The 10,000 Hour Meme

Since I'm blocked from the other thread on this topic, I decided to start my own.

I keep seeing references to 10,000 hours to become an expert.

Your 10,000 hour question comes from the research of Ericsson, Krampe & Tesch-Römer. They found that to become a world-class "expert" musician required about 10,000 hours of practice by the age of 20. That statistical study was expanded several times in other fields and was found to hold constant for expertise pretty much regardless of the field in question.

A guy named Malcolm Gladwell published a book called "Outliers" which looked at the research and generalized it into a pop-culture meme. However, Gladwell did not bring out the age function clearly in his work (if at all).

There is significant research that shows learning rate is relative to age, so the notion that 10,000 hours is a magic bullet at any age is probably false. However, 10,000 hours of work at any will certainly do wonder's for one's ability in any field.

I'm personally skeptical of 10,000 hours being sufficient to make an adult into a GM because of research in linguistics. There are about 8,000 hours in a year. Figuring that on average at least 2,000 of those are spent awake and engaged either directly or virtually with another person, then someone living in a culture where they are trying to function in the local language would require about 5 years to be 'expert' in the language.

However, research in native and non-native speakers shows that even after 10 years, non-native speakers who immigrated after the age of 25 can not process the language as effectively as native speakers. They still use concepts of their native language which do not directly map to the new language paradigm. For children, however, this time-line can be as short as 2-3 years, depending on how socially active the child is.

This means that adults simply can't learn to think entirely inside the domain of a new culture/language pair even after a decade of practice with native speakers.

I believe that chess is a lot like languages in that positions have semantic content.  A position 'tells us' something about what plans we should follow and so forth. It is, clearly, a very arbitrary and artificial language (not unlike mathematics) but is still in some ways analogous to one. Given the original studies on expertise, and the linguistic studies of non-native speakers, I'm extremely dubious that those over the age of 20 are likely to become GMs  with 10,000 hours of study. I suspect that most aren't likely to become GMs at all, and those that do will on average requires significantly more than 10,000 hours.

For those who started learning chess at a younger age, this might not apply depending on the amount and quality of their training.

Thanks for the post. I'm guessing brain plasticity is at work here to some extent. However, there are recent studies indicating that improvements in plasticity can occur beyond age 20. There is a nice chess book out on this topic, as well: Chess Master at Any Age, by Rolf Wetzell. 

If Dave took half the time he spends trying to make reality fit his personal POV and put it toward chess, he'd be an "expert" by now for sure .

Chess is pretty tough though, when someone refutes your ideas over the board you can't block them from repeating it in the future.

Will 10,000 hours of dicking around on tactics trainer, playing bullet games and posting bongcloud threads make someone a GM? Probably not...

Will 10,000 hours of serious study (calculation/visualization training, analysis of middlegame positions, studying master games, endgame practice, studying chess psychology etc. etc.) in addition to a healthy dose of OTB play make someone a GM? I find it hard to believe it wouldn't get them close...

Actually there are studies done with chess players also. The 10,000 hour figure is close to the average in those studies. I think the average was about 11,000 hours to reach master level. However, this is an average, nothing more. In one of the studies with chess players, I think one spend no less than 26,000 hours to reach master level, while someone spend only 3,000(!). Gladwell is a great writer, but he's a writer of popular "science" non-fiction, written for laymen, not for scientists. He utterly oversimplifies matters to push his arguments. It's entertaining to read, but not scientifically sound. 

When that's said, I think you're right that age matters a lot. I'm actually doing a project with chess, to empirically test the "10,000 Hour Hypothesis". I'm currently just over 500 hours, so there is quite a bit left. 

There's a fair bit of difference between master level and GM level.

Indeed there is !!  I have many drubbings at the hands of GMs to prove it too !!  

The studies I've read where looking at the hours to master level, not GM level.

I believe that there is research showing that a particular part of the brain learns language as a child, and then atrophies in adolecsence. Adults learn  a 2nd language with completely different parts of the brain then do children. There is no evidence that this holds true for mathematics, chess or anything else.

Mathematics is a language, eddiewsox.

That's why so many English majors get jobs as nuclear physicists. 

It's also interesting to think about the things you've spent 10,000 hours doing without the intention of becoming a master.

 Great point. lol!

English majors also don't get jobs a french professors. What's your point? That different languages are different?

Mathematics is an artificial formal language for a specific problem domain.

Mathematics is a completely different field from language.

some people can play 10000 hours without improving

"Mathematics is a completely different field from language."

http://en.wikipedia.org/wiki/Language_of_mathematics

I have no idea how long I studied in order to become master level. Here's some more food for thought. Note: "The 10,000 hour meme" is a great title.

People are different. You give the same amount (time, type) of instruction to people of the same level, and you will likely see two different amounts of improvement. In particular, if you take 100 novices, give them all a 10,000 hour course at chess, and you will see widely divergent results. In general, I believe that you will see rapid early improvement along with a plateau. But this is not news. It might be called Newton's Law of improvement, for its similarity with Newton's law of cooling, in shape and in spirit.

Where do people plateau?

What if instead of giving everybody the same exact 10,000 hour course, you gave everybody a 10,000 hour course suited exactly to their learning style? I would predict that you would see an overall higher level of play at the end, but still with a plateau and still with very different final values for playing strength.

I don't see it as "10,000 hours to get a GM title", or "10,000 hours to get a master title". I see it much more as "N hours of serious study to get to a plateau". And then one can measure at different points to predict what the final value will be.

Has any research been done in this area? Let me paint a picture of a very interesting future. Imagine a chess classroom where everybody has tablet computers loaded with lesson-specific software. The amount of data that can be collected and analyzed is immense. Is the student gravitating towards a certain type of position more than others? Are they gravitating towards positions which are each, so they can get more correct? Or are they doing the opposite? What types of mistakes do they make repeatedly? Do they prefer authors/articles which "talk down" or those which "talk up"? Do they prefer articles with diagrams every couple of moves? (Could the software have a hide/show diagram feature?) Not only could you imagine tailoring the learning process to each specific student, with student-specific homework as well, but you can also imagine the value of the data in aggregate.

-- Ozzie

 Cooking is an Art, therefore Van Gogh was a great chef.

Congratulations for making the most senseless analogy of the week!  Your prize is in the mail as we speak.