« Intelligent agent modeling | Main | Assorted links »
Chess is now a young man's game
Here are the latest ratings. You have to go down all the way to #61 (Nigel Short) before you find a player older than 40. It didn't used to be that way. For instance in 1963 Mikhail Botvinnik was world champion at age 53.
The game is more competitive, more players come from countries where chess is relatively new (China and India will give you young stars, not old stars), and there is great value from training with computers. If you didn't start training with computers until you were thirty-five years old, you are at a serious disadvantage.
Consistent with these hypotheses, there are also more and more prodigies in chess. Can you think of any other reasons for the falling ages of top chess players? I also see a general principle operating: the more exact a "science" the game becomes, the smaller is the value of accumulated experience relative to sheer skill.
Posted by Tyler Cowen on December 31, 2008 at 07:43 AM in Sports | Permalink
Comments
Tyler,
Being from India, I'd like to point out that Chess isn't a 'new' game in India. Some accounts believe that it started here.
However, what's likely is that Anand's success has only spurred a lot of competitive playing, which earlier many people didn't bother about, or take seriously. A parent in India would have not previously encouraged their wards to take up competitive chess seriously, but now it's 'accepted' to encourage a career in professional sports.
In response to your question, I'd like to say that the nature of the game at the highest level has changed, due to computers. Most players, when preparing for matches, actually have teams of players and computers who analyse the competitor comprehensively. What this does is reduce the edge that human intuition/accumulate experience as you've said. My assumption is that human intuition is directly related to experience.
A happy new year to you!
Dibyo
Posted by: Dibyo at Dec 31, 2008 7:59:43 AM
Which reminds me of a xkcd comic:
http://xkcd.com/447/
Posted by: Elaine at Dec 31, 2008 8:05:37 AM
Can you think of any other reasons for the falling ages of top chess players?
The Internet makes it easy for people, especially young people, who want to improve their game to find challenging human competition. Would-be chess prodigies no longer face the problem of being the undisputed master of their local club and having no one interesting to play with.
Posted by: Cyrus at Dec 31, 2008 8:23:11 AM
Something similar is happening in professional poker.
Posted by: AW at Dec 31, 2008 8:28:55 AM
I'm with Dibyo -- computers have short-circuited the path to obtaining experience. I suppose you could say that at least in tactics, computers have brought the game closer to an exact science.
Posted by: Alex at Dec 31, 2008 8:37:33 AM
There's also the end of the Soviet chess monopoly. Fischer wasn't entirely paranoid in saying the Soviets took it easy amongst themselves to beat up the rest of the world.
But I agree that computers are the main difference.
Posted by: Ted at Dec 31, 2008 8:38:25 AM
So, Tyler, how did you get to be such a prodigy
that you were the youngest ever New Jersey state
champ without playing against computers? Or
did you?
Posted by: Barkley Rosser at Dec 31, 2008 9:05:25 AM
It's the youngins that really seem into the more complicated variants as well, like bughouse.
Posted by: Aaron at Dec 31, 2008 9:09:23 AM
Another interesting development in chess is the revelations that have come out of computer-generated endgame tablebases. "Playing chess with God", forced wins that take 517 precisely necessary but seemingly random moves, with no rhyme or reason discernible to any mortal human even as it unfolds right in front of your eyes.
Posted by: at Dec 31, 2008 9:36:10 AM
When we say that younger players are better because its more of a science now, aren't we saying that with the training of computers both memory (being able to remember patterns/traps that you've seen before) and calculation ability are at much more of a premium than they were in the past.
Posted by: rifhen at Dec 31, 2008 9:48:06 AM
The Flynn effect?
Posted by: Richard at Dec 31, 2008 9:50:59 AM
computer chess allows for better cheating. no longer do you have to team up with your fellow comrades and fix matches.
Posted by: zombie_bot at Dec 31, 2008 10:16:26 AM
"I also see a general principle operating: the more exact a 'science' the game becomes, the smaller is the value of accumulated experience relative to sheer skill."
But I seem to recall you mentioning offhandedly in our Macro I class that many of us in the room probably had a better understanding of macroeconomics than did Keynes himself--not because we were more skilled, but because the intervening years had produced a division of labor in the discipline which greatly compounded our accumulated experience. Or am I misremembering?
Posted by: Rich at Dec 31, 2008 10:37:04 AM
It's not just computer programs and tablebases. It's also the ability to play constantly. Magnus Carlsen can just sit down with his computer and get a game against world-class competition over the internet whenever he wants to. That has given him playing experience it used to take years to obtain when you had to actually be in the same room as your opponent. Think of how much longer it would have taken him to play 1000 games against grandmasters if he had to travel to do it.
Posted by: Joe Kristan at Dec 31, 2008 11:21:59 AM
Why do most mathematicians (and economist) achieve success and fame for work done before they were 40?
Posted by: SS at Dec 31, 2008 1:47:05 PM
The second and third highest ranked players are 39 years old and Kasparov and Karpov are even older. I don't think it's true that most of the top players are young.
Posted by: Lungoficus at Dec 31, 2008 1:53:12 PM
There was an article in the Atlantic about how software has changed the game.
Posted by: Dave at Dec 31, 2008 2:02:41 PM
Where are the women? I'd sure hate to be a modern Amerikan or European woman, let alone the others, especially Muslim, excluded lifelong from chess, math, physics, engineering, cabinetmaking, deep-sea diving, skateboarding, hitchhiking and all the other fun things in life. Even the best in haute cuisine, haute couture, hairdressing, and movie making are almost exclusively men, and let's not talk of CEOs!
Hell if it weren't for sex, dancing, nursing and teaching kiddies, women would hardly participate at all in modern life.
Posted by: jimbino at Dec 31, 2008 2:47:38 PM
but who paid attention to kids fifty years ago? I think we're looking at the Beckerian quantity-quality tradeoff here.
Posted by: pat toche at Dec 31, 2008 3:16:50 PM
Tyler wrote: " the more exact a "science" the game becomes, the smaller is the value of accumulated experience relative to sheer skill"
Not necessarily, the required experience can now be accumulated at a much younger age. Indeed, the younger players may be the most experienced on the planet. Perhaps you could elaborate on how you differentiate between experience and skill. If by skill you mean innate talent rather than practiced technique, it is possible that even a very small loss in cognitive function in middle age would have a very noticeable effect among the top rankings.
Before passing judgment, however, I'd also like to know the base rates. That is, how many players are there at the various age groups. The age gradient could be an artifact of the base population of players.
Posted by: Bill Nichols at Dec 31, 2008 3:18:50 PM
Of the top sixty players, three are from India (b. 1969, 1981, and 1986) and four are from China (b. 1987, 1983, 1985, and 1989). I don't know how much of a shift there has been, but, while all but one of these guys are pretty young, there aren't enough to make a dramatic shift on their own, unless of course young players are better at playing other young players (perhaps due to the shift in tactics caused by computer training).
Dibyo: While it's not a new game in India, it may be new for Indians to start playing professionally (allowed them to spend more time playing=more practice) or internationally (allowing them into the rankings?)
I suspect this is even more true in China. China probably has always been somewhat keen to prove to the world their people are the best at chess, along with everything else, but they probably only trained the few they thought were worthy. Increased wealth and a less controlling government allowed anyone interested to play and rise up through the ranks, rather than just the government's chosen ones.
Posted by: Luca Masters at Dec 31, 2008 3:34:58 PM
If it used to be that older people were the best chess players, that implies that the mental peak for chess comes at a later age than the age of the current best chess players. So after the widespread adoption of these computer training techniques, the best is yet to come.
Also interesting to me is the advancement of machines versus the advancement of human minds in playing chess. That is, are the machines getting better in chess at a rater faster than humans? I'd be very interested to know.
Posted by: Justin at Dec 31, 2008 5:24:34 PM
it used to be that youth was wasted on the young, but now we leave that to computers.
Posted by: babar at Dec 31, 2008 5:42:54 PM
Motivation. Motivation combined with a larger knowledge base which therefore requires more work effort to remain at or near the top is the reason fewer older players remain near the top. Most of the few over forty players that are on the list were once much stronger (eg Karpov, Beliavsky, Bareev) than they are today. Botvinnik had a smaller skill set to maintain. Fischer and those who followed changed the training landscape. It is just harder to get up at forty and put in 12 or 16 hours of hard work. I bet none of us are as excited to read our recent Nobelist Mr Krugman’s work of the last 15 years (since his 40th birthday) as say what he did before that. It just used to be cheaper to remain a top chess player.
The reason there are more prodigies, I suspect, is that there is a better understanding of proper training methods combined with a wealth of material to train with. I suspect the plethora of books is more important than the computer, although I do not underrate the chess databases as a source of training material. A young chess player in the 1950s or 1960s (in the US at least) had access to a few games collections and maybe two below master tactics collections and two quality endgame books and no explanation of how to use any of them. The Soviet Union, of course, was much better, but still the most accessible way to improve was to play a game and analyze it (preferably some analysis with a trainer). Say four hours of play and six hours of quality analysis of the game would give ten hours per game. If we are aiming at a magic number of 10,000 hours of quality activity to reach mastery, then it takes a thousand games. Not far off the 700-800 games minimum for most players to reach grandmaster level. It is hard to play 100 quality slow games a year anywhere in the world previously or today. (Speed chess on the internet and otherwise does not count.) But today a young player can fill in the gaps with quality training thereby speeding the process. Note that computers make the access to trainers more widespread through the internet. So more quality training more widely available speeds the process to mastery and increases the quantity.
Posted by: montclus at Dec 31, 2008 7:16:20 PM
there is another hypothesis. maybe we just now happen to have a strong generation of chess players, whereas the preceding generation's talent went nowhere.
if this were a conversation about "writers" rather than "chess players", i imagine the dialog would be quite different.
Posted by: babar at Dec 31, 2008 7:34:35 PM
I would hypothesize that it has nothing to do with chess - it is just the scoring system. There's always going to be a few people each year who win a lot. A young guy who wins a lot shoots up the rankings, but an old guy with prior losses creeps up a little.
Posted by: Eastside at Dec 31, 2008 8:27:54 PM
I think that Ericson's (sp?) "10,000 hour rule" is very applicable to chess proficiency, with its heavy emphasis on pattern-recognition and the requirement to memorize long chains of moves and countermoves.
Computers can accelerate the accumulation of experience, plus improve the quality of practice. A young Grandmaster-to-be can play whatever ELO rating he/she likes many times per day, and even mix up the styles at will.
Posted by: Doug at Dec 31, 2008 9:53:22 PM
There's also the possibility of a genetic effect; I recall reading about aspergers going up in recent generations.
Software Industry: bringing nerds together to breed?
Posted by: Steve at Dec 31, 2008 10:04:03 PM
Putting aside even the vigor of youth and technological advances we find that there is a fundamental advantage to being young: you learn the best stuff. As theory evolves, after a certain point, experience becomes counterproductive. A prodigy today who's read two books (the latest and best) on will more rapidly come up with optimal moves then the prodigy of 30 years ago who's read 10 books on the subject, 7 of which are of mostly useless in current lines of play. Put another way, accumulating theory makes modern instruction innately superior to past instruction.
A simple example might be the old Evans Gambit. If you learned to play in the mid-19th century you probably studied this opening quite a bit. If you learned to play in the mid-20th century then you probably have looked into a few lines of the Lasker Defense but not "wasted" much of your time on the opening in general. {Perhaps by the mid-21st century the Evans will have completed it's comeback though}. Repeat this process for opening line after line and what you find is that the head of the new player is full of modern "optimal" ideas while the older player's head is only one third so filled.
Posted by: Steko at Dec 31, 2008 11:02:37 PM
As others have pointed out, 40 is just a number, and numbers 2 and 3 in the world are both 39. Gelfand (number 15) is now also 40 for sure (or will be in 32 minutes). For what it's worth, I think the stronger way to make your point would have been to focus on Magnus Carlsen, who is #4 at the age of either 18 or 19. A 2776 rating at that age? Now that's *really* frightening.
For that matter, I see that nobody has mentioned the somewhat notable fact that the top 100 players includes exactly one woman. Surely, this fact should be of interest to at least some of the labor economists out there?
Posted by: Jonathan King at Dec 31, 2008 11:40:06 PM
um, the comment about writers is dumb. chess is not only a game of skill, but one with zero chance. that is, it might as well be math (it's not longer anything close to an art (yes, yes, at it's highest level, math is an art, etc.)). and it's clearly just the ability to (i) learn from SW (the ability to play SW at the fisher level, and then take back moves and play things out until you either find a way out or realize why what you did was wrong is invaluable) and (ii) play as many games against great competition as you can stomach (the ICC and others are truly stunning). finally, for those who think they love chess, but haven't tried bughouse, please, please do. it's the greatest game ever (very social, great fun, makes you think in different ways that will end up helping your trad games).
Posted by: dj superflat at Jan 1, 2009 12:26:15 AM
"40 is just a number"
the mean age of the top one hundred is 28
Posted by: at Jan 1, 2009 4:18:43 AM
Jimbino, the top ranks of chess have ALWAYS had a paucity of women. I remember looking over a list of US rankings in Chess Life & Review probably 30 years ago and noting that if I had been female my rating would have put me in the top 10 women in the USA.
As a male, though, I wasn't even in the top 1000.
Posted by: Jens Fiederer at Jan 2, 2009 11:00:37 AM
