Monday, July 25, 2005
Kotov's Tree
Everyone knows that getting better at chess is one of the great "loves" of my life. Or rather, one of the sickness of my soul? Well, anyway you look at it, its something that compels me to do "things" to myself.

What I do mostly changes like a politicians "stand" on major issues. A few years ago, it was studying positional play. The past six months, it was the seven circles. And now? I have already constructed a tactics-centered plan awhile back - but has experimented with various things since. I felt that I did not properly understand what my problem is, and that I should explore first my options before commiting myself to another grand scheme.

I had been playing two-three games a day, and analysing my play. Not specific variations - but rather determining the holes in my game. Like, why do I choose the moves that I do. How do I choose my plan? Things like that. Trying to determine the most damaging element of my game.

It was very instructive to see my timesheet (FICS automatically records the time spent each move). The average lenght of time I used to determine a move was 30sec. 30 seconds? I didnt know I can think that fast. Chess is supposed to be a thinking game, and this shows that I had not been thinking enough. Clearly I need to medidate more on my moves. But what to think, and how do I "think"? Is there a correct way to do it?

Enter Kotov's tree
(Creepy sound ala star wars right here)

Yeah, I know - this type of thinking process has been "refuted" over and over again. By no less than Mr. Nunn, Mr Tisdall, Mr. Soltis, etc.. etc.. "I dont think like a tree, do you think like a tree" and all that stuff. But as with the seven circles - the goal is not to find the "best" thinking process or even the most efficient. But rather to just have a good process and master it by sticking at it.

But I think the implementation of Kotov's tree of anaylysis, would greatly help me consistently make good moves. I think, the only way to prove a move is good - is by comparing it with other moves. Only by comparison can we say something is better than something. Only by comparison can we go and say "Ah! this move is good". Anyway ,Here is the mechanics of my tree-of-analysis

1. Every move, select two candidate moves:
It is better to have three or maybe four?! but right now, asking that from myself is impossible. I'll start with two moves first and later on widen the search when I am more capable.

It is important to select the moves first and do not calculate.

2. Calculate the resulting lines to death:
After selecting the most promising two moves, we need to calculate to see which is better. Calculate until all forcing moves are exhausted. Or if there are no forcing moves - until I see a position I'm comfortable with. Maybe I see a slight space, or an outpost can be created.

It is important when calculating that we do not jump from one line to the next. Calculate until a definite evaluation is reached. I know when a definite evaluation is reached when my mind says something like this:

"This is an ok move - I dont see any way he can win material"
"This is an ok move - The position is playable"
"This is a bad move, I lose here"

This shows that I consider almost all moves to be good if I dont lose material or reach an obvious positional deficiency. All positions look playable to me =<. I dont think I can determine who is "slightly" better or worse.

Anyway, after I reach a definite conclusion, I can forget all those analysis and remember only if the the move is good or bad. In fact - I should forget the analysis and remember only the conclusion. Why? because I found out that trying to remember a variation while calculating another, causes the calculation to run much slower. My brain is not multi-threading so to speak.

Now, repeat for the next move and reach another conclusion. Compare the two conclusions. If one is bad, then the choice is easy. If both of them are good - just pick one randomly. They're both good anyway. Or according to what my gut tells me.

Gee - my kotov tree sounds easy. Where is the snag? What's the catch? For starters, calculating is really tiring. Consistently doing it everymove is hard. The other day, I was playing against a 1962 player - and I had the better position. But somewhere along move 35 or so - I got tired and made a "just" move. Instantly he equalized, and with great difficulty I had to scramble for a draw.

Also - I simply forget to do it. Sometimes, the position looks too obvious that I simple do the natural moves. But if I do that - then I start moving automatically the next move - faster and faster until I reach blitz speed. A more consistent approach would be the ideal to aim for.

Lastly, calculating variations aint all that easy. The seven circles have really helped me, but it is not enough. My calculation needs to be a little bit faster, a little bit deeper before I become comfortable with it.

Training Plan
So now that we know what I want to do, I have to train using it. Of course you dont just wake up, and decide to use the kotov tree cold-turkey, and expect to be successfull. I wasnt born a mean chess calculator, so i'm gonna have to work to get this ability. And, since i'm using the the Kotov tree - Might as well use the Kotov-recommended training plan. That is, get a game, go to the middlegame, and start calculating from there. Plus supplement with endgame studies for visualization. This two will keep me very busy for the next three months or so I think.

Well, this is the current plan. Now if it survives practice, is another matter altogether =>
posted by Nezha at 12:40 AM | Permalink |


  • At 4:10 AM, Blogger JavaManIssa

    looks to me like a thought process ;)

    I also tend to be a stronger player if just calculate a little bit longer, a little bit deeper. If you do tactics on CTS, try calculating 'a little further' than usual. Assuming you can calculate reasonably fast, you'll see it climb :) (mind you i don't calculate fast)

  • At 11:19 AM, Blogger CelticDeath

    I usually have the opposite problem. I take too long trying to come up with the ideal move and end up in severe time trouble. Often this means that I win the middlegame, only to blow it and lose in the middlegame/endgame transition. What this points to is opening training to a certain extent, and conquering my perfectionistic tendencies to perhaps an even greater extent.