This is a position from Shirov-Aronian, Game Two, with Shirov (White) to play. I looked at this briefly with Fritz 10 yesterday and noted that the software program strongly preferred 33 Ne4 to the move played, which was 33 Nc4.
This morning I read Mihail Marin's report on chessbase.com and was a bit surprised to find that he had skipped over this position without a comment.
So, was this a case of culpable negligence by Shirov (and Marin)? At first I thought so, but then looked closer. I also fed the position into a number of other chess engines (including Rybka 2.1, Hiarcs 10, Shredder 10). They all preferred 33 Ne4 to the move played. But then it dawned on me: if, after 33 Ne4, Black plays 33...Bxe4 34 Qxe4 Rd6, how can White ever win? Black is going to put his rook on e6, his king on g7 and then play Be7, Bf8, Be7 ad infinitum. Everything is defended. The white queen can go anywhere she wants but, without an attacking accomplice, cannot force a breakthrough with pawns or the king, and as far as I can see there is no potential zugzwang.
The finely-honed brains of Shirov and Marin would have figured this out pretty quickly, I guess. But it is a good example of software not being able to see the big picture. The human player sees the ideal placement of the black pieces to defend everything plus an adequate 'pass move' sequence to avoid zugzwang. Then a quick scan of the white side reveals no plausible breakthrough possibility.
It raises a general point, though, about annotators. They have a problematic task: they are good players who have to work out what weaker players will weant rexplaining even though it is may be obvious to them (which in itself brings the danger of talking down to people). Additionally, I think that now they will have to take into account moves that computers suggest - because people will be following games and analysis with the assistance of computers - even though the annotators themselves can see they are not particularly strong.
ReplyDeleteThis is another example of computers actually making life harder for chess professionals rather than easier. Which was the conclusion of my Masters thesis, I'll have you know.
Shirov lost with his chosen move. The computers reveal that problem for white is not how to win in this position, but how to draw.
ReplyDeleteIn some lines, White may be able to draw with Ne4. That, I gather, is why it is Rybka's chosen move. In this case I think the difference is a result of human, not computer, weakness.
I think that it is also a good example that annotators are explaining the positions in such a way that the reader can really learn from this annotations and use this knowledge in his own games. The software may be able to find a better move, but the ideas behind the move are somewhat hidden.
ReplyDeleteI think that plenty of humans, as well as computers, would have wondered why Shirov did not play Ne4, in view of the juicy prospect of getting the knight to f6, or alternatively getting rid of the B on b7. So maybe Marin should have mentioned it anyway, especially since Nc4 looks a bit tame and unshirovian. Ejh's point may be a good one but I'm not sure if this is a particularly good example.
ReplyDeleteRyan: the game ended in a draw, not a loss for White.
That thesis title in full:
ReplyDeleteMoving Too Fast?
Aspects of information overload and the study of opening theory in chess.
by Justin Horton
Submitted to the School of Information Studies at the University of Northumbria at Newcastle as part of the requirements for the MA in Information and Library Management
September 2001
I've been promising to write it up for Kingpin for six years now...