I just completed the following game. It is a French Defense (Advance Var.) against a somewhat higher-rated opponent. We agreed to a draw in a position that initial computer analysis says is a win for Black. I played White. My draw offer (which Black accepted) was not frivolous, despite being down the exchange. I'm pretty sure that White can hold in this position. But what do you think?

I've already lightly annotated it. Most of the later annotations are from the Chess.com analysis computer. I'm interested in others opinions of the final position, as well as of the moves surrounding the loss of the exchange. And of the whole game, actually.

Hrm. That last line got nuked. But you get the drift.

Looks like a dead draw to me.  White can just move the bishop back and forth between f2 and e1, and black will never be able to make any headway at all.  Black's king need to stay near e6 to keep white's e-pawn blockaded, so ...Kd8-c7-c6 wouldn't help black either.

What is winning? +3.00 on the computer?  Black is better. He might try infiltrating through b5 since White's King must watch the passed dpawn but if the bishop sets up on d4 and the king keeps watch on b4 making progress will be nigh impossible for black.

huh i think after 42 ke6 kxe6 looks strong

Hah. :P

For some reason, when the PGN was pasted in, it dropped a move, thus teleporting White's king to his doom. :)

After 37.Bc3, can't black play Rg8 38.Be1 Rg6 39,Bf2 Rh6 40.Bg1 Rh3? I think he can win by a kind of zugzwang.

In the final position, I'm hoping that some masters can weigh in and give us the scoop. I feel as if Black should be winning, but, like everyone else, I can't see the method. Any masters out there?

But Black should have remembered Silman's advice to never miss the opportunity to torture your opponent for many moves. Based on that, Black shouldn't have taken the draw

Computers evaluate badly these 'impenetrable' positions.

Playing on endlessly in a dead drawn position would be torture indeed, for both players perhaps.

But after 40.Bg1 Rh3 the B is tied to the h-pawn and can't move. For example, 41.Ke3 h5 42.Kd4 h4 43.gxh4 Rxh4 I think Black is making progress. And if the White K ever retreats too far, Black has d4! and the K can go after the c-pawn

But I'm not sure that we've established that it is a dead draw. True, I can't see a plan for Black, but I'm not that strong either

Computers are notoriously bad when it comes to evaluating 'impenetrable' positions because the '50 move draw' rule is always outside of the search horizon. Therefore if they can shuffle pieces back and forth while avoiding 3-fold repetitions they will think the value of the position can be determined by looking at the usual quality of pieces, the balance of material etc..

The game below is a striking example. The position is dead drawn, buy Fritz 11 was adamant that I was winning. In fact I wasn't until black allowed my pawn to escape to promotion!

That's why you use endgame tablebases for fritz. If you do, it will tell you that position is dead draw in a microsecond.

Agree with the "drawn" assessment that others have here as well.

Computers often cannot see past the fact that Black has a material advantage of rook vs. bishop. But in an ending especially, it is important that one of those becomes "the better piece" over the other. Most often (as we know) it is the rook that becomes (or starts out as) the better piece against a bishop, but in this case, neither piece is any good because neither has any true scope. The bishop is locked in by pawns of its own color, and the rook has no way to infiltrate and get behind the White pawns, which is the only way it can be effective.

Essentially, the pieces are non-existent and you have nothing but locked pawns. Therefore, a draw.

Interesting post!

Oh I see. paulgottlieb isn't looking at the final position that the original poster was asking about (which is a dead draw), but rather going back quite a few moves to fix black's blunder of 37...h5?  That does seem to be the move that throws away the win.

[COMMENT DELETED]

I'm aware of table bases, and you are right in this case because there are less than 7 pieces (including pawns) remaining..

The 7 piece tablebases are already very sizable, and I doubt most people would want to install an 8 piece one when it is ready.

My point was that the engine itself can't see the position is drawn, because of the reasons laid out by myself and others. Now if it can tap into a database that tells it the position is a draw, then it knows yes. But in the game discussed there's too many pieces to use such a solution.

Disregard, I need to look at this some more.

Clearly I have much to learn.