Tuesday, September 25, 2007

Which nominee will win?

Once you've got your candidate moves in mind, you need to select the one you will actually play. Soltis, in How to choose a chess move, identifies four strategies people tend to use, and the strengths and weaknesses of each one:

1. Prioritizing: focus on one particular move that catches your fancy, and analyze it thoroughly to make sure it is playable. If you find a flaw, go to the next candidate. Soltis says GMs often use this method (indeed, Tisdall says it is the best method as it is what our minds spontaneously want to do: explore the candidate that intuitively seems best). GMs can get away with it because their intuition is so damned good. On the down side, after spending a lot of time on one move, you might become irrationally committed to it, and if it turns out to be a bad move, it has stolen thinking time that you should have spent on the other moves.

2. Thinking like a Kotov: analyze each candidate in turn, examining each move tree only once, and after this is done, pick the candidate that yields the best consequences. The upside is that it is thorough, and you aren't as likely to miss a sneaky great move as you are with Prioritizing. On the down side, it is unrealistic. Nobody analyzes a line only once: at the very least, in crucial positions you need to recheck your analysis. Also, most people don't deeply analyze every candidate move. How many need to be analyzed deeply depends on the position.

3. Elimination: Try to find the biggest weakness in each candidate move, and select the move with the smallest weakness. That is, via process of elimination, whittle down to the least objectionable move. This is very helpful when you need to defend, and instead of being used exclusively, can be used to trim down the number of candidates down to a Kotov-able size. There is a danger of prematurely rejecting a candidate, so be careful. I really like this strategy: the least bad move is the best move!

4. Back and forth: Analyze one candidate a few moves deep. Then another. And another. Kotov shoveled derision on such a thought process as inefficient, but it does have its advantages. First, an analysis of one move can reveal features of the position that could come in handy for one of the other candidates. It would be strange to be stubborn and refuse to go back and reanalyze a move based on such new information. Also, since you are most likely to see tactics early in your thinking process, a quick search of all moves may reveal tactics you might not see if you already have spent 10 minutes on your first candidate (Soltis calls this tactical fatigue).

I use a mixture of all these methods. I think this is unavoidable, as they are not mutually exclusive but can all play a role in selecting a move. I usually start by doing a quick tactical scan before tactical fatigue sets in, and if something particularly promising pops out, I spontaneously will think it through to its conclusion (Prioritizing). The result of that analysis then becomes the basis of comparison for the other candidates. I am not very good at thinking like a Kotov, so reserve such effort for those cases where there are a couple of moves I can't decide between, and I have some reason to think one of them may be significantly better than the other (e.g., it leads to sharp play). If the position is quiet, and a superficial analysis doesn't reveal striking differences, I decide based on general principles and save my thinking time for sharp positions.

That's all the ideal. A more common thought process is "Three moves ago I planned on making this move now, so I'll just make it." [move] "Oh, crap, I just hung my knight. I didn't visualize the position very well three moves ago. I should have used Fine's modified rule."

Anyone out there think like a Kotov?


Blogger Tynicas said...

I think I tend to use a combination method as well, BDK. In an average middlegame position after I've pulled up my candidate moves, I'll often see one or two moves I'm quite positive about (intuitive and prioritizing), but I'll save those for later analysis, first plowing through checks, captures, and threats that didn't jump out at me, often ditching them quickly for tactical reasons but saving more playable ones(elimination). Then I'll Kotov my way through my saved short list by evaluating the first line, comparing the second to it, comparing the third to the winner of 1 and 2....and so on.

I definitely don't consciously evaluate every line, it seems when it's obviously bad (or super good) I know it and move on. It's the tight +.20 pawns evaluations I really have to think through in my conscious awareness.

9/26/2007 12:20:00 AM  
Blogger Temposchlucker said...

Method 1 is the only viable method behind the board. For the other methods there isn't enough time. Without out a good intuition you have to gamble.

In OTB practice I use a mix of gambling, intuition and trial and error.

How to develop intuition? That must be done in the study room. That is where the other 3 methods come in. Kotov's method is so unlike how the human mind works that you can only do it with pencil and paper.

Method 5, backwards thinking, is probably only suitable for the studyroom too.

9/26/2007 02:53:00 AM  
Blogger Blue Devil Knight said...

I use more than just the first OTB. Going through a process of elimination is very useful in practice, for instance. I do it quite frequently.

You are right that exhaustively thorough Kotov or any thinking method can't be done OTB, but I use practical (i.e., limited) versions of each method in real games. It's not all intuition: sometimes consciously crunching through variations helps, even if I'm not as good as Kotov.

Backwards thinking gets me candidate moves, but I still need forward thinking to make sure it will work. E.g., wow this invasion square looks great, and if I deflect his rook and open this line I'll have a great avenue: I just have to push my pawn. That pawn push is now a candidate move that needs to be 'forward analyzed.' It will likely have consequences other than the plan. The plan is typically a best-case scenario, and in analysis we need to find worst-case scenarios.

9/26/2007 03:12:00 AM  
Blogger Loomis said...

2 common thought processes that make me realize how bad I am at chess:

1. Analyze the same line several times without being able to conclude if it's any good. Make the move.


2. Analyze the same line several times before spotting the glaring refutation. Make the next move that pops into my head.

9/26/2007 10:56:00 AM  
Blogger Blue Devil Knight said...

Loomis: I use #2 a lot.

9/26/2007 11:28:00 AM  
Blogger Pale Morning Dun - Errant Knight de la Maza said...

You could solve all your troubles BDK fairly easily.

Just play blitz chess. None of this quibbling over candidate move this, canididate move that. The games gonna be over in 3 minutes anyway. If you lose, just fire up another one like Cheech & Chong at a Dead Show. Ha!

Seriously, the standard of your knowledge is slow games and I've found that my best games were ones where my thoughts were focused. I looked at a variation and then moved to the next, and didn't run back to the first variation over and over again knowing it's worthless.

That's an amazing thing about the mind. I will look at the same crappy continuation for a tactic ten times, even though I've conlcluded after the third run through it's a bust.

It will be quite some time before I get 3 hours for a long game. For now, I'm a blitz master. By the time the newborn is 3 months old, my tactical ability will be supreme, and all will kneel before me....

I'm getting off track here, not enough sleep. Goodbye.

9/26/2007 09:42:00 PM  
Blogger likesforests said...

Hmm. In simple opening and ending positions I tend to think like Kotov, but once the game gets rolling I use some mix of intuition and back and forth thinking. When I'm defending I often purposefully switch to the Elimination method... which sometimes leads me into surprises! You know, "Drat, that loses my queen, and that's a mate-in-3. I'll go there." and you step into a mate-in-1!

9/27/2007 08:23:00 PM  
Blogger likesforests said...

"Method 1 is the only viable method behind the board. For the other methods there isn't enough time."

Interesting and probably true. When you're at home with all the time in the world, think like a Kotov and study those variations deeply, But when you're facing an opponent over the board, play confidently with the intuitive knowledge of good and bad positions that your studying has bought you?

There are some exceptions: When you are struggling to survive an attack and in time trouble the Elimination method is a Godsend. Also, endgames often aren't +/- but won or lost. A Kotov-like examination of all the possibilities is sometimes needed.

9/27/2007 08:44:00 PM  
Blogger Liquid Egg Product said...

I favor a combination method, but with heavy inclusion of a 5th:

5. Mind wandering: After checking a move or two 3-ply deep, fall into a misty mind-stupor. Start thinking about dinner, that hot chick you want to date, or kick yourself that you forgot your iPod. Then play the first move you thought of instinctively whether it's any good or not.

9/28/2007 06:56:00 AM  
Blogger Blue Devil Knight said...

In practice I find prioritizing and elimination very similar. I prioritize, explore, and then find a problem, and then keep looking. However, even if I find no problem with a certain move I'll often keep looking to see if I find something better. The boundaries between these is very fluid.

9/28/2007 02:10:00 PM  

Post a Comment

<< Home