Thursday, January 28, 2010

Katar on the social dimension of chess improvement

In response to my previous post, which has lots of good comments, the following from Katar triggered a bunch of thoughts:
Environment is a big factor. If one has chess-playing friends it will speed up one's progress...

Somehow internet friends do not seem to have this effect, as far as i can tell. But i'll tell you this: I met a few cool guys at the club, they happened to be mostly around 1600 or 1700s. We played blitz and analyzed games after the club meetings. Within a year I was also rated in the 1700s, where i stayed for 2 years. Then my buddy jumped to 1870. Somehow I almost instantly jumped up to about 2000. The hard part is finding decent players (geographically close) who are interesting and diverse people away from chess. Then you can develop real friendships and what starts mainly based on chess can later exist independent of chess. Which is just a board game for gods sake.

There is something magical about going over games with humans, moving pieces around, throwing out ideas, making fun of each other, getting stumped on problems, doing postmortems with others watching. Humans are social creatures.

Imagine learning a language without interacting with other humans. We often stress the analogies between chess skill and language acquisition, but seem to forget the important social dimension of language acquisition. My hunch is the social dimension is radically under appreciated in the chess improvement blogosphere and chess improvement literature (for obvious reasons in the blogosphere, but also because books tend to focus on what they can directly provide the reader).

Coaching=Sex, Computer=Masturbation
Perhaps it is no coincidence that all the best players in any sport, including chess, have had extensive individualized coaching from real coaches. This is something that is stressed, to greater or lesser degrees, in the books. However, the books don't stand to gain all that much by telling you to get a coach rather than follow their generic chess path. And what can they do besides recommend that you get a coach, briefly explain why, and leave it at that?

It is usually stressed that it is the individual treatment is what is important in the coach relationship, but my hunch is that one-on-one social interactions with real people may be equally important. Our brains are wired to learn in a social way. Imagine an infant that hasn't learned to walk. Or speak. Learning such skills is an almost ineliminably social affair.

At the International chess school forum, I had a very vigorous debate with a guy who didn't do postmortems with his opponents because "Fritz is better" so what could he learn from them? I found this to be one of the most insane self-immolating sentiments I had ever heard about chess improvement. You can find the thread here, and it relates to this topic in ways I frankly hadn't even thought about yet.

At that post I said:
Key is to talk to chess peers after the game, go over it, hash over ideas, see what they were thinking, play around with pieces, refute each other. The more exposure to real people (including yourself), interaction, conversation about positions, with nobody involved allowed access to the computer nipple, the better. Then the computer later, if you must, will reveal some weaknesses of what everyone said, and that is a nice final evaluation of the position, but not necessarily superseding what you found in the conversation.

I have usually thought about this topic in terms of the importance of interacting with real people because it is so important to think like a real person, and be exposed to what other people think (because you will face real people in tournaments). This is in contrast to trying to emulate a computer that uses algorithms that do not resemble in the least the type of cognition humans use. I remember J'adoube tried to argue that computer emulation was the ideal, which I still think is just silly.

There are levels of possibilities here that go well beyond the need to have a human, rather than computer, approach to the game, and Katar's got me thinking about them.

Just putting my thumb on the scale

I don't want to suggest that everyone can get better if they take the social dimension of improvement into account. Some autistic folks probably would learn much worse if forced to communicate and interact socially with people to learn chess.

I'm just suggesting that the picture of the lone chess improver studying tactics on his computer, emerging like a butterfly from the cocoon of CT-Art problem sets to demolish the competition at the World Open, is pretty much bullshit. And it's bullshit in ways we in the chess improvement blogoshere, and especially Circles advocates, might not appreciate (not because it is bad to do a lot of tactics, but doing them with other people and talking about them and moving pieces around might be better).

I also am not suggesting that it is a mistake to work on chess by oneself, such as solitary opening study or whatever. I'm suggesting a counterbalance to the picture of solitary chess improvement, but not eliminating the quiet enjoyment of chess by oneself. I am, however, suggesting that improvement solely by solitary means is not the norm, but an aberration that most of us should not strive for as a realistic improvement technique. I'm basically putting my thumb on the scales, which are usually biased toward the solitary picture of improvement.

This all reminds me of something Davies once wrote that I cited here:
It really doesn’t matter what you study, the important thing is to use this as a training ground for thinking rather than trying to assimilate a mind-numbing amount of information. In these days of a zillion different chess products this message seems to be quite lost, and indeed most people seem to want books that tell them what to do. The reality is that you’ve got to move the pieces around the board and play with the position. Who does that? Amateurs don’t, GMs do.

Perhaps even better than moving pieces around in isolation (solitary chess) is the moving of pieces in an ongoing conversation about a position with another person. And what is a game of chess in a real-life tournament, if not a conversation over the board that involves moving the pieces?

What if you live in East Bumslap South Dakota?
To those that literally cannot find real people to play chess with, the computer and internet especially is a godsend. To what degree can we reproduce the benefits of chess facetime with others? I welcome your thoughts in the comments. E.g., is online coaching as good as face-to-face coaching? Is there some way to have a "chess club" atmosphere online?

Frankly I think just as a gymnast in East Bumslap would move to where the best gymnastic coaches are, that might be necessary for people that have high aspirations in chess improvement. You need to travel to play with real people, and perhaps even get coached by real people.

Or is that total bull? Can getting phone lessons be just as good?

(Aside: I wonder how many posts in a row can be seeded by comments from previous posts.)

Monday, January 25, 2010

Innate, acquired, or both?

The comments at my previous post were again excellent. Pretty much you guys are now writing my blog for me.

Anonymous said the following, which struck a nerve with me:
Some of us need to face the fact that we have no natural chess ability. From his rapid rating increase from the beginning it is obvious to me that MDLM had natural ability. Book smarts does not equate to natural ability in chess or else some of us would be Grandmasters.

I coached youth baseball for 15 years. Some kids had no natural ability at all and despite one-on-one instruction, showed no improvement at all over the course of their years on the diamond. They stuck with it for fun and to be with friends.

I look at it as two Gaussian distributions, one of innate talent, the other of time spent playing and trying to improve. I could beat someone who is way more talented than me, as long as they have only spent ten hours on the game in their entire life. However, there do seem to be those that are abnormally good, and when they work abnormally hard they destroy (e.g., Carlsen). These are the GMs. The rest of us with average talent who work really hard can probably reach around 2000.

Note also that "innate" talent Gaussian shifts markedly to the left as a function of onset of age of playing chess (I started at age 35). Hence, age, talent, and motivation/time all interact in interesting ways.

Unfortunately none of this is new, and all of it is obvious. You can't change your age or your innate talents. All you can do is train, play, and train some more. As I've said, chess improvement is like planting a flower. You can nourish it, do everything possible to have a healthy flower, but ultimately you can't force it and it will grow in its own way.

I certainly see the effects of "talent" in academia. There are kids I tried to teach formal logic, and they would go to office hours, do extra homework, and just couldn't seem to get the logic of the conditional, or indirect proofs. For some students, who never went to class, and were slackers, they just "got" it, and got the As, while many of the people who put in the hours would end up with crappy marks.

When it comes to chess practice, I often feel like those slow kids. I'm quick enough on the uptake when it comes to wordy explanations, the kind of crap that Rowson basically says is useless in practice. And he's right.

Thursday, January 21, 2010

Thanks, Robinson

Useful analysis of MDLM from previous post comment section:
I think one thing that is often overlooked in Michael de la Maza's success is the high number of rated games he played over a relatively short period of time. I've read many, many of the past Knights Errant posts and I don't find a lot of comments about the number of rated, slow games being played as part of their improvement programs. (Nor do I find a lot on people talking about doing the micro-level drills).

MDLM played 193 rated games in 45 events from July 1999-July 2001. He's only ever played 4 games that were quick rated. These quick-rated gamers were in the second-ever rated event that he played in and he received a provisional quick rating of 1649, which remains as his quick rating to this very day. So, 189 slow, rated games played over a 2-year period. This is an average of nearly two serious games per week. He accomplished this by playing a combination of a weekly regular game and a number of weekend Swiss events.

His rating after this period was 2041. During these 193 rated games, he played 143 different opponents for a record of 116-55-22 (W-L-D) (.658). His best upset was vs. Robert Egan (1713), 417 points higher rated (08-22-1999). His best draw was vs. Edwin Burnett (1556), who was 295 points higher rated at the time (08-08-1999). His worst loss was to a player 479 points lower rated, Samuel Atwood (1043) (02-29-2000), who was also the lowest rated player he ever lost to in a rated game.

The highest rated opponent he ever beat? Joel Johnson (2207) (10-31-2000). The highest rated opponent he ever drew was Herbert Carswell Jr. (1991) (7-05-2001).

Against players rated higher than him, he was 33-39-9 (.463). Against those rated 100 or more points above him, he was 17-32-3 (.356).

Against lower rated players, he was 83-16-13 (.799). Against those rated 100 or more points lower, he was 57-10-5 (.826).

MDLM was 41-13-14 (.706) against players who had a rating within 100 points plus or minus of his rating.

If we consider MDLM's lowest rating, which is a rating and not playing strength, as 1163 (his provisional rating after his first 4 games), then we can say his ratings climb was 878 points in about 730 days. However, since his ratings climb happened immediately upon beginning playing rated chess, it is very difficult to make any determination about actual playing strength at any given time. Though he says his rating made him a "Class D player because I played, well, like a Class D player," he is being a little bit disingenuous because his rating immediately went up from there and in four months later he had a 1421 rating and four months after that a 1600 rating.

Perhaps the circles didn't help him as much as playing slow rated games. Because he did both, we can't separate out the effects. Perhaps the best way to improve at tactics is to play slow games and go over them after.

That's one reason why, for beginners, I emphatically do not recommend the Circles, but simply playing lots of games (as described in my Chess Study Plan for Beginners). As I said there:
Much like learning to sing or swim well, chess skills are built up largely from extensive practice. Hence, as a beginner, it is most crucial to simply play chess. Some people attempt to improve at chess by focusing on everything but playing: they do puzzles, read lots of books, or (worst of all) spend lots of time studying openings. As I discuss below, some of these things are important, but if you are truly a novice you need to develop some intuitions about the mechanics of the game by playing as much as possible.

Monday, January 11, 2010

Reminds me of the good old days

Remember when everyone was doing the Circles? Remember the sometimes crazy heated discussions of them? How many problems per circle? How many circles? What software to use? That was fun.

Loomis provides a good reminder.

I need to think more about the circles, where I stand on the de la Maza methods. I started the juices flowing in a comment at Loomis' blog.

Chesstiger brought up one of the objections Knights heard frequently (that it is bad to do circles because you just memorize specific puzzles). It is far from clear that chesstiger is right. I wrote a detailed response to this objection here). However, it was nice to get thinking about this again, now that I've got quite a bit of perspective on the whole Circles thing. I will have to revisit it and think about it some more.