Tuesday, April 22, 2008

Evaluation Calibration

About five years ago, I was analyzing a game with an expert and I remember suddenly noticing something about how he talked about the game. I had only been playing chess for about a dozen years and I’d read my share of chess books, but I didn’t own ChessBase and had all but ignored those += and -/+ evaluations like some extinct language invented by the Informant magazine. Somehow I had always concentrated on individual exclams and blunders and material advantage. The expert spent a lot of time verbalizing judgments like “White is better. Black is better. I’d rather be playing Black.” It’s not that I’d never heard of evaluations, it’s just that I spent a lot of time looking at individual trees rather than appraising the whole forest. After I got Fritz and ChessBase in 2003 and learned what +-, +/-, +=, =, =+, -/+, -+ mean, I began to pay more attention to how they are applied. It hit me that strong players have an evaluation function that is quite keen and frequently if not always being used.

Temposchlucker and Phaedrus are always writing about “visualization”, “motorskills”, and “patterns” as the main elements of chess strength. There is usually some discussion at the periphery about whether move selection method is an element or some meta-component. But I wonder whether evaluation function is another independent element. "Calculation" I interpret as moving the pieces around in your head, but at the end of a bunch of movement, you have to use judgment to figure out if the positions are good, better, or best. I would separate evaluation as a skill that has its own merits. I’ve been thinking about my own Ultimate Chess Scheme/Framework, with a comparative glossary of terms and some incorporation of Wetzell of course.

When I study my own games, I lazily use Fritz as a pocket grandmaster to tell me what happened in my game. I figure I thought about the game already when it was happening, so just show me not only all the moves I missed, but tell me as my gold standard “objective evaluator” how the game seesawed. I know Fritz has limitations, but in the absence of a grandmaster at my beck and call, it’s the best thing around. If the numbers don’t make sense, then I go down the variations to see why Fritz evaluates the position. Why does it think that Be3 is 8 centipawns better than Bb2?

Last Thursday’s game was another interesting game against chess blogger extraordinaire Wahrheit who is on a bad streak lately. I didn’t let him break the streak, but things were fairly dicey in the late opening/early middlegame transition. I basically became impatient and broke open the center without full justification, only a desire to mix things up. My queen bishop became a development problem, but my opponent helped me by centralizing my problematic knight and then allowing me to get the two bishops. I pressed in the center and suddenly I had a raucous kingside attack. My opponent had some counterchances around this time, but the evaluation is debatable.

Here is a diagram of the moment to test your calculation and evaluation:

The most accurate line here is 17…Qb6+! 18.Kh1! (18.Rf2 stops Nxc4, but allows the annoying pin +/-) 18…Nxc4 19.Qg4! (19.bxc4 Qxb2 20.Rb1 Qxa2 21.Rxb7 with compensation +=) 19…Nxb2 20.Qxe6+ Rf7 21.Ng5 Bxg5 22.fxg5+- with winning play on the pinned rook. If you can match Fritz, the line continues 22...Qc7 23.Rf3 Qe7 (23...Raf8? 24.g6!! hxg6 25.Raf1 Nd3 26.Rxf7 Rxf7 27.Qe8+ Kh7 28.Rxf7!+-) 24.Bh3 Rae8 25.Raf1 Qxe6 26.Bxe6 Rxe6 27.Rxf7+-. Whew! Ten moves and twenty ply.

I’m annoyed that I missed mate in one. All that time spent at Chess Tempo and I still miss a mate in 1?! My main excuses are 1) that at this point in the game, I had to put on my TD hat and mediate a flag fall. 2) I was also firmly latched onto the mate in 4. 3) Perhaps I never considered moving the back rook because it’s a security blanket against Black’s queen. Positions with mate in 1 should be easy to evaluate, but I'm heartened that even the world's best can rarely miss mate in 1.


Temposchlucker said...

Good point.
It's true I focussed on motorskills lately. But in the back of my head I was well aware that there are a few topics that aren't adressed yet. I haven't posted about calculation and evalution for the main reason that I have not thought about them in the light of my new findings. I will try to adress this in a future post.

Blue Devil Knight said...

Excellent stuff, an important part of thought process, I think, which tempo and phaedrus have tended to downplay (and which I tend to overplay :)).