Komodo vs. Larry K on chess.com

[jp]

The standard position is playable between top engines as long as there are wins.
<5% wins is still not 0% wins and it is possible to give white more time to reduce the percentage of draws.

It may be interesting to know the percentage of draws in a match with no opening book from the opening position between stockfish and lc0 when white get 10 minutes per move against 1 minutes per move by black(of course no pondering).

Uri, if you have a match between two strong engines with white given a huge time advantage, you'll reduce the draws but also make sure there are zero black wins. Would you really like such a match?

[Marcus9]

The standard position is playable between top engines as long as there are wins.
<5% wins is still not 0% wins and it is possible to give white more time to reduce the percentage of draws.

It may be interesting to know the percentage of draws in a match with no opening book from the opening position between stockfish and lc0 when white get 10 minutes per move against 1 minutes per move by black(of course no pondering).

Uri, if you have a match between two strong engines with white given a huge time advantage, you'll reduce the draws but also make sure there are zero black wins. Would you really like such a match?

Even with same time in LTC, good hardware and top engine, white is basically the only winner, However I thought a similar match, Both engine have same number of games with time odd, And then compare result.
Eventually a bonus can be given to the victories Despite handicap, Using them as an example +3

For example:
Etheral 30 min vs laser 3 min
+70=25-5

Laser 30 min vs etheral 3 min
+75=15-10

Winner could be etheral with a score of 120-110 (+15 elo)

[jp]

Even with same time in LTC, good hardware and top engine, white is basically the only winner

Looking at TCEC (for example) you do see black wins.

[Marcus9]

Even with same time in LTC, good hardware and top engine, white is basically the only winner

Looking at TCEC (for example) you do see black wins.

They are rare, and most are with openings that benefit black, such as king's gambit, I'm talking about engines with similar elo

[jp]

They are rare, and most are with openings that benefit black, such as king's gambit, I'm talking about engines with similar elo

I looked at the TCEC archive and among Komodo, Houdini, Leela, SF, there are Black wins and out of tame openings that they evaluate as even.

[lkaufman]

They are rare, and most are with openings that benefit black, such as king's gambit, I'm talking about engines with similar elo

I looked at the TCEC archive and among Komodo, Houdini, Leela, SF, there are Black wins and out of tame openings that they evaluate as even.

Of course if the opening eval is near zero (I suppose that's what you mean by "even"), then color means nothing. But with normal openings played by strong grandmasters, White is always noticeably better after 6-8 moves or so. When the openings are limited to those seen in top GM play, I suspect that the White to Black win ratio between the top engines in long games on big hardware is huge. Giving White more time to increase wins is a reasonable idea, especially for human play (the Armageddon idea), but if the time ratio gets too big, like 10 to 1, it seems rather silly.

[jp]

Of course if the opening eval is near zero (I suppose that's what you mean by "even"), then color means nothing. But with normal openings played by strong grandmasters, White is always noticeably better after 6-8 moves or so. When the openings are limited to those seen in top GM play, I suspect that the White to Black win ratio between the top engines in long games on big hardware is huge.

I'd expect that in superGM games we'd see out of the opening a White advantage that engines would give maybe +0.3. Would you agree? (If White is able to get a +0.7 advantage according to engines I think he'd be very happy.)

What you see in closed superGM tournaments is that they get nowhere with sharp theory, because they've analysed the same lines with the same engines. You then see at least some Giuoco Pianos and d3 Ruys, which I'm guessing Komodo might give less than +0.3.

I agree the W:B win ratio between top engines will be huge, but that's not the same as B wins being zero.


This is the book position in the TCEC game Stockfish 190203-Houdini 6.03:

[d]rnbqk2r/ppp2pp1/4pb1p/3p4/2PP4/2N2N2/PP2PPPP/R2QKB1R w KQkq - 0 7

Straight out of the book, Stockfish's eval is +0.15. Houdini won the game as Black. Maybe we want White to have a +0.3 advantage, i.e. a bit bigger, but this is made up for by Stockfish 190203 being stronger than Houdini 6.03.

[pgn]1. Nf3 Nf6 2. d4 d5 3. c4 e6 4. Nc3 Be7 5. Bg5 h6 6. Bxf6 Bxf6 7. e3 O-O 8. Qd2 c6 9. h4 dxc4 10. Bxc4 c5 11. O-O-O cxd4 12. exd4 Bd7 13. g4 Bc6 14. Be2 Nd7 15. g5 Bxf3 16. Bxf3 hxg5 17. Bxb7 gxh4 18. Bxa8 Qxa8 19. Qf4 Rb8 20. Qe4 Qxe4 21. Nxe4 Be7 22. Kc2 f5 23. Nd2 g5 24. Nf3 Kg7 25. Rd3 Bf6 26. a4 Kg6 27. Rc3 Nb6 28. Rc6 Re8 29. Ne5+ Bxe5 30. dxe5 Nd5 31. Kd2 Kh5 32. Rhc1 g4 33. Rc8 Re7 34. b4 h3 35. b5 Rd7 36. R8c6 f4 37. Ke2 g3 38. Kf3 g2 39. Rc8 Rh7 40. Rg8 Ne7 41. Rxg2 hxg2 42. Kxg2 Ng6 43. Kf3 Rd7 44. Rh1+ Kg5 45. a5 Nxe5+ 46. Ke4 Ng4 47. Rh8 Nxf2+ 48. Kf3 Ng4 49. a6 Rd3+ 50. Ke4 Nf2+ 0-1[/pgn]

[jp]

The book position in the TCEC game LCZero v20.2-32930 vs. Stockfish 190203 (0-1):

[d]r1bqkb1r/pp1p1ppp/2n1p3/3nP3/3P4/5N2/PP3PPP/RNBQKB1R w KQkq - 0 7

Straight out of the book, LCZero's eval is +0.67 (although LCZero plays Bc4 and then Stockfish's eval is 0.00).


The book position in the TCEC game Komodo 2227.00 vs. Houdini 6.03 (0-1):

[d]rnbqkb1r/p2n1ppp/1p2p3/2ppP3/3P4/2PB4/PP1N1PPP/R1BQK1NR w KQkq - 0 7

Komodo's eval: +0.7.

(I just saw that the TC for these games was fast, though.)

[lkaufman]

Of course if the opening eval is near zero (I suppose that's what you mean by "even"), then color means nothing. But with normal openings played by strong grandmasters, White is always noticeably better after 6-8 moves or so. When the openings are limited to those seen in top GM play, I suspect that the White to Black win ratio between the top engines in long games on big hardware is huge.

I'd expect that in superGM games we'd see out of the opening a White advantage that engines would give maybe +0.3. Would you agree? (If White is able to get a +0.7 advantage according to engines I think he'd be very happy.)

What you see in closed superGM tournaments is that they get nowhere with sharp theory, because they've analysed the same lines with the same engines. You then see at least some Giuoco Pianos and d3 Ruys, which I'm guessing Komodo might give less than +0.3.

I agree the W:B win ratio between top engines will be huge, but that's not the same as B wins being zero.


This is the book position in the TCEC game Stockfish 190203-Houdini 6.03:

[d]rnbqk2r/ppp2pp1/4pb1p/3p4/2PP4/2N2N2/PP2PPPP/R2QKB1R w KQkq - 0 7

Straight out of the book, Stockfish's eval is +0.15. Houdini won the game as Black. Maybe we want White to have a +0.3 advantage, i.e. a bit bigger, but this is made up for by Stockfish 190203 being stronger than Houdini 6.03.

[pgn]1. Nf3 Nf6 2. d4 d5 3. c4 e6 4. Nc3 Be7 5. Bg5 h6 6. Bxf6 Bxf6 7. e3 O-O 8. Qd2 c6 9. h4 dxc4 10. Bxc4 c5 11. O-O-O cxd4 12. exd4 Bd7 13. g4 Bc6 14. Be2 Nd7 15. g5 Bxf3 16. Bxf3 hxg5 17. Bxb7 gxh4 18. Bxa8 Qxa8 19. Qf4 Rb8 20. Qe4 Qxe4 21. Nxe4 Be7 22. Kc2 f5 23. Nd2 g5 24. Nf3 Kg7 25. Rd3 Bf6 26. a4 Kg6 27. Rc3 Nb6 28. Rc6 Re8 29. Ne5+ Bxe5 30. dxe5 Nd5 31. Kd2 Kh5 32. Rhc1 g4 33. Rc8 Re7 34. b4 h3 35. b5 Rd7 36. R8c6 f4 37. Ke2 g3 38. Kf3 g2 39. Rc8 Rh7 40. Rg8 Ne7 41. Rxg2 hxg2 42. Kxg2 Ng6 43. Kf3 Rd7 44. Rh1+ Kg5 45. a5 Nxe5+ 46. Ke4 Ng4 47. Rh8 Nxf2+ 48. Kf3 Ng4 49. a6 Rd3+ 50. Ke4 Nf2+ 0-1[/pgn]

I would say a normal White edge is maybe 0.25 Komodo scale, 0.30 Stockfish scale, 0.40 Lc0 scale. Yes, Giuoco Pianos and d3 Ruys are quite normal top GM openings. White often gets a position with almost all of the winning chances, but only a small percentage of the games will end decisively. Some openings are more double-edged than others, so with clever selection (like TCEC tries to do) Black will win some of these games, but if the selection were limited to positions common in top GM play, Black wins between top engines would be pretty rare.

[Ovyron]

I would say a normal White edge is maybe 0.25 Komodo scale, 0.30 Stockfish scale, 0.40 Lc0 scale.

Interesting. For more than 3 years I have a 0.19 hard limit for a Stockfish scale advantage, that is, I haven't been able to break it and if Stockfish shows >0.19 it means black has already made some lousy moves that allowed white an edge this big. So if "0.30 SF" is considered normal then I guess people consider lousy black defenses normal.