mmt Vs. Ovyron (G4 D5 BG2)

[Ovyron]

Thanks for the challenge!

Here's white trying to defend 1...d5 2.g5
Here's from where these games are coming from

1. g4 d5 2. Bg2 Bxg4

[d]rn1qkbnr/ppp1pppp/8/3p4/6b1/8/PPPPPPBP/RNBQK1NR w KQkq -

[mmt]

I am using mostly LC0 with some Stockfish 11 help. I'll be running difficult moves overnight and "easier" ones for maybe an hour or two. Good luck to you! I actually hope you win, so we can say that 1. g4 (or at least 1. g4 d5 2. Bg2) very much appears like a losing opening. That'd be something!

3. c4

[d]rn1qkbnr/ppp1pppp/8/3p4/2P3b1/8/PP1PPPBP/RNBQK1NR b KQkq c3 0 3

if you play 3... c6 then 4. Qb3

[Spliffjiffer]

great news
at some point, say somewhere roughly between move 15 and 30, id prefere to rely more on SF rather than LC honestly...since this would require a very fast CPU though maybe even a cooperation might make sense at this point (if mmt doesnt use a fast one already)
whatever your decisions will be, good "luck" to both of you and a lot of fun, ill be curious to follow your game.
(glue the white mess together hehe)

thx for making the challenge public

[jp]

I actually hope you win, so we can say that 1. g4 (or at least 1. g4 d5 2. Bg2) very much appears like a losing opening. That'd be something!

It'd feel quite surprising if 1.g4 does lose.

I said in the other thread I feel favorable (philosophically) towards c4 and Bg2, instead of going down fussing about that g4 pawn. Just ditch the pawn and hang on for a draw!

[mmt]

at some point, say somewhere roughly between move 15 and 30, id prefere to rely more on SF rather than LC honestly...since this would require a very fast CPU though maybe even a cooperation might make sense at this point (if mmt doesnt use a fast one already)

Yes, I actually specifically saw it when playing out this opening. LC0 failed to see a nice tactical shot: http://talkchess.com/forum3/viewtopic.p ... 70#p826973. My CPU is just ok, 32000 score on Geekbench 4. I do have a lot of RAM though (512 GB) and 2 fast GPUs.

[mmt]

It'd feel quite surprising if 1.g4 does lose.

If I lose, I'll go over key moves (e.g. when SF and LC0 disagree or scores are close or seem to trend in one direction with time...) with even more time and detail and then maybe we can also play out an alternate variation if the best move changes or if it's very close.

I'll add some commentary based on engine output.

[Ovyron]

Back in October 2019 I did extensive analysis of 2.Bg2 for my game against Harvey, including analysis with CPU Leela, and even though she's the cutest little thing around here, averaging some 11111 nodes / 10 minutes, she still plays the same line shown with her 800M nodes. And I pounced on her until even she agreed 2.g5 was better, and I didn't use her at all after committing to 2.g5.

It'd be interesting to know what's her score for 2.g5, and if I can't win the game, if at the very least I managed to make her show a worse score for 2.Bg2, as I did around here.

I think Leela's line indeed ends in a draw and black is the one that needs to deviate for a win, but not yet

1. g4 d5 2. Bg2 Bxg4 3. c4 c6

[d]rn1qkbnr/pp2pppp/2p5/3p4/2P3b1/8/PP1PPPBP/RNBQK1NR w KQkq -

[mmt]

1. g4 d5 2. Bg2 Bxg4 3. c4 c6 4. Qb3

This is "Grob Opening: Grob Gambit, Fritz Gambit" or "Grob Gambit Accepted: Fritz Gambit" if anybody searches for it on Google. Wikipedia says "After 2.Bg2 Bxg4!?, White has an attack after 3.c4 and eventually Qb3, aiming at the weakened squares d5 and b7, an attack that may well be worth a pawn."

[d]rn1qkbnr/pp2pppp/2p5/3p4/2P3b1/1Q6/PP1PPPBP/RNB1K1NR b KQkq - 1 4

[Ovyron]

And our moves might be somewhat forced for a while...

1. g4 d5 2. Bg2 Bxg4 3. c4 c6 4. Qb3 e6

[d]rn1qkbnr/pp3ppp/2p1p3/3p4/2P3b1/1Q6/PP1PPPBP/RNB1K1NR w KQkq -

[mmt]

I will analyze it for some hours. Stockfish thinks that some other options besides 5. Qxb7 are not awful so I'll guide LC0 towards checking them. Specifically, 5. cxd5 and 5. h3 with 6. Qxb7 can use a closer look.