Thursday 12 June 2008

The problem of assessment

One of the self-imposed disciplines when playing correspondence chess is that you cannot use computers to analyse the game. While I am (a) fully supportive of the rule and (b) believe not relying on computers makes you a better player, it does expose some bad habits that I (and I guess others) have.
One is in the area of opening analysis. If I play an OTB (over the board) game I'll often run it through the computer, paying special attention to the assessment of the first 'non-book' move. This is especially important if I have lost the game, as I may have made a mistake in the opening and do not wish to repeat it. In correspondence chess, you have to wait until the game is completely over (or at least I do), before putting the silicon brain to work on any position from the game. This may mean weeks (or months) between seeing an interesting opening idea, and checking if it is really sound. And if by the time you've finished the first game you are playing another game that uses the same opening, honour demands you wait until the second game is complete.
The other weakness it exposes is in assessing the position. How many times have you analysed a game with your opponent at the club and said "I might be better here, but I'm not sure. Fritz will tell me when I get home who is winning". I tend to do this a lot, and I'm sure this has made it harder for me to accurately analyse variations. Given enough time (such as in a CC game) I can usually find enough moves, and variations but often I reach the end of a line and am unsure of whether the position is better, equal or worse for me.
For example, in the diagrammed position I (as Black) have just sacrificed a couple of pieces, although I have got one back by capturing the White Knight on f3. When I had envisaged this position a few moves back I figured that I could move the Black Rook to h3 and achieve a winning attack. But once the position arose on the board I saw that White could play f3 in reply, defending the Queen and providing an escape route for the White King. So I looked at whole lot of forcing lines, without being able to decide whether they were playable or not.
15. ... Rh3 16.f3 Rxh2 17.fxg4 Rh1+ 18.Kf2 and now I've have a number of choices to look at. I looked at Rf8+, Qf7+ Rxf1+ and Qh4+. The latter leads to 18. ... Qh4+ 19.Ke3 Qg3+ 20.Qf3 Qxf3 21.Rxf3 Rxc1 By this stage I decided that I was just down a piece, White could eventually develop the b1 knight, and White is just winning. Fritz 8 on the other hand gives White a tiny advantage at best. I on the other hand only counted material and abandoned the line entirely.
Indeed I eventually discarded a number of moves, based on not seeing an advantage for me, and eventually chose the sub-optimal 15. ... h6 (to prevent tricks based on Bg5+). This is actually worse for me but my opponent didn't find the strongest follow up and I won 9 moves later.


TrueFiendish said...

I certainly bloody hope they're not using computers; otherwise I'm spending a lot of time and a fair bit of moolah playing against a machine.

Shaun Press said...

Based on my post-facto analysis on my CC games over the years, I'm pretty certain that my opponents weren't receiving compute assistance. In fact most of the "CC stands for Computer Cheating" type comments that I get come from players who don't actually play CC.

Anonymous said...

I think that many (most)CC Leagues draw a distinction between the use of chess engines for analysis which is usually prohibited (the only exception are leagues devoted to "advanced chess")and the use of chess literature, and chess databases, which is generally permitted. That is, you can't use Fritz but you can use Chessbase. That is why playing idiosyncratic or unusual openings is a good idea in CC games. It is why the most treasured possessions of players like Umansky are their (self-annotated) databases.

TrueFiendish said...

Shaun, yeah. The fact that I'm not losing suggests to me that, if people are using silicon assistance, it isn't very good. ;-)