Are chess players getting better over time?

Mostly.  Kenneth W. Regan and Guy McC. Haworth analyze games move-for-move, using chess-playing computer programs.  The result:

…we conclude that there has been little or no ‘inflation’ in ratings over time—if anything there has been deflation. This runs counter to conventional wisdom, but is predicted by population models on which rating systems have been based…The results also support a no answer to question 2. In the 1970’s there were only two players with ratings over 2700, namely Bobby Fischer and Anatoly Karpov, and there were years as late as 1981 when no one had a rating over 2700 (see [Wee00]). In the past decade there have usually been thirty or more players with such ratings. Thus lack of inflation implies that those players are better than all but Fischer and Karpov were.

Comments

Ah, data with inflation adjustments. I missed posts with that kind of data. Glad they're back. On the other hand, it still goes up!

"Data" and "projections" are two different things!

Makes sense. I was a strong A player about 15 years ago, this holds true at amateur levels IMHO also. Players of a 1900 standard today are noticeably stronger than back then. I would further propose that this is true in basically any field that has remained competitive for the past 20-40 years.

I was in Dr. Regan's Programming Languages class. He's brilliant

It's all about the number of games played, particularly online play (both against computers and humans). You see this in texas hold 'em too; a kid will have played as many hands in his life by the time he's 21 and can legally enter a Vegas tournament as an 60 year old pro ten years ago. This can't help but improve play.

Championship golf courses have been lengthened about 10% over the last couple of decades, and the greens are dramatically faster than the were a generation ago.

Is there a Flynn effect with chess players?

I thought Kasparov would have had the highest rating?

This should be expected given that we now have computers to aid chess-player's study of the game. Thanks to computers, not only do top players know the problems with thousands of variations that were popular 40 years ago, but their endgame play has improved dramatically too.

There is an interesting argument in this post 'Two Hour Marathon in 2045' that increased competition will lead to a linear improvement in sports records. http://allendowney.blogspot.com/2011/04/two-hour-marathon-in-2045.html
It could be similar in chess

I'm not sure what new information is in the linked paper. The numerical ranking of any player is only relative to other players at the same time, not to an arbitrary standard. So as computer programs became generally adopted, players improved - but not relative to the other players who were also using computers. Don't forget.. in any assertion that humans are "getting better" as computers are being used, you've left out the skill improvements that older players would of course have seen if they were able to harness teh internets for their study.

Give me Tal/Alekhine/Capablanca every day of the week.

I remember Raymond Keene doing something similar (Using Fritz as his computer muse) and checking an early British Tournament (Hastings? Late 1920s early 1930s) and coming to the same conclusion. The play was riddled with tactical errors.

Wasn't it John Nunn who did that? In one of his puzzle books?

Yes, it was John Nunn. Good catch.

By the same token, Capablanca had the fewest average errors per game

http://en.wikipedia.org/wiki/Comparing_top_chess_players_throughout_history#Actual_moves_played_compared_with_computer_choices

I think I agree with unsympathetic.

in any case, I don't think that chess players today have more latent talent - it's just that whatever talent they do have is being more fully reached, if they are "getting better".

The article cited in the blog post specifically mentions the Crafty study as being somewhat flawed and why, FWIW.

Better training methods will apply to chess just as much as they do in track and field.

I would expect they are, but I also expect that even so within 20 years a $1000 will be able to beat any human player 9/10 games.

$1000 PC*

This is already true today, and the computers are getting better much faster. The top chess programs are unbeatable by humans without enormous concessions.

Really? Even on a PC? That's surprising, I would have thought the best players were still considerably better than an everyday PC.

Yeah, in fact the difference between the PCs and the supers in not that much. Chess isn't particularly easy to make parallel.

The key algorithm, Alpha-Beta, is inherently serial, and trying to split the work into parallel tracks involves doing things less efficiently.

This is just the 'standing on the shoulders of' effect, of course the current generation is always the "best ever", objectively.*

Watch some film of Vince Lombardi's great Packers teams -- they look like a good high school team today, And they were about the physical size of one. They would be destroyed by a decent college team today with a 40-pound advantage at every line position.

That's why the best way to choose "the best" is the gap between number #1 and say #2 through #10. Give me Paul Morphy any day by that measure.

* The exception for some reason is horse racing. Thoroughbreds haven't gotten any faster for a long time.

Probably because the pool of horses got smaller as they were replaced as the primary mode of transport.

Comments for this post are closed