Friday, October 09, 2009

Look around, not ahead

It is amazing that nearly every game in the lower levels of chess are decided by people literally not seeing obvious interactions between pieces on the board. Leaving a piece open to be taken, or forgetting a piece is pinned, or even not seeing mate in one.

Yet many of us in these lower levels still insist on looking many moves ahead in silly lines. We aren't losing because we fail to look ahead enough moves. We are losing because we don't look ahead one or two moves in the correct lines. We look ahead six moves in the line that might get us a nice outpost, but don't see that our rook is about to be taken due to a one or two move tactic.

So, instead of looking ahead in a particular game tree, first relax and inspect the trunks of the different trees. Look around. See what is going on in the position, its most elementary tactical and strategic contours. What if your opponent was a beginner? What kinds of mistakes would he make? Perhaps he was kind and made such a mistake. You won't know if you don't look. Don't assume he knows what he is doing. Lord knows I never know what the hell I'm doing, but my opponent gives me too much credit and misses the little treats I leave him on the board.

This can be really hard to do when you feel "momentum" in your game, when you are caught up in plans you made four moves ago. But yesterday's plan can easily become today's disaster. 'Long analysis, wrong analysis' as they say. You can see the board in front of you now much better than you could visualize it four moves ago. Slow down, don't be like a hyperactive teenage boy shown a boobie for the first time. Keep it in your pants. Look around. Take stock of the situation in front of you. Don't let momentum push you about.

Note added: since we can't explore all of the game tree all the way to the end, we have to choose whether to cut our exploration short by not considering as many shallow trees, or by not considering as many branches on one tree. My argument here is that it is better to err on the side of considering too many trunks, rather than too many leaves. Of course, some extremely sharp lines are so promising they need to be visualized, so in really sharp positions you probably want to err on the side of thinking ahead in fewer lines (e.g., if he is about to mate you, then there may be only two or three lines worth considering). The quieter the positon, the more trees the better, but not deeply, just a quick walk around the forest.

This is already in my Chessplanner stuff (see blog highlights), but not in the same language. If I ever update it perhaps I should phrase things in terms of width versus depth, looking around versus looking ahead.

As an added bonus, below is my favorite new video game song from Portal. The lady singing is GLaDOS (Genetic Lifeform and Disk Operating System), who is your guide in the Portal world...This game is intellectually challenging, but unlike chess I was actually able to figure out how to beat it. If you have played the game, the song is hilarious. If you haven't I'm frankly not sure how it will come off.