a great advantage practically guarantees a win for the Engine with the big advantage, so I agree with you completely.
Don't know what Book Thoresen is using, but its certainly not in the same class as his Hardware !!

I still wonder how this opening lines issue isn't yet resolved especially when we are talking about the final stage of the TCEC

In less than an hour I can provide the TCEC author with hundreds of optimzed opening linss from my private databases that would eliminate these unfair opening lnes we are witnessing right now....

It's that simple when you know what you are doing regards,
Dr.D

[Adam Hair]

You can read about the Season 6 book at TCEC. Look under the 'Help' tab.

The opening PGN being used was extracted from Norm Pollock's PGN of GM games and truncated to 8 moves originally by me. It has been modified more than once with Stockfish, Komodo, and Houdini at lower depths to remove obvious bad openings. For TCEC, Nelson Hernadez and myself spent two months collecting depth 24 (24 plies or 12 moves deep) evaluations from Houdini 4 and Komodo TCEC, and depth 28 evaluations from Stockfish DD for all 44,264 positions. Nelson uses these evaluations and the statistics from his huge database of games to select openings that match his selection criteria. Here is a description of his criteria:

Code: Select all

Frequency (or commonness) - how often has the candidate position been seen in human and engine games?

Play-balance - how closely does the candidate position align to the relative advantage seen in the opening position in empirical and evaluative terms?

Draw-tendency - how often has the candidate position resulted in a draw in human and engine databases relative to each of those databases?

Dissonance - to what degree do leading engines disagree in deep evaluations of the position?

My opening selection process was based on quantifying these four dimensions and converting the resulting analysis into percentile-ranks. In other words, every position was boiled down to four simple numbers on a 0-100 scale. I then assigned varying weights to each of these numbers for each Stage.

In all Stages no position was selected that had not been seen at least 100 times in human and engine games; in Stage 1 this number is at least 300 times. I did not choose very rare openings because the metrics for play balance and draw-tendency would have been unreliable.

As already mentioned play-balance is critically important in Stages 1 and 2 because they are both single round-robin. Every game in those Stages will slightly favor white as is seen in the opening position. Play-balance remains quite important through the other Stages, but not as much, because engines will get to play both sides of each position.

Attempting to reduce the draw-rate is not important in Stage 1 because engines of widely varying strength will be competing resulting in many decisive games. From Stage 2 onward all openings selected have below-average draw-rates. Our theory, which was demonstrated in the Superfinal last Season, is that you can knock down draw-rates by as much as 15-20% by selecting balanced openings that have historically produced low draw-rates.

Dissonance is irrelevant to my opening selection until Stage 4 and the Superfinal. In those Stages I purposely picked positions where the leading engines displayed sharp evaluation disagreements while not being heavily unbalanced in favor of one color. I am hoping that this will both further depress the draw-rate as well as lead to some really wild and exciting games. (Or it could lead to some big names getting steamrolled because they flat-out do not understand the starting position, in which case it will be very revealing!)

To perfect the openings, we would need evaluations that are as deep as what the engines reach under TCEC conditions. Unfortunately, we only had 8 year old servers to devote to this project.

[Adam Hair]

but still those last 6 games somehow reminded us again of the influence of books.

When you have a score more than half a pawn, say 60-70cps right after the opening, this means the game has already finished.

My thoughts exactly !
Even in the game which Stockfish won, which Ansari posted, I was surprised to see sub-optimal Book moves in the first 8 moves itself, compared to strong Books like HIARCS 14i or l33t.
I know from my own experience in online games that coming out of the Book with such a great advantage practically guarantees a win for the Engine with the big advantage, so I agree with you completely.
Don't know what Book Thoresen is using, but its certainly not in the same class as his Hardware !!

I still wonder how this opening lines issue isn't yet resolved especially when we are talking about the final stage of the TCEC

In less than an hour I can provide the TCEC author with hundreds of optimzed opening linss from my private databases that would eliminate these unfair opening lnes we are witnessing right now....

It's that simple when you know what you are doing regards,
Dr.D

Hi Wael,

I believe that it is broadly acknowledged, especially at the Rybka forum, that Nelson Hernadez knows what he is doing.

[Lyudmil Tsvetkov]

You can read about the Season 6 book at TCEC. Look under the 'Help' tab.

The opening PGN being used was extracted from Norm Pollock's PGN of GM games and truncated to 8 moves originally by me. It has been modified more than once with Stockfish, Komodo, and Houdini at lower depths to remove obvious bad openings. For TCEC, Nelson Hernadez and myself spent two months collecting depth 24 (24 plies or 12 moves deep) evaluations from Houdini 4 and Komodo TCEC, and depth 28 evaluations from Stockfish DD for all 44,264 positions. Nelson uses these evaluations and the statistics from his huge database of games to select openings that match his selection criteria. Here is a description of his criteria:

Code: Select all

Frequency (or commonness) - how often has the candidate position been seen in human and engine games?

Play-balance - how closely does the candidate position align to the relative advantage seen in the opening position in empirical and evaluative terms?

Draw-tendency - how often has the candidate position resulted in a draw in human and engine databases relative to each of those databases?

Dissonance - to what degree do leading engines disagree in deep evaluations of the position?

My opening selection process was based on quantifying these four dimensions and converting the resulting analysis into percentile-ranks. In other words, every position was boiled down to four simple numbers on a 0-100 scale. I then assigned varying weights to each of these numbers for each Stage.

In all Stages no position was selected that had not been seen at least 100 times in human and engine games; in Stage 1 this number is at least 300 times. I did not choose very rare openings because the metrics for play balance and draw-tendency would have been unreliable.

As already mentioned play-balance is critically important in Stages 1 and 2 because they are both single round-robin. Every game in those Stages will slightly favor white as is seen in the opening position. Play-balance remains quite important through the other Stages, but not as much, because engines will get to play both sides of each position.

Attempting to reduce the draw-rate is not important in Stage 1 because engines of widely varying strength will be competing resulting in many decisive games. From Stage 2 onward all openings selected have below-average draw-rates. Our theory, which was demonstrated in the Superfinal last Season, is that you can knock down draw-rates by as much as 15-20% by selecting balanced openings that have historically produced low draw-rates.

Dissonance is irrelevant to my opening selection until Stage 4 and the Superfinal. In those Stages I purposely picked positions where the leading engines displayed sharp evaluation disagreements while not being heavily unbalanced in favor of one color. I am hoping that this will both further depress the draw-rate as well as lead to some really wild and exciting games. (Or it could lead to some big names getting steamrolled because they flat-out do not understand the starting position, in which case it will be very revealing!)

To perfect the openings, we would need evaluations that are as deep as what the engines reach under TCEC conditions. Unfortunately, we only had 8 year old servers to devote to this project.

Hi Adam, you have done a fantastic job, congratulations!
The current book is really much much better than older ones.

4 plies more at real TCEC really matters, I do not know what to say, maybe minimise the maximum score threshold to some 30cps at most? However, even that way, we learn a lot about certain positions that we otherwise would not have known.

[Nelson Hernandez]

I am the responsible party for the TCEC openings. What I would suggest is that you first read my notes on the openings in TCEC Help. That will provide insight into how I approached the problem of selecting hundreds of suitable openings given the parameters and priorities decided by Martin Thoresen.

Then you would do well to consider the limited resources we had in terms of time (two months) and CPU power (two old servers). Our job was not to over-promise and under-deliver but rather the reverse. In other words, given that we needed to complete the task in such a short period of time, we had to size the scope of the job so that we would deliver the best product we could in the time allotted. I think we succeeded.

Was every one of the hundreds of openings a success, in that it produced a fair and entertaining contest? I would say "no" immediately, while stating that a high percentage of TCEC attendees agree that the openings as a whole have been a success, even while they personally may have different ideas and approaches.

It has proved a particularly daunting task to coax decisive outcomes out of the Superfinal without employing biased starting positions given the strength of the finalists. Were we to employ optimal openings exclusively the draw-rate might well climb to 80-90% and if it did I would be criticized on those grounds. (What you may not realize is that I was hired for this position in the first place in order to bring down the draw-rate in a controlled and sensible manner.)

In any event there have been lessons learned from this Season. There are areas of possible improvement, and I hope to incorporate them next time around. But I am under no illusions that I will be able to please everyone!

[arjuntemurnikar]

I am surprised this game was not mentioned here.

It is a drawn game, but nonetheless one of the most interesting draws I have ever seen, so it deserves to be highlighted.

[pgn]


[Event "TCEC Season 6 - Superfinal"]
[Site "http://tcec.chessdom.com"]
[Date "2014.05.22"]
[Round "20"]
[White "Komodo 7x"]
[Black "Stockfish 170514"]
[Result "1/2-1/2"]
[BlackElo "3164"]
[ECO "D10"]
[EventDate "2014.05.19"]
[GameDuration "03:01:18"]
[Opening "Slav: Winawer Countergambit, 4.cxd5 cxd5 5.dxe5"]
[PlyCount "78"]
[TerminationDetails "3-fold repetition"]
[TimeControl "7200+30"]
[WhiteElo "3155"]

{ WhiteEngineOptions: Protocol=UCI; Hash=16384; Table Memory=256; Threads=16; SyzygyPath=C:/EGTB/Syzygy/; Drawscore=0; OwnBook=false; Ponder=false;, BlackEngineOptions: Protocol=UCI; Hash=16384; Threads=16; SyzygyPath=C:/EGTB/Syzygy/; Emergency Base Time=1000; OwnBook=false; Ponder=false; }
1. d4 { book } d5 { book } 2. c4 { book } c6 { book } 3. Nc3 { book }
e5 { book } 4. cxd5 { book } cxd5 { book } 5. dxe5 { book } d4 { book }
6. Ne4 { book } Qa5+ { book } 7. Nd2 { book } Nc6 { book } 8. Nf3 { book }
Bg4 { book }
9. a3 { d=27, pd=Rc8, mt=00:03:02, tl=02:01:27, s=14644 kN/s, n=2678690081, pv=a3 Rc8 h3 Bf5 Qb3 Be6 Qxb7 Rc7 Qa8+ Rc8 Qb7, tb=3, R50=47, wv=0.00, }
Rc8 { d=34, pd=h3, mt=00:03:55, tl=02:00:34, s=17981 kN/s, n=4233869497, pv=Rc8 h3 Bf5 Qb3 Be6 Qxb7 Rc7 Qa8+ Rc8 Qb7, tb=12, R50=50, wv=0.00, }
10. h3 { d=29, pd=Bf5, mt=00:08:02, tl=01:53:54, s=18270 kN/s, n=8808345142, pv=h3 Bf5 Qb3 Be6 Qxb7 Rc7 Qa8+ Rc8 Qb7, tb=44, R50=50, wv=0.00, }
Bf5 { d=39, pd=Qb3, mt=00:02:06, tl=01:58:57, s=18699 kN/s, n=2368495546, pv=Bf5 Qb3 Be6 Qxb7 Rc7 Qa8+ Rc8 Qb7, tb=4, R50=50, wv=0.00, }
11. g4 { d=29, pd=Nb4, mt=00:04:07, tl=01:50:17, s=19123 kN/s, n=4736620478, pv=g4 Nb4 axb4 Qxa1 Nb3 Bxb4+ Nfd2 Bxd2+ Kxd2 Qa4 gxf5 Qb4+ Kd3 Qc4+ Kd2 Qb4+, tb=1916, R50=50, wv=0.00, }
Nb4 { d=40, pd=axb4, mt=00:04:00, tl=01:55:27, s=19665 kN/s, n=4722402198, pv=Nb4 axb4 Qxa1 Nb3 Bxb4+ Nfd2 Qxc1 Qxc1 Rxc1+ Nxc1 Bc2 Bg2 Ne7 Bxb7 h5 Na2 Ba5 b4 Bc7 O-O d3 exd3 Bxd3 Re1 hxg4 hxg4 Rh4 f3 Ng6 Ne4 Bxe4 Rxe4 Nxe5 Kg2 Kf8 Nc1 g5 Bd5 Ng6 Ne2 Ne7 Bc4 Bd6 Rd4 Rh6 b5, tb=6621, R50=50, wv=-0.10, }
12. axb4 { d=27, pd=Qxa1, mt=00:00:39, tl=01:50:07, s=20001 kN/s, n=786311751, pv=axb4 Qxa1 Nb3 Bxb4+ Nfd2 Bxd2+ Kxd2 Qa4 gxf5 Qb4+ Kd3 Ne7 Ke4 d3+ Kf3 dxe2 Bxe2 Nxf5 Kg2 Qe4+ Bf3 Nh4+ Kg3 Nf5+ Kg2, tb=359, R50=50, wv=0.00, }
Qxa1 { d=41, pd=Nb3, mt=00:02:03, tl=01:53:54, s=22582 kN/s, n=2794409202, pv=Qxa1 Nb3 Bxb4+ Nfd2 Qxc1 Qxc1 Rxc1+ Nxc1 Bc2 Bg2 Ne7 Bxb7 h5 Na2 Ba5 b4 Bc7 O-O d3 exd3 Bxd3 Re1 hxg4 hxg4 Rh4 f3 Ng6 Ne4 Bxe4 Rxe4 Nxe5 Kg2 Kf8 Nc1 g5 Bd5 Ng6 Ne2 Ne7 Bc4 Bd6 Rd4 Rh6 b5, tb=30472, R50=50, wv=-0.10, }
13. Nb3 { d=26, mt=00:00:59, tl=01:49:38, s=20401 kN/s, n=1209747227, pv=Nb3, tb=749, R50=50, wv=-0.15, }
Bxb4+ { d=40, pd=Nfd2, mt=00:02:04, tl=01:52:19, s=21602 kN/s, n=2692605297, pv=Bxb4+ Nfd2 Qxc1, tb=6496, R50=50, wv=-0.10, }
14. Nfd2 { d=26, pd=Bxd2+, mt=00:01:14, tl=01:48:54, s=20622 kN/s, n=1538770765, pv=Nfd2 Bxd2+ Kxd2 Qa4 gxf5 Qb4+ Kd3 Ne7 Ke4 O-O Kf3 Qa4 Bg2 Nxf5 Qd3 f6 e6 g6 Kg4 h5+ Kf4 Qb4 Kf3 Qb6 e4 dxe3 Bxe3 Qxe6 Nd4 Qd5+ Qe4 Nxe3 fxe3 Qxe4+ Kxe4 Rf7 Kf4 Rd7 Be4 f5 Bf3 Kf7 Rg1 Re8 h4 a6, tb=1881, R50=50, wv=-0.16, }
Bxd2+ { d=38, pd=Kxd2, mt=00:04:56, tl=01:47:53, s=20791 kN/s, n=6155702451, pv=Bxd2+ Kxd2 Qa4 gxf5 Qb4+ Kd3 Ne7 Ke4 O-O Rg1 f6 Nxd4 fxe5 Qb3+ Qxb3 Nxb3 Rc4+ Kd3 Rb4 Bg5 Nxf5 Kc3 Rb6 Bg2 Rc8+ Kd2 Kf8 Be4 Nd6 Bd5 Nc4+ Bxc4 Rxc4 Kd3 Rc8 Nd2 Rxb2 Be3 Rd8+ Kc3 Ra2 Rg5 Rd5 f4 b5 Rxe5 b4+ Kxb4 Rdxd2 Bxd2, tb=10377, R50=50, wv=-0.13, }
15. Kxd2 { d=24, pd=Qa4, mt=00:01:44, tl=01:47:39, s=21424 kN/s, n=2237580653, pv=Kxd2 Qa4 gxf5 Qb4+ Kd3 f6 Rg1 Ne7 exf6 Qb5+ Kd2 Qb4+ Kd3, tb=1157, R50=50, wv=0.00, }
Qa4 { d=40, pd=gxf5, mt=00:02:03, tl=01:46:19, s=21581 kN/s, n=2673564596, pv=Qa4 gxf5 Qb4+ Kd3 Ne7 Ke4 O-O Rg1 f6 Nxd4 fxe5 Qb3+ Qxb3 Nxb3 Rc4+ Kd3 Rb4 Bg5 Nxf5 Kc3 Rb6 Bg2 Rc8+ Kd2 Kf8 Be4 Nd6 Bd5 Nc4+ Bxc4 Rxc4 Kd3 Rc8 Nd2 Rxb2 Be3 Rd8+ Kc3 Ra2 Rg5 Rd5 f4 b6 Rf5+ Kg8 fxe5 g6 Rg5 Kf7, tb=8295, R50=50, wv=-0.20, }
16. gxf5 { d=25, pd=Qb4+, mt=00:00:37, tl=01:47:31, s=20283 kN/s, n=770494298, pv=gxf5 Qb4+ Kd3 f6 Rg1 Ne7 Rxg7 Nxf5 exf6 Qc4+ Kd2 Qb4+ Kd3, tb=1393, R50=50, wv=0.00, }
Qb4+ { d=38, pd=Kd3, mt=00:02:04, tl=01:44:44, s=21755 kN/s, n=2716433056, pv=Qb4+ Kd3 Ne7 Ke4 O-O Rg1 f6 Nxd4, tb=6981, R50=50, wv=-0.20, }
17. Kd3 { d=27, pd=Ne7, mt=00:00:42, tl=01:47:18, s=20684 kN/s, n=886720926, pv=Kd3 Ne7 f6 gxf6 exf6 Ng6 Bh6 Qb5+ Kd2 Qb4+ Kd3, tb=1895, R50=50, wv=0.00, }
Ne7 { d=37, pd=Ke4, mt=00:02:05, tl=01:43:08, s=21843 kN/s, n=2749569462, pv=Ne7 Ke4, tb=5297, R50=49, wv=-0.20, }
18. f6 { d=27, pd=gxf6, mt=00:01:58, tl=01:45:50, s=20727 kN/s, n=2453955413, pv=f6 gxf6 exf6 Nf5 Rg1 Nh4 Ke4 d3+ Nd4 Kd8 Be3 Re8+ Kxd3 Ng6 Rg5 Qc4+ Kd2 Qb4+ Kd3, tb=10953, R50=49, wv=0.00, }
Ng6 { d=35, pd=fxg7, mt=00:03:19, tl=01:40:19, s=20655 kN/s, n=4111928822, pv=Ng6 fxg7 Rg8 f4 Rxg7 Nxd4 Qc4+ Ke3 f6 exf6 Rd7 Bg2 Nxf4 Bxb7 Nd5+ Bxd5 Qxd5 Kf2 Qxd4+ Qxd4 Rxd4 Be3 Rd6 Bxa7 Rxf6+ Ke3 Rc2 b4 Re6+ Kd3 Rexe2 Be3 Ra2 h4 Rg2 Re1 Ra3+ Kc4 Rg4+ Kd5 Kf7 Bc5 Rd3+ Kc6 Rxh4 Re7+ Kg6 b5 Rc3 Kd6, tb=48729, R50=50, wv=-0.41, }
19. fxg7 { d=25, pd=Qb5+, mt=00:01:15, tl=01:45:04, s=21123 kN/s, n=1592274771, pv=fxg7 Qb5+ Kd2 Qb4+ Kd3, tb=6002, R50=50, wv=0.00, }
Rg8 { d=38, pd=f4, mt=00:03:14, tl=01:37:35, s=21645 kN/s, n=4210398467, pv=Rg8 f4 Rxg7 Nxd4 Qc4+ Ke3 f6 exf6 Rd7 Bg2 Nxf4 Bxb7 Nd5+ Bxd5 Qxd5 Kf2 Qxd4+ Qxd4 Rxd4 Be3 Rd6 Bxa7 Rxf6+ Kg3 Rc4 Ra1 Rg6+ Kf2 Rf4+ Ke3 Rh4 Kf2 Rg7 Ra3 Rb7 Kg3 Rhb4 Ra6 Rxb2 Bd4 R2b3+ e3 Rf7 Kg4 Re7 Kf3 Rf7+, tb=326854, R50=50, wv=-0.49, }
20. f4 { d=27, pd=Rxg7, mt=00:04:20, tl=01:41:14, s=21676 kN/s, n=5655325491, pv=f4 Rxg7 Nxd4 Qxd4+ Kxd4 Rd8+ Ke4 Rxd1 Be3 Ra1 Bd4 Ne7 Bg2 Ra4 Bf3 Rg3 e3 Rb4 Bg4 Nc6 Rh2 Rg1 Kd5 Rc1 Rd2 Nxd4 exd4 Ke7 Bf5 h6 Bd3 Rc7 Ke4 Rd7 Bf1 a6 f5 Rd8, tb=88153, R50=50, wv=-0.31, }
Rxg7 { d=40, pd=Nxd4, mt=00:04:27, tl=01:33:37, s=22743 kN/s, n=6089149564, pv=Rxg7 Nxd4 Qxd4+ Kxd4 Rd8+ Kc3 Rxd1 Bd2 Ne7 Kc2 Ra1 Bg2 Rxh1 Bxh1 b6 Bf3 Rg3 Bg4 h5 Bxh5 Rxh3 Bg4 Rg3 Bf3 Nf5 Bc3 Ke7 Kd2 Ke6 b4 Ne7 Be4 Rg4 e3 Rg3 Kd3 Rh3 b5 Nf5 Bd2 Rg3 Ke2, tb=856194, R50=50, wv=-0.52, }
21. Nxd4 { d=29, pd=Qxd4+, mt=00:04:31, tl=01:37:12, s=22006 kN/s, n=5969100000, pv=Nxd4 Qxd4+ Kxd4 Rd8+ Ke4 Rxd1 Be3 Ra1 Bd4 Ne7 Bg2 Ra4 Bf3 Rg3 e3 Rb4 Bg4 Nc6 Rh2 Rg1 Rc2 Re1 Be2 Rh1 Bc4 Nxd4 exd4 Kf8 Ba2 Ra1 Bc4 Rb1 Bd3 R1xb2 Rc7 Ke8 f5 Rh2 f6 Rxh3 Re7+ Kf8, tb=289590, R50=50, wv=-0.31, }
Qxd4+ { d=41, pd=Kxd4, mt=00:03:28, tl=01:30:38, s=22693 kN/s, n=4731510830, pv=Qxd4+ Kxd4 Rd8+ Ke4 Rxd1 Be3 Ra1 Rg1 Kf8 Bc5+ Kg8 Rh1 b6 Ba3 a5 e3 Re1 h4 Nf8 Kf3 h5 Kf2 Rc1 Rh3 Ne6 Bd3 Nc5 Bf5 Rd1 Ke2 Rgg1 Bxc5 bxc5 Rh2 Rh1 Rg2+ Kf8 Rg5 Rd8 Rg2 c4 Kf3 Rb8 Rc2 Rxh4, tb=689633, R50=50, wv=-0.60, }
22. Kxd4 { d=28, pd=Rd8+, mt=00:00:39, tl=01:37:03, s=22400 kN/s, n=881323640, pv=Kxd4 Rd8+ Ke4 Rxd1 Be3 Ra1 Bd4 Ne7 Bg2 Ra4 Bf3 Rg3 e3 Rb4 Bh5 Rxe3+ Kxe3 Nf5+ Ke2 Nxd4+ Ke3 Ne6 f5 Nd4 Rf1 Ke7 Bg4 f6 Rf2 Nc6 e6 Rb3+ Kf4 h5 Bxh5 Rxh3 Bg4 Rb3 Rh2 Rb4+ Kg3 Ne5, tb=39194, R50=50, wv=-0.20, }
Rd8+ { d=41, pd=Ke4, mt=00:04:24, tl=01:26:44, s=23509 kN/s, n=6215026815, pv=Rd8+ Ke4 Rxd1 Be3 Ra1 b3 Ne7 Bg2 Ra3 Bf3 Rxb3 Bxa7 b6 Bb8 Rb4+ Ke3 Nf5+ Kf2 Rxf4 e6 Rb4 Be5 fxe6 Bc3 Rb3 Bxg7 Nxg7 Rc1 Kd7 Ra1 Ke7 Ra7+ Kf6 e4 Rb2+ Ke3 b5 h4 h6 Bg4 Rb3+ Kd2 Rg3 Be2 Ke5 Bd3 Ne8 Rh7 Nd6 Rxh6 Nc4+, tb=1940820, R50=50, wv=-0.40, }
23. Ke4 { d=30, pd=Rxd1, mt=00:02:59, tl=01:34:33, s=21377 kN/s, n=3843518180, pv=Ke4 Rxd1 Be3 Ra1 Bd4 Ne7 Bg2 Ra4 Bf3 Rg3 e3 Rb4 Bh5 Kf8 Kd3 b6 Bc3 Ra4 Bd1 Ra1 Ke4 Rc1 h4 a5 h5 b5 Bf3 Rxh1 Bxh1 b4 Bd4 a4 Kd3 Rh3 Be4 Rxh5, tb=526340, R50=50, wv=-0.36, }
Rxd1 { d=42, pd=Be3, mt=00:02:02, tl=01:25:11, s=23328 kN/s, n=2866651094, pv=Rxd1 Be3 Ra1 b3 Ne7 Bg2 Ra3 Bf3 Rxb3 Bxa7, tb=1118811, R50=49, wv=-0.53, }
24. Be3 { d=29, pd=Ra1, mt=00:00:51, tl=01:34:12, s=21995 kN/s, n=1122014655, pv=Be3 Ra1 Bd4 Ne7 Bg2 Ra4 Bf3 Rg3 e3 Rb4 Bh5 Kf8 Kd3 b6 Bc3 Ra4 Bd1 Ra1 Ke4 Rc1 h4 a5 Rf1 Rh3 Bd4 Nc6 Bxb6 Rh2 Be2 Rxf1 Bxf1 Rxb2 Bc5+ Kg7 h5 a4 Ba3, tb=119179, R50=50, wv=-0.31, }
Ra1 { d=40, pd=b3, mt=00:03:38, tl=01:22:03, s=21384 kN/s, n=4669007999, pv=Ra1 b3 Ne7 Bg2 Ra3 Bf3 Rxb3 Bxa7 b6 Bb8 Rb4+ Ke3 Nf5+ Kf2 Rxf4 e6 Rxf3+ Kxf3 fxe6 Be5 Rg5 Ra1 Ne7 Ra8+ Kf7 Bd4 Rf5+ Ke4 b5 Ra7 Ke8 Rb7 Rh5 Bf2 Nd5 h4 Nf6+ Kd3 Rd5+ Kc2 h5 Be1 Rf5 Kd3 Nd7, tb=969159, R50=50, wv=-0.40, }
25. Bd4 { d=30, pd=Ne7, mt=00:04:21, tl=01:30:21, s=21479 kN/s, n=5614492081, pv=Bd4 Ne7 Bg2 Ra4 Bf3 Rg3 e3 Rb4 Rc1 Nc6 Bg4 h5 Bxh5 Rxh3 Be2 Rh2 Rc2 Ke7 Rd2 a5 Rc2 Kf8 Rd2 Kg7 Rc2 a4 Rd2 b6 Rc2 Nxd4 exd4 b5 Rd2 Rb3 f5 b4 f6+ Kg6 Kd5 Kg5 Kc5, tb=1054158, R50=49, wv=-0.39, }
Ne7 { d=38, pd=Bg2, mt=00:03:33, tl=01:19:00, s=21904 kN/s, n=4665698122, pv=Ne7 Bg2 Ra4 Bf3 Rg3 e3 Rb4 Bh5 Rxe3+ Kxe3 Nf5+ Kf2 Nxd4 Ke3 Nf5+ Kf3 Rxb2 Rc1 Ne7 Kg4 Rb5 Kg5 a5 Kf6 Nd5+ Kg7 Nxf4 Bxf7+ Kd7 Kf6 Rb6+ Kf5 Ne6 Rd1+ Ke7 Bg8 Ng7+ Ke4 a4, tb=479289, R50=49, wv=-0.55, }
26. Bg2 { d=29, pd=Ra4, mt=00:03:20, tl=01:27:30, s=21101 kN/s, n=4230362529, pv=Bg2 Ra4, tb=1109890, R50=48, wv=-0.28, }
Ra4 { d=40, pd=Bf3, mt=00:02:51, tl=01:16:38, s=22544 kN/s, n=3868681987, pv=Ra4 Bf3 Rg3 e3 Rb4 Bh5 Rxe3+ Kxe3 Nf5+ Kf3 Nxd4+ Ke3 Nf5+ Kf3 Rxb2 Rc1 Nd4+ Ke3 Nc6 e6 Rb3+ Ke4 Kf8 exf7 a5 Rg1 Rb4+ Ke3, tb=1239587, R50=48, wv=-0.63, }
27. Bf3 { d=28, pd=Rg3, mt=00:02:22, tl=01:25:38, s=22015 kN/s, n=3137745123, pv=Bf3 Rg3 Bh5 Kf8 e3 Nc6 f5 Rb4 Rh2 a5 Bg4 Nxd4 exd4 f6 exf6 Rgb3 Rc2 Kf7 Rc7+ Kxf6 Rxh7 Rxb2 Rh6+ Kg7 Rg6+ Kh7 Bf3 Rd2 Ke5 Rdxd4 Bxb7 a4 Bd5 Rd3 Bg8+ Kh8 Bd5 Rb5 Rd6 Re3+ Kf4 Rxh3 f6, tb=869834, R50=47, wv=-0.35, }
Rg3 { d=40, pd=Bh5, mt=00:05:18, tl=01:11:50, s=23776 kN/s, n=7574960859, pv=Rg3 Bh5 Nc6 e3 Kf8 Bg4 Rb4 f5 Rg2 Kf3 Rc2 Bc3 Rxc3 bxc3 Nxe5+ Kg3 Rb3 f6 Rxc3 Bf5 Rc6 Be4 Rxf6 Bxb7 Rg6+ Kf2 Rh6 Kg3 a5 Ra1 Nc4 e4 Rb6, tb=3118762, R50=47, wv=-0.72, }
28. Bh5 { d=28, pd=Kf8, mt=00:06:04, tl=01:20:03, s=22110 kN/s, n=8060995835, pv=Bh5 Kf8, tb=2136581, R50=46, wv=-0.14, }
Nc6 { d=39, pd=e3, mt=00:05:24, tl=01:06:55, s=24768 kN/s, n=8039866278, pv=Nc6 e3 Kf8 Bg4 Rb4 f5 f6 exf6 Nxd4 exd4 Rxb2 Rc1 Kf7 Kf4 Rd3 Ke5 Rb5+ Ke4 Rg3 Rc7+ Kxf6 Rxh7 Rg1 Rh6+ Kg7 Rg6+ Kf7 Rh6 Re1+ Kd3 Rf1 Kc4 a6 Rh7+ Kf8 Rh6 Rc1+ Kd3 Rb3+ Ke4 Re1+ Kd5 Rb5+ Kc4 Kg7 Rg6+ Kf7 Rh6 Rc1+, tb=2673660, R50=46, wv=-0.55, }
29. e3 { d=29, pd=Rc4, mt=00:00:54, tl=01:19:39, s=22311 kN/s, n=1210537141, pv=e3 Rc4 e6 Nxd4 exd4 Ke7 exf7 Rc6 Kd5 Re3 Rg1 Rh6 Rg5 Rd6+ Kc4 Rc6+ Kb4 Rb6+ Kc5 Rc6+ Kd5 Rd6+, tb=946999, R50=45, wv=0.00, }
Kf8 { d=39, pd=Bg4, mt=00:05:26, tl=01:01:58, s=24326 kN/s, n=7950005146, pv=Kf8 Bg4 Rb4 f5 f6 exf6 Nxd4 exd4 Rxb2 Rc1 Kf7 Rc7+ Kxf6 Rxh7 Rg1 Rh6+ Kg7 Rg6+ Kf7 Rh6 Re1+ Kd5 Rb6 f6 Rxf6 Rh7+ Kg6 Rxb7 Ra6 Rb8 Kf6 Rc8 Raa1 Rc6+ Kg5 Rc5 Kf4 Rc8 a5 Rf8+ Kg3 Rf3+ Kh4 Kc4 a4 d5, tb=3284660, R50=50, wv=-0.32, }
30. Bg4 { d=29, pd=Rb4, mt=00:02:06, tl=01:18:02, s=22089 kN/s, n=2798426473, pv=Bg4 Rb4 f5 Rg2 Rc1 Rgxb2 f6 Rd2 Rc5 a5 e6 fxe6 Bxe6 Rf2 Kd3 h6 Bc3 Rb1 Rh5 Rb6 Rxh6 Ne5+ Bxe5 Rxe6 Rh8+ Kf7 Rh7+ Kg8 Rg7+ Kf8 Bd4 b6 Ra7 Rf3 Ra8+ Re8 Ra7 Rf2 h4 Rd8 Ke4 Re8+ Kd5 Rf5+ Kd6 Rf3 Rg7, tb=1728484, R50=50, wv=0.00, }
Rb4 { d=39, pd=f5, mt=00:05:32, tl=00:56:56, s=24351 kN/s, n=8096493739, pv=Rb4 f5 f6 exf6 Nxd4 exd4 Rxb2 Rc1 Kf7 Kf4 Rgb3 Ke5 Re3+ Kf4 Rd3, tb=5041314, R50=49, wv=0.00, }
31. f5 { d=30, pd=Nxd4, mt=00:01:15, tl=01:17:17, s=22408 kN/s, n=1688771593, pv=f5 Nxd4 exd4 f6 exf6 Rxb2 Rc1 Kf7 Kf4 Rgb3 Ke5 Re3+ Kf4 Rd3 Kg5 h6+ Kxh6 Kxf6 Rc8 Rg2 Rf8+ Ke7 Rb8 Rb3 Kg5 Rb5 h4 b6 h5 Rd5 h6 Rxd4 f6+ Kf7 Rb7+ Kf8 Rb8+ Kf7, tb=2271203, R50=49, wv=0.00, }
Rg2 { d=41, pd=Rc1, mt=00:02:17, tl=00:55:08, s=24149 kN/s, n=3322418263, pv=Rg2 Rc1 Rgxb2 f6 Rd2, tb=2457956, R50=50, wv=-0.06, }
32. Rc1 { d=30, pd=Rgxb2, mt=00:01:22, tl=01:16:24, s=22371 kN/s, n=1851733149, pv=Rc1 Rgxb2 f6 Rd2 Rc5 a5 e6 fxe6 Bxe6 Rf2 Kd3 h6 Bc3 Rb1 Rh5 Rd1+ Ke4 Rd6 Bc4 Rfxf6 Bxf6 Rxf6 Rb5 Ke7 Rxb7+ Kd6 Bb3 Rf8 Rh7 Rf6 Rg7 Kc5 Rg1 Rd6 Rc1+ Kb4 Bd5 Ne7 Rb1+ Kc5 Rc1+ Kb4, tb=2323158, R50=50, wv=0.00, }
Rgxb2 { d=41, pd=f6, mt=00:02:19, tl=00:53:19, s=22320 kN/s, n=3104910847, pv=Rgxb2 f6 Rd2 Bf5 a5 Bxh7 Ke8 Bf5 Rg2 Kd5 Rg5 Bg4 Kd8 Kd6 Rxg4 hxg4 Nxd4 exd4 Rxd4+ Kc5 Rxg4 e6 fxe6 Kd6 Ke8 Kxe6 Re4+ Kf5 Re2 Rc7 Rf2+ Ke6 Re2+ Kf5, tb=2648659, R50=49, wv=0.00, }
33. f6 { d=29, pd=Rb1, mt=00:01:10, tl=01:15:44, s=22578 kN/s, n=1580801805, pv=f6 Rb1 Rc5 Rb5 Bd7 R1b4 Kd5 Rb1 Ke4, tb=1270870, R50=50, wv=0.00, }
Rb1 { d=44, pd=Rc5, mt=00:02:16, tl=00:51:33, s=23518 kN/s, n=3209527983, pv=Rb1 Rc5 Rb5 Bd7 R1b4 Kd5 a5 Kd6 Rxc5 Bxc5 Rb5 Bxc6 Rxc5 Kxc5 bxc6 e4 h5 e6 fxe6 e5 h4 Kd6 Kf7 Kc5 Ke8 Kc4 Kf7, tb=3046668, R50=50, wv=0.00, }
34. Rc5 { d=33, pd=Rb5, mt=00:03:55, tl=01:12:18, s=22723 kN/s, n=5358351885, pv=Rc5 Rb5 Bd7 R1b4 Kd5 Rb1 Kc4 a6 Kd5 a5 Kc4 Rb6 Kd5 Ra6 Rc4 Rb5+ Rc5 Rb3 e6 fxe6+ Bxe6 Nd8 Rc8 Rb5+ Kc4 Rb4+ Kd3 Rd6 Bg4 Rbxd4+ exd4 Kf7 Rc7+ Kxf6 Rxh7 Nc6 Rh6+ Ke7 Rh7+ Kd8 Rxb7 a4 d5 Rxd5+ Kc4 Na5+ Kxd5, tb=7140822, R50=50, wv=0.00, }
Rb5 { d=46, pd=Bd7, mt=00:02:45, tl=00:49:17, s=24131 kN/s, n=3990382782, pv=Rb5 Bd7 R1b4 Kd5 a5 Kd6 Rxc5 Bxc5 Rb5 Bxc6 Rxc5 Kxc5 bxc6 e4 h5 e6 fxe6 e5 h4 Kd6 Kf7 Kc5 Ke8 Kc4 Kf7 Kc5, tb=3941491, R50=49, wv=0.00, }
35. Bd7 { d=35, pd=R1b4, mt=00:02:45, tl=01:10:03, s=23079 kN/s, n=3809152003, pv=Bd7 R1b4 Kd5 Rb1 Kc4 R5b4+ Kd5 Rb5, tb=8684550, R50=49, wv=0.00, }
R1b4 { d=51, pd=Kd5, mt=00:01:39, tl=00:48:08, s=22316 kN/s, n=2211707110, pv=R1b4 Kd5 a5 Kd6 Rxc5 Bxc5 Rb5 Bxc6 Rxc5 Kxc5 bxc6 e4 h5 e6 fxe6 e5 h4 Kd6 Kf7 Kc5 Ke8 Kc4 Kf7 Kc5, tb=5556911, R50=48, wv=0.00, }
36. Kd5 { d=34, pd=a5, mt=00:01:12, tl=01:09:21, s=24282 kN/s, n=1761089440, pv=Kd5 a5 Kd6 Rxc5 Bxc5 Rb5 e6 fxe6 Bxe6 b6 Ba3 Nb4 e4 Rc5 Bxb4 axb4 e5 Rc2 Bd5 Rd2 e6 Rxd5+ Kxd5 b3 Kd6 b2 Kd7 b1=Q e7+ Kf7 e8=Q+ Kxf6 Qe6+ Kg7 Qe5+ Kh6 Qe3+ Kh5 Qe5+ Kh4 Qf4+ Kxh3, tb=8113687, R50=48, wv=0.00, }
Rb3 { d=51, pd=Ke4, mt=00:01:37, tl=00:47:01, s=23880 kN/s, n=2322824671, pv=Rb3 Ke4 R3b4, tb=3790708, R50=47, wv=0.00, }
37. Ke4 { d=37, pd=R3b4, mt=00:03:35, tl=01:06:15, s=24547 kN/s, n=5300341655, pv=Ke4 R3b4, tb=19213897, R50=47, wv=0.00, }
R3b4 { d=64, pd=Kd5, mt=00:02:13, tl=00:45:18, s=24668 kN/s, n=3291030183, pv=R3b4 Kd5, tb=7505889, R50=46, wv=0.00, }
38. Kd5 { d=41, mt=00:03:47, tl=01:02:57, s=23942 kN/s, n=5446445032, pv=Kd5, tb=13719956, R50=46, wv=0.00, }
Rb3 { d=58, pd=Ke4, mt=00:03:14, tl=00:42:33, s=25212 kN/s, n=4898404333, pv=Rb3 Ke4, tb=18952007, R50=45, wv=0.00, }
39. Ke4 { d=36, mt=00:06:03, tl=00:57:24, s=23590 kN/s, n=8580059067, pv=Ke4, tb=13312885, R50=45, wv=0.00, }
R3b4 { d=65, pd=Kd5, mt=00:02:45, tl=00:40:18, s=24185 kN/s, n=4003436528, pv=R3b4 Kd5, tb=12353528, R50=44, wv=0.00,, Draw by 3-fold repetition }
1/2-1/2


[/pgn]

The Slav-Winawer Countergambit, an interesting opening choice to being with.

[d]r3kbnr/pp3ppp/2n5/q3P3/3p2b1/5N2/PP1NPPPP/R1BQKB1R w KQkq - 6 9

Here Stockfish plays Nb4!? a very interesting tactic and a series of forcing moves follow...

[d]2r1kbnr/pp3ppp/8/q3Pb2/1n1p2P1/P4N1P/1P1NPP2/R1BQKB1R w KQk - 1 12

11..Nb4 12. axb4 Qxa1 13. Nb3

[d]2r1k1nr/pp3ppp/8/4Pb2/1b1p2P1/1N3N1P/1P2PP2/q1BQKB1R w Kk - 0 14

13...Bxb4+ 14. Nfd2 Bxd2+

But what is most interesting (from a human perspective) is that to everyone's shock in the chatroom, Komodo plays 15. Kxd2!

[d]2r1k1nr/pp3ppp/8/4Pb2/3p2P1/1N5P/1P1KPP2/q1BQ1B1R b k - 0 15

15...Qa4 16. gxf5 Qb4+ and then follows the forced 17. Kd3

[d]2r1k1nr/pp3ppp/8/4PP2/1q1p4/1N1K3P/1P2PP2/2BQ1B1R b k - 2 17

What an interesting position! Now, how many games have you ever seen in your life where the king ventures so far down the middle of the board, in the middle of a tactical firefight, and with queens and both rooks and two minor pieces still on board! No human GM would conceive of, let alone play this game this way... and to think of it here 2 of the strongest engines on this planet are playing in this seemingly erratic fashion. An incredible sight! Since I was in the chat during that pleasantly shocking time, I can say first hand that "the chat went nuts" is putting it lightly...

Stockfish expected here 17... Ne7 @ 18. Ke4!!

which I am sure would have completely brought the TCEC server down because people would have exploded in excitement! Unfortunately/fortunately, Komodo chose to play 18. f6 instead...

[d]2r1k2r/pp2nppp/5P2/4P3/1q1p4/1N1K3P/1P2PP2/2BQ1B1R b k - 0 18

...which Stockfish thought was a mistake at first and it's eval started rising.

But soon after a couple of forcing lines: 18. f6 Ng6 19. fxg7 Rg8 20. f4 Rxg7 21. Nxd4 Qxd4+ 22. Kxd4 Rd8+ 23. Ke4 Rxd1 24. Be3 Ra1 25. Bd4 Ne7 26. Bg2 Ra4 27. Bf3 Rg3 28. Bh5 Nc6 29. e3 Kf8 30. Bg4 Rb4

the position settled into 0.00

[d]5k2/pp3p1p/2n5/4P3/1r1BKPB1/4P1rP/1P6/7R w - - 3 31

The game was adjudicated as draw after 39. Reb4 after a 3-fold repetition.

[d]5k2/pp1B1p1p/2n2P2/1rR1P3/1r1BK3/4P2P/8/8 w - - 13 40

Don't you think this was a very interesting game, albeit a draw? Some draws are truly worth preserving, and this one is one of those immortal draws!

I hope you enjoyed the game as much as I did

[CRoberson]

Hi Adam, you have done a fantastic job, congratulations!
The current book is really much much better than older ones.

4 plies more at real TCEC really matters, I do not know what to say, maybe minimise the maximum score threshold to some 30cps at most? However, even that way, we learn a lot about certain positions that we otherwise would not have known.

Limiting the eval to 30 cps will not work. The issue is the "Horizon Effect". There isn't a way to predict the impact of the Horizon Effect at any number of plies in any way shape or form. What looks even at 30 ply might be a forced win/loss at 31 ply. What looks like a win at 24 ply might be a draw at 25 ply.

[chessmobile]

That game was interesting, thanks for posting. I wonder how things would have turned out if Ke4!! was played.