by [anonymous]
1 min read22nd Sep 20118 comments

3

Little over a week ago my work watched this video for a "self-improvement" seminar.
I hadn't seen this linked anywhere on LW yet, and thought it might be relevant, given lukeprogs' article on motivation.

New to LessWrong?

New Comment
8 comments, sorted by Click to highlight new comments since: Today at 3:12 PM

Has there been any work on how top chess players manage not to be distracted (or at least not badly distracted) by high stakes contests?

[-][anonymous]13y130

I've played in several competitive tournaments myself and had a chess coach for a number of years. I don't know of any general studies or research (so I'll be interested if anyone finds anything more principled than my anecdotal evidence). In my experience, the main goal of chess coaching and training was to teach you how to act like you were a computer. Any kind of "intuitive" play or even creative play was harshly criticized from a young age. The first goal was to memorize a massive amount of opening theory and what is known as 'book' knowledge. Once a student has a reasonable amount of book knowledge, then you move on to techniques to focus you on calculating quickly.

In my later years, I've come to realize that the heuristics that chess teachers employ cause you to think in a way that's analogous to simulated annealing. You scan the board and make simulated "proposal" moves in your head, and in the early stages of your turn you should be very open (high temperature) to suggesting any move, no matter how preposterous, to yourself. But you also have to just very quickly look at the move's surface level consequences and basically classify it as "worth more of a look" or "worthless". You do this a few times, and assuming some independence of the distribution of your brain's ability to pick out decent moves, you'll have whittled down to maybe 5-10 respectable moves, and perhaps a few other riskier moves worth a second look.

At this point you really dive in deep and attempt to examine long chains of moves, maybe 5-6 moves deep for each of the 5-10 moves you've got going in your head. Your annealing temperature is much lower, so you're less likely to spend time thinking about weird variants or moves that don't have an obvious upside. You systematically focus on understanding two things, the immediate consequences of the moves and the level of complication each line of hypothetical play will bring about. The reason complication matters is that human chess is all about resource management. Given enough time, every player in the tournament is good enough to see the best moves, potentially even the best moves that a computer would see. The difference in human performance, though, is decision fatigue, carelessness, and inability to calculate long, complicated lines in memory (some kind of human cache limit, basically).

During this deeper analysis stage, you should ultimately come down to 3 moves that seem to be the best. For these moves, you'll go back over the 5-10 move sequences, maybe even up to 15 moves deep if you're a grandmaster, and you'll not only detect the complication, but also the larger strategic principles. Will having two bishops in this particular variation give you more mobility? Will your pawn structure effectively lock down your pieces? After the dust settles in a large exchange sequence, will your opponent have a passed pawn and can you stop it without losing a whole piece to babysit that pawn? Will there be pressure on the dark (or light) squares because of an exploitable tactic. From this, you eventually emerge with your chosen move.

There is also some meta-analysis that goes into it. Since you are analyzing the amount of complications expected in certain lines of play, you also develop some experience for knowing when the really critical points of the game are happening. You obviously cannot spend the effort of the paragraphs above on every single move. You have to miniaturize it and hurry it up for most moves, and to do that safely you need to be confident that you're not making fast moves in a deeply complicated portion of the game. This can be accomplished by studying lots of classical games, or by knowing a lot about your opponent's games and styles prior to a match, Most grandmasters will have only two or three moves per game that take longer than 15 minutes to make, but each of these will take 45+ minutes. If you spend 45 minutes on a move, you have to get more bang for your buck out of that move, so the subsequent moves should be able to go faster. If you spent 45 minutes on a move and then your opponent surprises you and you have to spend another large chunk of time, you're in trouble.

At any rate, my experience was that the nerves and the situation don't play much of a role at all because, if you have practiced well enough, you dive right into this deep Zen of computation, busily reducing the game down into the conscious simulation of a serial algorithm plus memory lookups, according to what your coach / training has taught you.

In my experience, the main goal of chess coaching and training was to teach you how to act like you were a computer. Any kind of "intuitive" play or even creative play was harshly criticized from a young age. The first goal was to memorize a massive amount of opening theory and what is known as 'book' knowledge. Once a student has a reasonable amount of book knowledge, then you move on to techniques to focus you on calculating quickly.

This hasn't been my experience at all. At what level do you believe that memorization of opening theory is the first goal? I've seen coaches state again and again that most players under 2000 (i.e., most tournament chess players) spend too much time memorizing opening theory, when they would get far more benefit from working on tactics and middlegame technique, playing through lots of master games, and playing more slow chess. This is what my coach has recommended to me (I'm only about 1700 ICC standard, probably much less than you), and I've heard it stated again and again that too much emphasis on opening theory is a serious problem for sub-2000 players.

[-][anonymous]13y20

Every chess player has to memorize opening theory or they can't make progress. I agree that this is often over-emphasized, generally because it's the easiest thing for a coach to assign. I do think opening theory is pretty fundamental because it is a constructive way to teach someone about controlling the center of the board and developing pieces, which need to be learned concurrently with tactics. The same errors that people fall into with over-emphasizing openings are also prone to occur when people over-emphasize solving chess puzzles or replaying GM games. If you spend too much time on any one aspect in the developmental stage, it's not good. But still, 1200 level players must really study a few basic openings just to even get a handle on engaging in a game of chess.

My preferred approach is to teach one or two main variations for a couple of openings for both black and white. Then, after these are reasonably understood, I like to spend time focusing on the goals of the opening, like controlling the center of the board and developing minor pieces to good posts. That's usually the point where tactics and combinations start to become relevant, so its natural to work on that. This is just my opinion, of course. I'm sure many chess instructors have their own opinions.

Playing slow chess is by far the most problematic issue. Chess requires so much patience, and thinking like a computer when your opponent is thinking about their own move is really hard, especially for children. Most kids don't have many options to practice other than internet chess, which is generally blitz dominated.

At any rate, I never believe that memorizing opening theory is an important goal unto itself, not even at the highest level. It goes in spurts. Maybe study the Ruy Lopez and Queen's Gambit to illustrate basic ideas about central control, then take a break from openings and play dozens of Ruy Lopez or Queen's Gambit games to build up the ability to systematically study positions and compute moves, then go back and extend some opening knowledge to study piece development, then play more games and study tactics and combinations.

My only point in my original comment was that because of the systematic approach to move calculation and focus of attention, top chess players are usually not concerned about distraction so much. It is exactly avoidance of distraction that they have been training for to get them into the big tournament games in the first place.

Thanks for the explanation. Your explanation accords with what I've heard from my coach and what I've read. What surprised me in your original comment was this sentence in particular:

The first goal was to memorize a massive amount of opening theory and what is known as 'book' knowledge.

That sounded to me like much more than "studying the Ruy Lopez and Queen's Gambit to illustrate basic ideas about central control". It sounded more like "try to memorize every line of every variation of the Ruy Lopez that is in MCO".

[-][anonymous]13y20

Yes, I see your point. It was strongly worded, I think I was just typing quickly and over-emphasized that. In my mind, I was lumping a lot of things together as 'book theory' but it is good to point out that for developing players, it's not good to devote too much time to memorizing anything, whether it's openings or solutions to chess puzzles.

What I mean about "intuitive play" being stifled early on is that one of the first things I was taught was that playing moves that "look good" or "seem right" is not the right way to learn. Very few people can be successful playing with this sort of intuition. Ironically, though, this is why many chess players list Mikhail Tal as their favorite world champion.. he frequently played by intuition and would specifically choose technically unsound positions just because they had much more complication, which just personally interested him more.

This is the best description of a fairly strong chess player's thought process that I've read. If it were worth the effort, I would link every person who asked me, "How many moves deep do you calculate in chess?", to your comment.

[-][anonymous]13y00

This is tremendously interesting. I regret that I have only one upvote to give this comment.

[This comment is no longer endorsed by its author]Reply