A french defence game

[Lyudmil Tsvetkov]

so far, for the f4 line I have:

f4 b6 Bb5 Bd7 Bd3 Ba4 Nf3 a6(to change bishops via b5 with draw) Rb1 Nd7 0-0 c4 Be2 0-0 g4:

[d]r2q1rk1/3nnppp/pp2p3/3pP3/b1pP1PP1/P1P2N2/2P1B2P/1RBQ1RK1 b - g3 0 9

and white has almost won

any suggestions for improvements for black?

I think I can draw this line as well. So again, I suspect the line is not busted and that a high white eval is not warranted. However, after white's Nf3, I'd play Nbc6

[d]r2qk2r/p3nppp/1pn1p3/2ppP3/b2P1P2/P1PB1N2/2P3PP/R1BQK2R w KQkq - 0 11

Your turn.

you can draw nothing, I have checked this thousands times; have played thousand games with SF and Komodo, and have substantially analysed it.

that said, I have analysed only 90% of the lines, some 10% always tend to escape my attention, but I think this is just about natural.

you can not know and prove everything.

so far, you have not drawn even a single game against me.

if you have been looking at the games played with Louis, in both games white has the advantage, but I played some weak moves in the first game, that is why the advantage slipped away.
that does not mean the line is bad for white, just that I played some average moves.

the second game, with the f4 line is still not finished, white has the advantage there as well, no matter what SF will show.

when I started this, I stated the Winaver is lost, at least in some 90% of all lines, and I can easily prove that.
for example, black castling short is lost, and that is how some 80% of human games are played, so I am very correct on that.

Louis/Botvinnik's line with castling long gives best saving chances, I never explored it till the end, might be draw with perfect play, but I am not certain, white still retains quite some advantage.

Louis' SF played 2 excellent/over-outstanding moves, f5 and 0-0-0, without any of those 2 black is simply busted.

sorry, but will post some lines only later, very busy now, not clear-headed, so my first move might be a mistake.

if you want to repeat Louis line, with f5 and castling long, then it does not quite make sense to repeat it, but, if you will castle short, no matter if f5 is played or not, I can certainly demonstrate a very convincing white win.

[Lyudmil Tsvetkov]

so far, for the f4 line I have:

f4 b6 Bb5 Bd7 Bd3 Ba4 Nf3 a6(to change bishops via b5 with draw) Rb1 Nd7 0-0 c4 Be2 0-0 g4:

[d]r2q1rk1/3nnppp/pp2p3/3pP3/b1pP1PP1/P1P2N2/2P1B2P/1RBQ1RK1 b - g3 0 9

and white has almost won

any suggestions for improvements for black?

I think I can draw this line as well. So again, I suspect the line is not busted and that a high white eval is not warranted. However, after white's Nf3, I'd play Nbc6

[d]r2qk2r/p3nppp/1pn1p3/2ppP3/b2P1P2/P1PB1N2/2P3PP/R1BQK2R w KQkq - 0 11

Your turn.

0-0:

[d]r2qk2r/p3nppp/1pn1p3/2ppP3/b2P1P2/P1PB1N2/2P3PP/R1BQ1RK1 b kq - 0 1

if you castle long, we are transposing to the game with Louis, and it does not matter to analyse another one like this, and apart from this it is not even finished.

if you castle subsequently short, then I will quickly demonstrate how white wins.

anyway, 80% of games played with this variation feature short castles.

My SF also wants to castle short at first, but maybe VLTC makes it to change its mind.

[Werewolf]

So is there any antidote to the simple:

7.f4 Qa5 8.Bd2 cd 9.cd Qa4 10.Nf3 b6 11.c4 Qd1 12.Rd1 dc 13.Bc4 Bb7 and black is fine I think.

there are many antidotes, white can play c3 in stead of c4, with clearly better game, or even earlier Nf3 instead of Bd2, etc., but I can not play multiple games at the same time.

will be waiting for Louis to return in half a light year or so, to finish the game, and then we will see what happens.

Take your time, truth matters more than speed.

In the line: 7.f4 Qa5 8.Bd2 cd 9.cd Qa4 10.Nf3 b6 11.c3 Qd1

I don't see any advantage for white. After 12.Kd1 (or Rd1) Ba6!
Swapping bishops.

White is left with:
- a bad bishop
- weak light squares
- backward pawn on a half open file
Though the position is so closed it's probably =

8.Nf3 is interesting. But black is ok after 8...cd

[mbabigian]

so far, for the f4 line I have:

f4 b6 Bb5 Bd7 Bd3 Ba4 Nf3 a6(to change bishops via b5 with draw) Rb1 Nd7 0-0 c4 Be2 0-0 g4:

[d]r2q1rk1/3nnppp/pp2p3/3pP3/b1pP1PP1/P1P2N2/2P1B2P/1RBQ1RK1 b - g3 0 9

and white has almost won

any suggestions for improvements for black?

I think I can draw this line as well. So again, I suspect the line is not busted and that a high white eval is not warranted. However, after white's Nf3, I'd play Nbc6

[d]r2qk2r/p3nppp/1pn1p3/2ppP3/b2P1P2/P1PB1N2/2P3PP/R1BQK2R w KQkq - 0 11

Your turn.

0-0:

[d]r2qk2r/p3nppp/1pn1p3/2ppP3/b2P1P2/P1PB1N2/2P3PP/R1BQ1RK1 b kq - 0 1

if you castle long, we are transposing to the game with Louis, and it does not matter to analyse another one like this, and apart from this it is not even finished.

if you castle subsequently short, then I will quickly demonstrate how white wins.

anyway, 80% of games played with this variation feature short castles.

My SF also wants to castle short at first, but maybe VLTC makes it to change its mind.

If we've been miscommunicating, I apologize, but I'm not concerned if the Winawer is trappy, sharp or dangerous to play for black - only if it is lost. I agreed on the outset after looking at it, that 1) engines don't understand the threats, 2) often fail to see pawn storms, and 3) it looks like an unnecessarily dangerous line for black to play.

There are many lines that are sharp and must be played carefully for one side to avoid loss early on. Even if all lines except one are lost, the correct evaluation is still 0.00. You stated that you thought engines misevaluate the position and it should be 0.8 for white. Engines don't concern themselves with how many bad lines they find, only whether their best line works and how that line evaluates. After 54 quad CPU days of analysis since we started discussing the Winawer, I see no reason to doubt the near equal evaluation of white and black's positions.

The IDeA system has now analyzed more than 100K positions at between (1 and 6min/move) in this line, and some of those lines contain O-O-O, some do not, depending on white's play. That's chess!

The biggest flaw found in the original line you gave above was 10 ... a6 which seeds the advantage to white. Nbc6, Qc7, c4, and Qd7 seem to all maintain the balance.

I can't tell you for sure that after a few moves I won't castle queenside as I don't know your moves. As I said, some lines contain queenside castling, some don't.

Again as of right now I see no evidence that black cannot hold in the Winawer or that it should be evaluated 0.8 for white.

That being said, my next move would be Qc7 should you wish to continue. The database of data suggests you have 7 reasonable responses. With scores between -0.02 to -0.09.

Mike
P.S. I have not been reading the other people's posts in this thread. It is too big and watching engine games in a line engines don't understand wouldn't interest me even if someone found a line by chance that doesn't lose.

[Werewolf]

If we've been miscommunicating, I apologize, but I'm not concerned if the Winawer is trappy, sharp or dangerous to play for black - only if it is lost.

For over-the-board chess though, this is quite important. It's highly unlikely something as well-known and main line as the French Winawer will be LOST. But if 90% of lines lead to a black loss, that's a big deal for human chess.

At the moment, I just can't see black is in any trouble.

[mbabigian]

Exactly

[Lyudmil Tsvetkov]

so far, for the f4 line I have:

f4 b6 Bb5 Bd7 Bd3 Ba4 Nf3 a6(to change bishops via b5 with draw) Rb1 Nd7 0-0 c4 Be2 0-0 g4:

[d]r2q1rk1/3nnppp/pp2p3/3pP3/b1pP1PP1/P1P2N2/2P1B2P/1RBQ1RK1 b - g3 0 9

and white has almost won

any suggestions for improvements for black?

I think I can draw this line as well. So again, I suspect the line is not busted and that a high white eval is not warranted. However, after white's Nf3, I'd play Nbc6

[d]r2qk2r/p3nppp/1pn1p3/2ppP3/b2P1P2/P1PB1N2/2P3PP/R1BQK2R w KQkq - 0 11

Your turn.

0-0:

[d]r2qk2r/p3nppp/1pn1p3/2ppP3/b2P1P2/P1PB1N2/2P3PP/R1BQ1RK1 b kq - 0 1

if you castle long, we are transposing to the game with Louis, and it does not matter to analyse another one like this, and apart from this it is not even finished.

if you castle subsequently short, then I will quickly demonstrate how white wins.

anyway, 80% of games played with this variation feature short castles.

My SF also wants to castle short at first, but maybe VLTC makes it to change its mind.

If we've been miscommunicating, I apologize, but I'm not concerned if the Winawer is trappy, sharp or dangerous to play for black - only if it is lost. I agreed on the outset after looking at it, that 1) engines don't understand the threats, 2) often fail to see pawn storms, and 3) it looks like an unnecessarily dangerous line for black to play.

There are many lines that are sharp and must be played carefully for one side to avoid loss early on. Even if all lines except one are lost, the correct evaluation is still 0.00. You stated that you thought engines misevaluate the position and it should be 0.8 for white. Engines don't concern themselves with how many bad lines they find, only whether their best line works and how that line evaluates. After 54 quad CPU days of analysis since we started discussing the Winawer, I see no reason to doubt the near equal evaluation of white and black's positions.

The IDeA system has now analyzed more than 100K positions at between (1 and 6min/move) in this line, and some of those lines contain O-O-O, some do not, depending on white's play. That's chess!

The biggest flaw found in the original line you gave above was 10 ... a6 which seeds the advantage to white. Nbc6, Qc7, c4, and Qd7 seem to all maintain the balance.

I can't tell you for sure that after a few moves I won't castle queenside as I don't know your moves. As I said, some lines contain queenside castling, some don't.

Again as of right now I see no evidence that black cannot hold in the Winawer or that it should be evaluated 0.8 for white.

That being said, my next move would be Qc7 should you wish to continue. The database of data suggests you have 7 reasonable responses. With scores between -0.02 to -0.09.

Mike
P.S. I have not been reading the other people's posts in this thread. It is too big and watching engine games in a line engines don't understand wouldn't interest me even if someone found a line by chance that doesn't lose.

I already told you, in the first post I posted I stated that I am certain 90% of lines are winning for white, and I abide by that.

white wins all lines where black castles short, and usually, in engine as well as human games, that would be over 80%.

perfect play might still be a draw, but with a very narrow path for black: necessarily castling long and playing f7-f5.

let us finish the game with Louis to see how it goes.

for black should be losing:
- castling short without f5
- castling short with f5
-castling long without f5, etc.

so, only castling long + f5 might be holding, but I am not certain of that either, white might have missed better moves.
in any case, on the vital moves, Louis spent some 5 hours or so analysis.

if you want to continue, play immediately 0-0, or a bit later, and I will show you how white wins.

after Qc7, black will probably castle long.

[Lyudmil Tsvetkov]

If we've been miscommunicating, I apologize, but I'm not concerned if the Winawer is trappy, sharp or dangerous to play for black - only if it is lost.

For over-the-board chess though, this is quite important. It's highly unlikely something as well-known and main line as the French Winawer will be LOST. But if 90% of lines lead to a black loss, that's a big deal for human chess.

At the moment, I just can't see black is in any trouble.

human databases give 41 vs 30 % white wins, that should tell something.

this is well beyond the average opening success for white, counting Sicilians, Ruy Lopez and any other system around.

play short castle, and I will show you how white wins.

PS. sorry that I can not pay more attention to you, but am simply currently doing too many distracting things.

[zullil]

Ng2:

[d]5qrr/pk2n1p1/1pn1p2p/3pPpP1/b1pP1P2/P1P4P/2PQBBNR/6RK b - - 0 3

OK, back to the game. Gave asmFish a long search to "remember" what was going on before the break.

27... Ng6
[D] 5qrr/pk4p1/1pn1p1np/3pPpP1/b1pP1P2/P1P4P/2PQBBNR/6RK w - - 7 28

Code: Select all

-0.38 27... Ng6 28. Qc1 Nb8 29. Bf3 Na6 30. Nh4 Nxh4 31. Bxh4 Qe7 32. Kg2 Be8 33. Qb2 Nb8 34. Kf2 Nc6 35. a4 Bg6 36. Ra1 Ka8 37. Qa3 Qf7 38. gxh6 Rxh6 39. Bg5 Rhh8 40. Qd6 Rc8 41. Rhh1 Qe8 42. Rab1 Rb8 43. Kg3 Bh5 44. Bg2 Rb7 45. Rhe1 Bf7 46. Ra1 Qc8 47. Qa3 Rd7 48. Rab1 Na5 49. Kf2 Rc7 50. Ra1 Be8 51. Qd6 Bg6 52. Kg1 Bh5 53. Qa3 Nc6 54. Qd6 Rf7 55. Qa3 Qe8 56. Qd6 Rb7 57. Bh4 Bg6 58. Bg5 (depth 59, 12:12:00)

-0.34 27... Nb8 28. Nh4 Na6 29. g6 h5 30. Rhg2 Rh6 31. Bf3 Nc6 32. Qc1 Rgh8 33. Be3 Nc7 34. Qb2 Nb5 35. Ra1 Rg8 36. Rg5 Qd8 37. Rb1 a6 38. Kh2 Qd7 39. Bf2 Rgh8 40. Ra1 Ne7 41. Rb1 Rb8 42. Rbg1 Rg8 43. Rb1 Rgh8 44. Rg2 Nc6 45. Rg5 Na5 46. Kg1 Ra8 47. Rc1 Nc6 48. Kh2 Rc8 49. Rb1 Ne7 50. Kg1 Ra8 51. Kh2 Rd8 52. Rgg1 Nc6 53. Rg5 Qe8 54. Rg2 Qe7 55. Rg5 Qxa3 56. Qxa3 Nxa3 (depth 59, 12:12:00)

-0.34 27... Kc8 28. Qc1 Nb8 29. Nh4 Na6 30. g6 h5 31. Rhg2 Qe8 32. Nf3 Rh6 33. Nh4 Kb7 34. Qb2 Nc7 35. Qc1 Bb5 36. Qb2 Rgh8 37. Bf3 Nc6 38. Rc1 Ba4 39. Rg5 Nb5 40. Rb1 Qd7 41. Kh2 Rc8 42. Kg1 Rd8 43. Kf1 Rdh8 44. Rg2 Nc7 45. Kg1 Qe8 46. Rg5 Ne7 47. Kh2 Rg8 48. Rgg1 Nc6 49. Qa2 Qf8 50. Qb2 Rgh8 51. Rg5 Qe7 52. Kg1 Nb5 53. Ra1 Re8 54. Kh1 Qd7 55. Rb1 Nc7 56. Kh2 Reh8 (depth 59, 12:12:00)

[Lyudmil Tsvetkov]

Ng2:

[d]5qrr/pk2n1p1/1pn1p2p/3pPpP1/b1pP1P2/P1P4P/2PQBBNR/6RK b - - 0 3

OK, back to the game. Gave asmFish a long search to "remember" what was going on before the break.

27... Ng6
[D] 5qrr/pk4p1/1pn1p1np/3pPpP1/b1pP1P2/P1P4P/2PQBBNR/6RK w - - 7 28

Code: Select all

-0.38 27... Ng6 28. Qc1 Nb8 29. Bf3 Na6 30. Nh4 Nxh4 31. Bxh4 Qe7 32. Kg2 Be8 33. Qb2 Nb8 34. Kf2 Nc6 35. a4 Bg6 36. Ra1 Ka8 37. Qa3 Qf7 38. gxh6 Rxh6 39. Bg5 Rhh8 40. Qd6 Rc8 41. Rhh1 Qe8 42. Rab1 Rb8 43. Kg3 Bh5 44. Bg2 Rb7 45. Rhe1 Bf7 46. Ra1 Qc8 47. Qa3 Rd7 48. Rab1 Na5 49. Kf2 Rc7 50. Ra1 Be8 51. Qd6 Bg6 52. Kg1 Bh5 53. Qa3 Nc6 54. Qd6 Rf7 55. Qa3 Qe8 56. Qd6 Rb7 57. Bh4 Bg6 58. Bg5 (depth 59, 12:12:00)

-0.34 27... Nb8 28. Nh4 Na6 29. g6 h5 30. Rhg2 Rh6 31. Bf3 Nc6 32. Qc1 Rgh8 33. Be3 Nc7 34. Qb2 Nb5 35. Ra1 Rg8 36. Rg5 Qd8 37. Rb1 a6 38. Kh2 Qd7 39. Bf2 Rgh8 40. Ra1 Ne7 41. Rb1 Rb8 42. Rbg1 Rg8 43. Rb1 Rgh8 44. Rg2 Nc6 45. Rg5 Na5 46. Kg1 Ra8 47. Rc1 Nc6 48. Kh2 Rc8 49. Rb1 Ne7 50. Kg1 Ra8 51. Kh2 Rd8 52. Rgg1 Nc6 53. Rg5 Qe8 54. Rg2 Qe7 55. Rg5 Qxa3 56. Qxa3 Nxa3 (depth 59, 12:12:00)

-0.34 27... Kc8 28. Qc1 Nb8 29. Nh4 Na6 30. g6 h5 31. Rhg2 Qe8 32. Nf3 Rh6 33. Nh4 Kb7 34. Qb2 Nc7 35. Qc1 Bb5 36. Qb2 Rgh8 37. Bf3 Nc6 38. Rc1 Ba4 39. Rg5 Nb5 40. Rb1 Qd7 41. Kh2 Rc8 42. Kg1 Rd8 43. Kf1 Rdh8 44. Rg2 Nc7 45. Kg1 Qe8 46. Rg5 Ne7 47. Kh2 Rg8 48. Rgg1 Nc6 49. Qa2 Qf8 50. Qb2 Rgh8 51. Rg5 Qe7 52. Kg1 Nb5 53. Ra1 Re8 54. Kh1 Qd7 55. Rb1 Nc7 56. Kh2 Reh8 (depth 59, 12:12:00)

where have you gone Louis?

I thought you were on a visit to another galaxy.

Qc1:

[d]5qrr/pk4p1/1pn1p1np/3pPpP1/b1pP1P2/P1P4P/2P1BBNR/2Q3RK b - - 0 1