1.g4 opening is losing?

[Ovyron]

I'd forgotten to turn off contempt and had to restart

A contempt like 18 will be more useful than 0 as it will find better lines faster. A trick is to keep scores from the white side separate from scores of the black side (which would be lower, so the sides can have favored lines that differ, white with a mainline that has the most positions where black is likely to blunder, and black with one where white is most likely to blunder, which aren't the same.) What is the real score depends on the position being won or drawn for black.

[jjoshua2]

LS14 that Pohl tested 46 elo stronger than LS13 after go 50M nodes.

Code: Select all

info depth 37 seldepth 90 time 1772501 nodes 27383748 score cp 2597 hashfull 1000 nps 15142 tbhits 0 multipv 1 pv f1g2 c8g4 c2c4 c7c6 d1b3 e7e6 b3b7 b8d7 b1c3 g8e7 c4d5 e6d5 d2d4 a8b8 b7a6 b8b6 a6d3 e7g6 h2h3 g4e6 g1f3 f8d6 h3h4 h7h6 g2h3 e6h3 h1h3 e8g8 h4h5 g6e7 b2b3 c6c5 h3h1 d8c8 c3a4 b6c6 d4c5 d7e5 d3c3 e5f3 c3f3 d6e5 c1b2 e5b2 a4b2 c6c5 b2d3 c5c2 e1f1 e7f5 d3b4 c2c1 a1c1 c8c1 f1g2 c1g5 g2h3 f8e8 f3g4 e8e4 g4g5 h6g5 b4d5 e4e2 h3g4 f5h6 g4f3 e2a2 h1c1 a2a3 c1c3 a7a5 f3g2 g8h7 c3c5 a3b3 c5a5 h6g4 d5e7 f7f6 e7f5 b3b2 g2g3 g4h6
info depth 37 seldepth 90 time 1772501 nodes 27383748 score cp 2507 hashfull 1000 nps 15142 tbhits 0 multipv 2 pv c2c4 e7e5 f1g2 d5c4 b1c3 b8c6 d1a4 g8e7 g1f3 e7g6 a4c4 c8e6 c4b5 e6d7 b5c4 g6f4 h1g1 f8d6 d2d3 e8g8 g2h1 c6a5 c4e4 f8e8 h2h4 d7c6 e4e3 b7b6 c3e4 a5b7 f3g5 d8d7 e4d6 c7d6 h1c6 d7c6
info depth 37 seldepth 90 time 1772501 nodes 27383748 score cp 2253 hashfull 1000 nps 15142 tbhits 0 multipv 3 pv h2h3 h7h5 g4h5 e7e5 d2d3 b8c6 f1g2 g8f6 b1c3 f8e7 c1g5 c8e6 e2e4 d5e4 d3e4 f6h5 g5e3 h5f4 e3f4 e5f4 c3d5 d8d6 d1d2 e8c8 e1c1 c6b4 a2a3 b4d5 e4d5 e6f5 g1e2

So winrate dropped 2.74% or 16.8 elo per %. So linearly, we need 436 elo stronger engine/search to solve to 0% winrate

LS14.1 about 8 elo stronger but has quite different winrate ideas than 14.0. It sees f1g2 as only possible try to save. Sligthly higher cpuct than I used for 14.0 so depth is worse but similar number of total nodes.

Code: Select all

info depth 31 seldepth 85 time 1227259 nodes 24826537 score cp 3276 hashfull 1000 nps 19462 tbhits 0 multipv 1 pv f1g2 c8g4 c2c4 c7c6 d1b3 e7e6 b3b7 b8d7 b1c3 g8e7 c4d5 e6d5 d2d4 a8b8 b7a6 b8b6 a6d3 e7g6 h2h3 g4e6 g1f3 f8d6 h3h4 h7h6 g2h3 e6h3 h1h3 d7f6 h4h5 d8c8 h3h1 g6f4 c1f4 d6f4 e2e3 f4d6 e1c1 e8g8 d1g1 g8h8 g1g2 c8b7 b2b3 c6c5 d4c5 d6c5 h1g1 c5a3 c1d1 f6h5 d3d4 a3e7 f3e5 e7f6 g2h2 g7g6 f2f4 f6g7 d4d5 b7e7 h2d2
info depth 31 seldepth 85 time 1227259 nodes 788396 score cp 2477 hashfull 1000 nps 19462 tbhits 0 multipv 2 pv c2c4 e7e5 c4d5 d8d5 g1f3 c8g4 b1c3 d5a5 f3g5 b8c6 d1b3 c6d8 h1g1 g8f6 d2d3 h7h6 g1g4 f6g4 g5f7 d8f7 b3e6 f8e7 e6g4 e8g8 f1g2 e7b4 c1d2 f7d6 e1c1 b4c3 d2c3 a5c5 d1g1 f8f7 c1b1 a8e8 g2f3 e8e7 c3b4 c5f2 f3d5 g8h8 d5f7 e7f7 h2h4 d6f5 b4c3 f5d4 c3d4 e5d4 a2a3 c7c5 h4h5 b7b6
info depth 31 seldepth 85 time 1227259 nodes 227965 score cp 2225 hashfull 1000 nps 19462 tbhits 0 multipv 3 pv h2h3 h7h5 g4h5 e7e5 d2d3 b8c6 f1g2 g8f6 b1c3 c8e6 c1g5 f8e7 h3h4 d8d6 e2e4 d5d4 g5f6 g7f6 c3d5 e8c8 g1e2 f6f5 e2g3 f5e4 d5e7 c6e7 g3e4 d6b6 e4g5 d8g8 g2f3 e6d5 f3d5

[jp]

(Of course, I have no proof that the final position here is genuinely drawn - but it is evaluated at 0.00 at depth 40, so clearly black has already blundered by this point)

Or else this is a drawing line for White. People who believe Black wins by force are free to point out where they think Black erred.

[Exa65536]

Or else this is a drawing line for White. People who believe Black wins by force are free to point out where they think Black erred.

Some quick backwards analysis suggests that by move 11 white was already a long ways towards equalizing, and that 10... Na6 was better than 10... dxe2

(But this isn't a tremendously sharp opening - I could keep pointing out better moves for black and then finding ways to draw with that error corrected until the cows come home without making any real progress, due to combinatorial explosion. And I say that with confidence because I have been doing that, finding drawing lines against CDBCN's preferred lines, submitting them to CDBCN, which makes it change its response, and repeating. I've seen no evidence of either my running out of drawing lines or CDBCN running out of candidate moves.)

[Zenmastur]

(Of course, I have no proof that the final position here is genuinely drawn - but it is evaluated at 0.00 at depth 40, so clearly black has already blundered by this point)

Or else this is a drawing line for White. People who believe Black wins by force are free to point out where they think Black erred.

1.g4 d5 2.c4 d4?!

[Ovyron]

(But this isn't a tremendously sharp opening - I could keep pointing out better moves for black and then finding ways to draw with that error corrected until the cows come home without making any real progress, due to combinatorial explosion. And I say that with confidence because I have been doing that, finding drawing lines against CDBCN's preferred lines, submitting them to CDBCN, which makes it change its response, and repeating. I've seen no evidence of either my running out of drawing lines or CDBCN running out of candidate moves.)

That's why playing a game is so important, even without time limit you eventually have to commit to a move, and in this format nobody has been able to find any white defense that draws.

[Exa65536]

That's why playing a game is so important, even without time limit you eventually have to commit to a move, and in this format nobody has been able to find any white defense that draws.

Yeah. My personal opinion is that 1. g4 probably is lost by force - but if you committed to some deterministic procedure by which black would decide on a move, and let white determine the output of that procedure arbitrarily many times on arbitrary positions, I'd guess that that procedure would need to be really quite close to perfect play before white couldn't find a drawing line (that is, white is lost, but just barely). Clearly it'd need to be stronger than 1 Gn/move stockfish, and as an admittedly wholly unsupported guess I'd say that the required nodecount would be many orders of magnitude higher than that.

[Ovyron]

From everything I've seen since October 2019 (back when I thought it could be saved), black wins easily. The rest is just about how badly unassisted engines suck (so it makes sense how 1G/nodes fails, as would higher node counts.)

Measuring by nodes does not convince me, though, I'd still like to see a draw against Depth 39 (on Game 1 I was still able to produce potential drawing lines against Depth 38, up until the very end, so D39 is where the line could be drawn and it'd be enough to defeat any defense.)

[jp]

That's why playing a game is so important, even without time limit you eventually have to commit to a move, and in this format nobody has been able to find any white defense that draws.

From everything I've seen since October 2019 (back when I thought it could be saved), black wins easily.

If playing a game is so important, then so far you have only two pieces of data on this forum to base your opinion on. Yet you talk about "everything I've seen", etc.

[Nordlandia]

1. b4 is the mirror in FRC #534 starting position.

K contempt at -13
SF contempt at +24
TC: 30m+30s
CPU: 5930K 4.5GHz 5-cores per engine.
Hash: 8192 mb

[pgn][Event "?"]
[Site "i7-5960X 4.5GHz"]
[Date "2020.04.05"]
[Round "?"]
[White "komodo-13.3-64bit-bmi2"]
[Black "stockfish_20040215_x64_bmi2"]
[Result "0-1"]
[FEN "rnbkqbnr/pppppppp/8/8/1P6/8/P1PPPPPP/RNBKQBNR b KQkq - 0 1"]
[GameDuration "01:53:25"]
[GameEndTime "2020-04-05T12:06:55.007 Vest-Europa (sommertid)"]
[GameStartTime "2020-04-05T10:13:29.546 Vest-Europa (sommertid)"]
[PlyCount "124"]
[SetUp "1"]
[Termination "adjudication"]
[TimeControl "1800+30"]
[Variant "fischerandom"]

1... e5 {+1.09/32 61s} 2. Bb2 {-1.10/30 49s} Bxb4 {+1.15/35 67s}
3. f4 {-1.18/33 80s} f6 {+1.20/33 19s} 4. Qg3 {-1.13/31 49s} d6 {+1.35/32 59s}
5. Qxg7 {-1.11/32 83s} Ne7 {+1.71/32 31s} 6. Nf3 {-1.11/31 71s}
Nbc6 {+1.38/34 93s} 7. a3 {-0.82/33 185s} Ba5 {+1.50/33 44s}
8. Nc3 {-0.68/32 60s} Rf8 {+1.74/32 49s} 9. e4 {-1.08/28 51s} Be6 {+1.64/31 28s}
10. f5 {-1.07/31 119s} Bg8 {+1.85/33 55s} 11. Bd3 {-0.99/35 137s}
O-O-O {+1.76/30 42s} 12. O-O {-1.05/31 43s} d5 {+1.69/34 79s}
13. Qg4 {-1.17/33 80s} Kb8 {+1.64/33 59s} 14. Rae1 {-1.20/31 39s}
h5 {+1.90/32 20s} 15. Qh3 {-1.19/32 44s} d4 {+1.95/36 40s}
16. Nd1 {-1.23/33 54s} Nc8 {+2.03/35 32s} 17. Nf2 {-1.17/32 43s}
Nb6 {+2.01/33 34s} 18. Rb1 {-1.32/31 42s} a6 {+2.06/40 97s}
19. Be2 {-1.18/32 85s} Na4 {+2.15/34 24s} 20. Bc1 {-1.37/31 39s}
Ba2 {+2.16/39 53s} 21. Ra1 {-1.51/30 57s} Bf7 {+2.15/37 124s}
22. Rb1 {-1.52/32 116s} Ba2 {+2.23/39 167s} 23. Ra1 {-1.11/32 59s}
Bf7 {+2.15/37 33s} 24. Rb1 {-1.47/32 42s} Ka8 {+2.13/34 42s}
25. Bd1 {-1.53/34 226s} Ne7 {+2.17/37 92s} 26. Qh4 {-1.63/31 34s}
Rd6 {+2.14/33 22s} 27. Be2 {-1.57/33 115s} Rg8 {+2.20/35 92s}
28. Ne1 {-1.51/33 142s} Qc6 {+2.46/34 32s} 29. Kh1 {-1.71/33 173s}
Nc8 {+2.62/33 33s} 30. Rg1 {-1.87/34 132s} Ncb6 {+2.53/34 45s}
31. Ra1 {-1.94/34 86s} Nc4 {+2.87/34 34s} 32. Nf3 {-1.95/33 24s}
Rdd8 {+2.76/38 81s} 33. Rb1 {-1.96/32 124s} Ncb6 {+2.84/33 51s}
34. Ne1 {-1.83/29 17s} Rg7 {+2.82/35 48s} 35. Ned3 {-1.83/32 42s}
Rdg8 {+2.74/34 45s} 36. Nb4 {-1.82/33 68s} Qd6 {+2.54/37 80s}
37. Nd1 {-1.93/32 36s} c5 {+3.19/34 38s} 38. Nd3 {-1.93/30 17s}
c4 {+3.12/34 80s} 39. Nb4 {-1.80/30 18s} Nc8 {+3.35/34 47s}
40. d3 {-1.78/32 32s} c3 {+3.39/34 52s} 41. Bf3 {-1.73/35 16s}
Ne7 {+3.57/36 76s} 42. Nf2 {-1.87/35 36s} Bd8 {+3.78/37 45s}
43. Bh6 {-1.86/37 61s} Rh7 {+3.94/38 41s} 44. Bc1 {-2.00/35 29s}
Ka7 {+4.12/39 127s} 45. Ra1 {-2.23/32 31s} a5 {+4.44/33 25s}
46. Nd5 {-2.36/32 19s} Nxd5 {+4.64/36 35s} 47. exd5 {-2.35/34 34s}
Bxd5 {+4.77/36 29s} 48. Qh3 {-2.53/33 30s} b5 {+4.99/36 36s}
49. Ne4 {-2.66/34 38s} Qc6 {+5.12/35 32s} 50. Re1 {-2.81/34 30s}
Kb6 {+5.48/39 117s} 51. Kg1 {-2.91/32 30s} Nc5 {+5.66/35 19s}
52. Kf1 {-3.11/30 30s} Be7 {+5.78/35 27s} 53. Re2 {-3.22/32 30s}
Nxe4 {+6.49/36 29s} 54. Bxe4 {-3.41/34 30s} Bxe4 {+7.76/32 28s}
55. dxe4 {-4.89/32 30s} Qc4 {+8.33/32 29s} 56. Qd3 {-5.37/32 19s}
h4 {+8.79/33 67s} 57. Kf2 {-6.00/34 41s} Kc6 {+9.03/33 26s}
58. Re1 {-6.40/35 30s} Qxd3 {+9.52/30 41s} 59. cxd3 {-6.73/33 19s}
b4 {+9.69/31 36s} 60. axb4 {-6.96/32 40s} axb4 {+10.20/30 32s}
61. Kf3 {-7.40/32 30s} b3 {+10.59/30 30s} 62. Ra6+ {-7.09/32 30s}
Kb5 {+10.84/32 27s} 63. Ra7 {-8.12/33 26s, Black wins by adjudication} 0-1
[/pgn]