Komodo 11.01 vs asmFish 170522

[Jeroen]

OK, I completely missed this match. 40' is of course slower than 15', but it is still quite a difference to the TCEC tempo.

BTW, it seems that at that time SF was winning against Komodo by similar margins using different testsets. See the comments in the thread.

If the distance between SF and Komodo is now smaller than at the end of 2016, I would expect a (much) closer match using such short opening lines.

[Jeroen]

TCEC superfinal 8 used shorter opening lines with level scores. The result: 89% draw rate. Nine of these wins were in lines where one side had a small advantage. Only 2 wins were in lines with scores around 0.00. Two wins out of 66 games, which is a winning percentage of 3%.

TCEC superfinal 9 used a mixed bag of opening lines, longer and shorter, with imbalances. The result: 75% draw rate.

You can twist it any way you want, but the length of an opening line doesn't say anything about the winning chances.

To give you a simple example:

Line A) 1.e4 e6 2.d4 d5 3.exd5 exd5

Line B) 1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 g6 6.Be3 Bg7 7.f3 O-O 8.Qd2 Nc6 9.Bc4 Bd7 10.O-O-O

Which one of the two would you pick to generate the highest probability of a win?

[Jeroen]

What is more: short and balanced lines between two engines of similar strength will never lead to brilliant games like this one:

[pgn][Event "TCEC Season 9 - Superfinal"]
[Site "http://tcec.chessdom.com"]
[Date "2016.11.15"]
[Round "17"]
[White "Stockfish 8"]
[Black "Houdini 5"]
[Result "1-0"]
[ECO "B78"]
[WhiteElo "3228"]
[BlackElo "3182"]
[PlyCount "143"]
[EventDate "2016.11.11"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 g6 6. Be3 Bg7 7. f3 O-O 8. Qd2
Nc6 9. Bc4 Bd7 10. O-O-O Rc8 11. Bb3 Ne5 12. h4 Nc4 13. Bxc4 Rxc4 14. h5 Nxh5
15. g4 Nf6 16. Kb1 Re8 17. b3 Rc8 18. Nd5 Nxd5 19. exd5 e5 20. dxe6 fxe6 21.
Qh2 Qf6 22. a4 b6 23. Qxh7+ Kf7 24. g5 Qe5 25. Rh6 Qxe3 26. Rxg6 Rg8 27. f4 d5
28. f5 exf5 29. Nxf5 Qxb3+ 30. cxb3 Bxf5+ 31. Ka2 Rc2+ 32. Ka3 Bxg6 33. Rf1+
Ke7 34. Qxg8 Bb2+ 35. Kb4 Bc3+ 36. Kb5 Bd3+ 37. Kc6 Be5+ 38. Kb7 Rc7+ 39. Ka8
Bxf1 40. Qxd5 Bg7 41. Kb8 Rd7 42. Qe4+ Kf8 43. Qf5+ Rf7 44. Qc8+ Ke7 45. Qc7+
Ke6 46. Qc6+ Kf5 47. Qd5+ Be5+ 48. Ka8 Rf8+ 49. Kxa7 Be2 50. b4 Bh5 51. Kxb6
Bf7 52. Qf3+ Bf4 53. Qc6 Rb8+ 54. Ka7 Kxg5 55. Qd7 Bh5 56. b5 Re8 57. b6 Be3
58. Qd5+ Kh4 59. a5 Re7+ 60. Ka8 Re8+ 61. Kb7 Re7+ 62. Kc8 Bg4+ 63. Kb8 Bf4+
64. Ka8 Kg3 65. Qg8 Re5 66. a6 Re6 67. Kb7 Be3 68. a7 Rxb6+ 69. Kc7 Ra6 70.
a8=Q Bf4+ 71. Kb7 Rxa8 72. Kxa8 1-0[/pgn]

[Lyudmil Tsvetkov]

TCEC superfinal 8 used shorter opening lines with level scores. The result: 89% draw rate. Nine of these wins were in lines where one side had a small advantage. Only 2 wins were in lines with scores around 0.00. Two wins out of 66 games, which is a winning percentage of 3%.

TCEC superfinal 9 used a mixed bag of opening lines, longer and shorter, with imbalances. The result: 75% draw rate.

You can twist it any way you want, but the length of an opening line doesn't say anything about the winning chances.

To give you a simple example:

Line A) 1.e4 e6 2.d4 d5 3.exd5 exd5

Line B) 1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 g6 6.Be3 Bg7 7.f3 O-O 8.Qd2 Nc6 9.Bc4 Bd7 10.O-O-O

Which one of the two would you pick to generate the highest probability of a win?

you seem not ot read what I write, so it is difficult to discuss.

with same same, +50cps scores, for a 2-movelong and a 12-move-long opening, the shorter book will achieve much higher winning rate.

beautiful, high-quality games are everywhere, but of course, most spectacular and deepest play is possible in the early opening.

so far, the most spectacular stage of TCEC was stage 4 of, I already forgot what edition of TCEC, when a short 2-moves book was used. very intriguing play, very high winning rate.

of course, if you want high winning percentage, you might simply start from move 20 and +300cps score.

[Jeroen]

It is indeed difficult to discuss something with somebody who only claims something without any proof. And who changes the subject from 'equal scores' to '+0.5 advantage'. In your previous posting you wrote:

with equal scores, a short book will return higher winning percentage, because engines will make more mistakes

Now you are speaking about a 2-move book with +0.5 scores. That's an entirely different subject. Besides, in that case the 2-move book needs to have a lot of off-beat lines to get these +0.5 book exits. Not really interesting for a TCEC superfinal.

Far more interesting is f.e. a King's Indian Mar del Plata, where you give the engines the first 12-14 moves. They will say white is clearly better, which is highly exaggerated.

A last point: a 2-move off-beat book for the TCEC superfinal would have been dismissed by Cato. The same goes for a short book with scores around 0.00. In both cases all people following the TCEC superfinal would be complaining about the 'crap openings'. Rightly so.

[Jeroen]

You also didn't answer my question, if you would choose line A or B. Of course B will have much better winning chances, which you are probably very well aware of.

Once again my statement: it is the POSITION that counts, NOT the number of moves. As an 2180 elo player you should know that.

For me as an opening expert, 2-move book games by engines are just of limited interest. In that case you will almost never see gambits, sharp lines, highly topical main lines, castling on opposite sides, and so forth. They will mostly play classical and avoid the main lines, that are interesting for most players, GM's and so on. And IF they play a main line, you might as well have given them the first 10-20 moves (if that is what they are going to play anyway).

[Lyudmil Tsvetkov]

You also didn't answer my question, if you would choose line A or B. Of course B will have much better winning chances, which you are probably very well aware of.

Once again my statement: it is the POSITION that counts, NOT the number of moves. As an 2180 elo player you should know that.

For me as an opening expert, 2-move book games by engines are just of limited interest. In that case you will almost never see gambits, sharp lines, highly topical main lines, castling on opposite sides, and so forth. They will mostly play classical and avoid the main lines, that are interesting for most players, GM's and so on. And IF they play a main line, you might as well have given them the first 10-20 moves (if that is what they are going to play anyway).

do not have time to discuss with a person who distorts facts.

I have nothing to do with TCEC, it is their business, but shorter lines would definitely give better and more interesting play.

last remark: so far, my highest winning percentage is when I play 2 engines without book at all. SF vs Komodo with no book at all usually end at somewhere 65%+ winning games. this is at 1 min. TC. with any opening book, they score less.

and games are random, non-deterministic, even with single core.

[beram]

This is only blitz, in which games are more tactical and you get more decisive results. Besides, the other reason this set produces quite a lot of wins, is the inclusion of short lines that engines will never play themselves (and they also do not particularly like them).

In a LTC match a la TCEC the number of draws will obviously increase significantly, maybe there is someone on this forum with strong hardware who'd like to try this out.

Well perhaps but now at 5 times longer time control 15m10s and 14cpu(!) it makes no difference
as with the match pedantfish 170426 versus K10.4 no lesser draws after 10 games but again allready +4 with 6 draws for the fish

after 35 games I have +13, 22 draws and yet not a single loss for asmfish170522 which means 68,5%

[beram]

This is only blitz, in which games are more tactical and you get more decisive results. Besides, the other reason this set produces quite a lot of wins, is the inclusion of short lines that engines will never play themselves (and they also do not particularly like them).

In a LTC match a la TCEC the number of draws will obviously increase significantly, maybe there is someone on this forum with strong hardware who'd like to try this out.

Well perhaps but now at 5 times longer time control 15m10s and 14cpu(!) it makes no difference
as with the match pedantfish 170426 versus K10.4 no lesser draws after 10 games but again allready +4 with 6 draws for the fish

after 35 games I have +13, 22 draws and yet not a single loss for asmfish170522 which means 68,5%

and finally after 36 games the first victory for Komodo +13 -1 =22 is now 63,5%

[Leo]

Nice hardware. Nice tournament.