A french defence game

[Lyudmil Tsvetkov]

btw., even online databases of human games show 40% white wins vs 29% black wins, so definitely much more difficult to play for black than the average solid black defence(for example the Sicilian and even Ruy Lopez).

[Lyudmil Tsvetkov]

well, if you are interested in the main line after 7...Qc7, it should involve something like the following: Ne2 b6 Bh3(this is also a possibility) Ba6 0-0 Nc6 Re1 Rc8 Nf4 0-0 Nh5:

[d]2r2rk1/p1q1nppp/bpn1p3/2ppP2N/3P4/P1P3PB/2P2P1P/R1BQR1K1 b - - 0 8

and I guess even the dullest of fishes should see here quite some white edge.

white continues with Ra2, to defend the c2 pawn, and launches a mighty kingside attack with f2-f4, etc.

of course, SF will never see that line, even after months of analysis, as it all the time wants to play other moves, which are much wekaer.

[Werewolf]

Hi Lyudmil,

I'm following this thread with interest, since the French is one opening I hate playing against.

After the line:

1.e4 e6 2.d4 d5 3.Nc3 Bb4 4.e5 c5 5.a3 Bxc3+ 6.bxc3 Ne7 7.Nf3

You're claiming an advantage for white. After 7...Nbc6, how would you show that? In some of your lines you have the plan g3 + Bh3 + Nh5 + f4 etc, though 7.Nf3 doesn't seem to lend itself well to that plan (how does the knight get to h5?)

Also, in just about all your lines, black castles - but does he have to?

[Werewolf]

that is absurd.

g3 getting a negative score and placing only 15th or so.

it is at least the 3rd-best move, maybe even better.

similarly for f4.

I understand what the program is doing, but the problem is, how you know when to input alternative moves to the engine, and when not?

inputting just couple of alternatives here and there will not be sufficient, as this line involves some 30 plies of accurate playing, where most probably you should correct the engine on each and every ply.

concerning email correspondence, thanks a lot for the suggestion, but I will pass, I am really quite busy at the moment.

we can definitely though continue here for a while with some of the most important issues.

I think your STATIC evaluations of these lines is very good. But the problem with the g3 plan is it's quite slow:

This is the line I've come up with (using IDeA) and it seems to be fine for black:

1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 c5 5. a3 Bxc3 6. bxc3 Ne7 7. g3 Qa5!?
8.Bd2 Qa4 and black has time to attack the centre - it's hard to organise Ne2 + Bh3 + o-o + Re1 + Nf4-h5 and f4

[mbabigian]

I would like nothing more than to prove you right! However, looking at games where black loses isn't a reasonable way to do that. Especially since black played 10 ... c4?

I propose we play out the Nf3 variation which you believe is strong. I will play the black side and have the system generate data as we go. In your example, the white score got over 0.40 in just 9 moves. I propose we set some limit, say 25 moves (you pick the move limit) for you to get my system to show greater than 0.40. After each move by you, I'll post what my system suggested for white (if it differed from your move) based on positions it had time to analyze. If you are interested, I play 7 ... h6 with a current score of 0.00.

[pgn]
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]

1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 c5 5. a3
Bxc3+ 6. bxc3 Ne7 7. Nf3 h6 *
[/pgn]

Your turn!

[Lyudmil Tsvetkov]

Hi Lyudmil,

I'm following this thread with interest, since the French is one opening I hate playing against.

After the line:

1.e4 e6 2.d4 d5 3.Nc3 Bb4 4.e5 c5 5.a3 Bxc3+ 6.bxc3 Ne7 7.Nf3

You're claiming an advantage for white. After 7...Nbc6, how would you show that? In some of your lines you have the plan g3 + Bh3 + Nh5 + f4 etc, though 7.Nf3 doesn't seem to lend itself well to that plan (how does the knight get to h5?)

Also, in just about all your lines, black castles - but does he have to?

too many lines, Carl, do you have an optimal one for black, I could refute?

I am certain Qa5-a4 is a weak manoeuver, white should not have hard time refuting it.

Nf3 Nc6 supposedly should give white even larger edge.

currently, I have other more urgent things to do, but will be back with some output as soon as possible.

[Lyudmil Tsvetkov]

that is absurd.

g3 getting a negative score and placing only 15th or so.

it is at least the 3rd-best move, maybe even better.

similarly for f4.

I understand what the program is doing, but the problem is, how you know when to input alternative moves to the engine, and when not?

inputting just couple of alternatives here and there will not be sufficient, as this line involves some 30 plies of accurate playing, where most probably you should correct the engine on each and every ply.

concerning email correspondence, thanks a lot for the suggestion, but I will pass, I am really quite busy at the moment.

we can definitely though continue here for a while with some of the most important issues.

I think your STATIC evaluations of these lines is very good. But the problem with the g3 plan is it's quite slow:

This is the line I've come up with (using IDeA) and it seems to be fine for black:

1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 c5 5. a3 Bxc3 6. bxc3 Ne7 7. g3 Qa5!?
8.Bd2 Qa4 and black has time to attack the centre - it's hard to organise Ne2 + Bh3 + o-o + Re1 + Nf4-h5 and f4

idea does not work, time to accept that.

will post later some feedback.

[Lyudmil Tsvetkov]

I would like nothing more than to prove you right! However, looking at games where black loses isn't a reasonable way to do that. Especially since black played 10 ... c4?

I propose we play out the Nf3 variation which you believe is strong. I will play the black side and have the system generate data as we go. In your example, the white score got over 0.40 in just 9 moves. I propose we set some limit, say 25 moves (you pick the move limit) for you to get my system to show greater than 0.40. After each move by you, I'll post what my system suggested for white (if it differed from your move) based on positions it had time to analyze. If you are interested, I play 7 ... h6 with a current score of 0.00.

[pgn]
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]

1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 c5 5. a3
Bxc3+ 6. bxc3 Ne7 7. Nf3 h6 *
[/pgn]

Your turn!

yeah, STC games/shootouts often present a distorted picture, but are still indicative; the engine makes mistakes for both sides, after all.

I am not certain if f4 or Nf3 is stronger and how big an advantage g3 gives.

all I know all 3 lines give white quite some edge.

I would be very happy to take your challenge, my problem is I don't quite have the time at the moment for a full consistently-attended-to game, all I can do is give some feedback at random.

maybe, if you give longer variations for differnt lines, I can suggest where white could improve.

[mbabigian]

Unfortunately improving the Winawer wasn't high on my priority list for the machines. That would take a long time. Finding a line that holds isn't as much effort.

Perhaps when you have the time we could give it a go. I'm fine with playing after f4 also.

Anyway, your thoughts and insights were very illuminating.

Thanks,
Mike

[Lyudmil Tsvetkov]

that is absurd.

g3 getting a negative score and placing only 15th or so.

it is at least the 3rd-best move, maybe even better.

similarly for f4.

I understand what the program is doing, but the problem is, how you know when to input alternative moves to the engine, and when not?

inputting just couple of alternatives here and there will not be sufficient, as this line involves some 30 plies of accurate playing, where most probably you should correct the engine on each and every ply.

concerning email correspondence, thanks a lot for the suggestion, but I will pass, I am really quite busy at the moment.

we can definitely though continue here for a while with some of the most important issues.

I think your STATIC evaluations of these lines is very good. But the problem with the g3 plan is it's quite slow:

This is the line I've come up with (using IDeA) and it seems to be fine for black:

1. e4 e6 2. d4 d5 3. Nc3 Bb4 4. e5 c5 5. a3 Bxc3 6. bxc3 Ne7 7. g3 Qa5!?
8.Bd2 Qa4 and black has time to attack the centre - it's hard to organise Ne2 + Bh3 + o-o + Re1 + Nf4-h5 and f4

your plan with Qa5 and Qa4 does not help at all.

white continues with Qb1:

[d]rnb1k2r/pp2nppp/4p3/2ppP3/q2P4/P1P3P1/2PB1P1P/RQ2KBNR b KQkq - 0 4

threatening Bb5 and Qb3 change(after Qb3 cb3 endgame should be won for white), and proceeding further with kingside attack.