1.g4 opening is losing?

[Ovyron]

A lot of pages in this thread. Please tell me what was the conclusion, 1. g4 loses?

Yes.

[Uri Blass]

Well, my efforts got stuck improving 40.Bd3, so no wonder I couldn't get 40.Bd3 to a mate in 28 at the root. 40.Bd3 is good enough to mate any defense, unfortunately it takes too much time to find its shortest mate (which in my approach was necessary before solving 40.Bf4 because, how could you say your minimal mate is 28 if 40.Bd3 could mate in 27?)

I think it also takes a long time to find the shortest mate for 40.Bf4

Even checking that it is no more than 27 is hard when basically you want a tree when every mate is verifiable by the mate solver chest assuming we cannot trust stockfish with its hash collisions(I know that chest store all the board).

Here is some analysis that I did and I am not close to proving only that the line 40.Bf4 is mate in 23 or that there is no shorter mate(for no shorter mate you need to use chest or reliable mate solver that search for the shortest mate).

The tree is based on many hours of analysis with slow hardware and I am not close to proving by the analysis that 40.Bf4 is mate in at most 23.

[pgn][Event "Computer chess game"]
[Site "URIBLASS-THINK"]
[Date "2020.03.05"]
[Round "?"]
[White "UriBlass"]
[Black "Stockfish_20030109_x64_modern"]
[Result "*"]
[BlackElo "2000"]
[Time "07:27:05"]
[WhiteElo "2000"]
[TimeControl "300+5"]
[SetUp "1"]
[FEN "2b3k1/r2r2p1/p1N1q3/1B6/3p4/2b1B2R/PPP5/2K4R w - - 0 36"]
[Termination "unterminated"]
[PlyCount "29"]
[WhiteType "human"]
[BlackType "human"]

36. Rh8+ Kf7 37. Rf1+ Qf5 38. Rxf5+ Kg6 39. Rg5+ Kf6 40. Bf4 axb5 41. Nxa7
Rxa7 42. Rf8+ Rf7 43. Rxc8 Be1 44. Rc6+ Ke7 45. Rc7+ Kf6 46. Rxb5 Rxc7 47.
Bxc7 Ke6 {no mate in 10 chest} (47. .. Bf2 48. a4 {mate in 15 if to believe
stockfish} Ke7 (48. .. Ke6 49. a5) (48. .. Be3+ 49. Kd1) 49. a5) (47. .. g5
48. a4 {mate in at most 15 if you believe stockfish} Ke7 (48. .. Ke6 49.
a5) (48. .. d3 49. a5) 49. a5 {mate in at most 13 moves if you believe
stockfish and chest} Bxa5 50. Rxa5 {mate in exactly 12 moves} Kd7 (50. ..
Ke6 51. b4 {mate in exactly 11 moves chest}) (50. .. Kf6 51. Ra6+ {mate in
exactly 11 moves chest}) (50. .. Kf8 51. b4 {mate in exactly 11 moves
chest}) (50. .. d3 51. cxd3 {mate in exactly 11 moves chest}) (50. .. g4
51. c4 {mate in exactly 11 moves chest}) (50. .. Kf7 51. b4 {mate in
exactly 11 moves chest}) (50. .. Ke8 51. b4 {mate in exactly 11 moves
chest}) 51. Bg3 {mate in exactly 11 moves chest} Kc6 52. b4) (47. .. Kf7)
(47. .. Ke7) (47. .. d3) (47. .. g6) 48. Re5+ {mate in at most 15 moves}
Kd7 {no possible mate in 11 based on chest(better hardware need to prove no
mate in 12 or 13)} (48. .. Kf7 49. Rxe1 {mate in exactly 11 moves chest})
(48. .. Kf6 49. Rxe1 {mate in exactly 11 moves} g5 (49. .. d3 50. Bd8+
{mate in exactly 10 moves chest}) (49. .. g6 50. Bd8+ {mate in exactly 10
moves}) (49. .. Kf7 50. c4 {mate in exactly 10 moves chest}) (49. .. Kf5
50. b4 {mate in exactly 10 moves chest}) (49. .. Kg6 50. c4 {mate in
exactly 10 moves chest}) (49. .. Kg5 50. b4 {mate in exactly 10 moves
chest}) 50. c4 {mate in exactly 10 moves chest}) 49. Rxe1 {mate in exactly
14 moves conclusion} Kxc7 (49. .. d3 50. Bg3 {mate in exactly 10 moves
chest}) (49. .. Kc6 50. Bg3 {mate in exactly 10 moves chest}) (49. .. Kc8
50. Bg3 {mate in exactly 10 moves chest}) (49. .. g6 50. Bg3 {mate in
exactly 10 moves chest}) (49. .. g5 50. Bg3 {mate in 10 chest}) 50. a4
{mate in exactly 13 moves chest} *
[/pgn]

[Ovyron]

I think it also takes a long time to find the shortest mate for 40.Bf4

Yes, it's a consolation that not even Zenmastur could do it!

After what I've seen, at the rate I was advancing, I estimate that it'd have taken me around 4 more months of analysis (counting from the time I gave up) to find fastgm's mate. Had I known in advance that Zenmastur didn't actually have the fastest mate I'd have continued digging just to see his face when I finally showed a mate 1 move shorter than what he had (estimated completion time: 2020 June 23 11:11 pm GTM+3).

But then I read about Neville Goddard and surrendering to what you want, about how we should stop doing anything that requires too much effort. And indeed a few days later I had the fastest mate as posted by someone else without needing to do anything, so I should try surrendering more often.

Letting go of control.

[jp]

A lot of pages in this thread. Please tell me what was the conclusion, 1. g4 loses?

There is no conclusion, so what you want is the current summary:

One correspondence game on this bulletin board has been lost by White.
Another game is ongoing, but everyone predicts White will lose.

And that's about it.

... hard when basically you want a tree when every mate is verifiable by the mate solver chest assuming we cannot trust stockfish with its hash collisions (I know that chest store all the board).

Okay, so that's the reason you trust Chest and not SF. I see.

[Ovyron]

And that's about it.

Not really.

Thousands of hours have been spent analyzing 1.g4 by many members of the community and none have been able to find a single promising like that white could draw.

The only reason we don't share anything is because someone else like mmt could appear in the future, claiming they can draw 1.g4, and then we'd use all the analysis we've amassed to bust whatever they got. Otherwise all the analysis was just a waste of time.

[mmt]

I'd say 90% chance that it's lost. What should be done is training LC0 with all games starting at 1. g4. I might do it sometime but I still have to set some things up and it will take a lot of GPU time. Maybe somebody else can join or do it instead of me. Or training LC0 just for situations with a lot of "firepower" left on the board and letting SF takeover self-play when getting close to the endings.

[Ovyron]

You could save a lot of time by just inserting the winning lines for black, as, unlike drawn positions, black only needs a single line to win, so for instance, you don't need to check 1.g4 e5 (or any other move but 1...d5), or 1. g4 d5 2. Bg2 Nc6 (or any other move but 2...Bxg4). You just need to learn black's winning moves yourself and cut that big chunk from the training.

[zullil]

A lot of pages in this thread. Please tell me what was the conclusion, 1. g4 loses?

There is no conclusion, so what you want is the current summary:

One correspondence game on this bulletin board has been lost by White.
Another game is ongoing, but everyone predicts White will lose.

And that's about it.

Just for fun, I've started a game between Stockfish-dev (with Contempt = 0) and Stockfish-dev (with default Contempt = 24). The starting position has Black to move after 1. g4 d5 2. g5. Each side has 20 threads, 32 GB hash and 6-man Syzygy tables. Pondering is disabled. The time control is game-in-12-hours for each side.

Here's where things are now. Probably very few updates before the game is concluded.

1. g4 book d5 book
2. g5 book e5 +1.30/46 2484s
3. d4 -1.21/47 1834s exd4 +1.40/44 217s

[Hai]

It's funny to see what's going on here.

I only wanted to mention I have 1.g4 down to -1.11 with LC0.
It's a draw.
And I have completely other variations.

[zullil]

It's funny to see what's going on here.

I only wanted to mention I have 1.g4 down to -1.11 with LC0.
It's a draw.
And I have completely other variations.

Great! I believe 1.g4 is a theoretical draw. Please provide your analysis, if you are able to do so.