Monday, June 23, 2014

ISEs: From the Books II

White to play
Seirawan - Kozul, Wijk aan Zee 1991

Yasser’s got something truly absurd for us today. Swap queens, 'lose' the exchange and then just carry on playing.

I can kind of see the point - Black’s bishop ends up with no diagonals, the rooks have no files - but White’s pieces don’t seem to be doing anything either. Well, what do I know? Seirawan wins anyway, although he (and Watson: I found this position in Secrets of Modern Chess Strategy) thinks White can do even better a little while along the line with 25 Rf1.

2014 ISE Count: 43
TISE Index


Anonymous said...

Working to the narrative that Black shouldn't lose, I tested both the initial position after the exchange sacrifice and than subsequently to see where the evaluation changes. According to the engine, the initial position at move 22 favours Black, although not be very much. That continues to be the case until Black's breakout attempt by 32. .. f5. Black can just move his rooks around, as it's breakthroughs became possible with 34. c5 that he has to be wary of.

The engine would also exchange queens on move 19, but the boring way with Qe1 so as not to drop the exchange. This is assessed as in White's favour.

There's another potential theme for a series, games where one player made a sacrifice and won after inaccurate defence, but where the sacrifice wasn't the best available move.


Dave Ewart said...

Is this the sort of position where engines can evaluate it accurately, though? The two black pawns on e4 and e3 are far advanced, but are more of a liability than anything else. Also, this sacrifice is certainly one of a long-term positional nature: well beyond event horizon, unless the engines were analysing *very* deep.

Comment Moderator said...

I also wondered about the computer’s evaluation ability here. Possibly true of exchange sacs in general, but maybe particularly so with this one.

Anonymous said...

I was starting from the viewpoint of Black not trying to win. There's the breakthrough as seen in the game, but if Black does nothing in a constructive manner, what is White's plan?

Some engines are notorious for suggesting exchange sacrifices at the slightest opportunity. I expect it's how they are programmed but an evaluation of Rook = Minor piece + pawn + a half seems empirically likely. Thus if they can see "pawn and a half" as compensation, they go for it.

On the topic of engine evaluation, there's a line advocated in a recent book about the Tarrasch (against the Queens Gambit) where Black plays .. c4 and then allows a position where he has isolated pawns on c4 and c6 but a Bishop pair. Engines superficially think Black is OK, but I wonder if this is another horizon effect. As a result of the book, people are trying the line, so a consensus may emerge. To return to the topic, if you check the game Gormally-Holland from the 2012 British, the variation features an exchange sacrifice by Holland to reach a position where he was able to draw.