Preliminary Results of FIDE Chess Survey

Thanks to everyone who took part in this survey. Here are the preliminary results. The sample consist of 581 players to date, with five grandmasters, 25 international masters, 67 FIDE masters, two woman’s grandmasters, two woman’s international masters, and two woman’s FIDE masters. The results are only preliminary, however.

Some highlights:

Players learned the moves at a median age of eight years old (masters about two years younger). The median age of starting serious play and taking part in the first rated tournament is 14, 12 for masters. Most players have had coaching. Players average around five or six hours of chess study a week, but the range is huge (0 to 60 hours). Number of hours of study of chess material is a factor in expertise level but only a relatively minor one.

Most players firmly believe in natural talent for chess and most believe that top ten players have some special traits, that few really can reach that level. However, many believe that a lot of study and practice can take a player a long way. Some believe that almost everyone can get to FIDE master with enough practice and study.

Views on what natural talent for chess consists of vary, but some common ideas are good spatial ability, high IQ, good memory, creativity, high motivation, a strong will to win, control over emotions, and psychological hardiness.

Eventual grandmasters take a median 390 FIDE-rated games from rating list entry to gain the title. Most players do not play anywhere near enough rated games in their careers to have a realistic chance of becoming a grandmaster. About two thirds of those who do play over 900
games actually succeed in becoming a grandmaster. However, those who play over 740 games without becoming a grandmaster on average seem to strike an impassable barrier at around 2400 level.

Analysis of rating data of players who played over 900 FIDE-rated games show that eventual top ten players indeed are identifiable from list entry. They get on the rating list much younger on average, get the grandmaster title much younger and much faster, and rise in the
ratings much faster than other grandmasters.

Most believe that playing rated games and studying are equally important in developing skill.

The full article is below.

The material above is not copyrighted and may be used freely in any medium.

The material below is copyright 2008 by Robert Howard but may be used freely in part or
whole on any website with attribution.

___________________________________________________________________

Background

Can anyone become an expert at any activity with enough practice and/or training?
Or, does natural talent limit ultimate performance level? The idea of natural talent derives
from common everyday observations of some people acquiring a skill much more rapidly
than others, and/or very much younger than average, and then reaching a higher ultimate
performance level. Since at least the 19th Century, it has been assumed that natural talent
differences among individuals set differing, fixed maximum performance limits. Training can
improve skill only to a certain level. This idea still largely dominates formal education and
sports coaching, where the search is on for people with evident natural talent to develop
further.

What natural talent actually consists of might differ across domains of expertise and
may consist of a set of abilities and even personality traits, which are partly innate. Complex
intellectual domains may require only a threshold IQ score of around 120 for high
achievement, beyond which creativity and motivation differences create performance
differences. In the sciences and the arts, generally a few people excel and make most of the
contributions. Studies show that they tend to have a similar set of ability and personality
characteristics; IQ score over 120, and such personality traits as high motivation and non-
conformity.

However, in the last few decades, an increasingly popular view has arisen which disputes the
importance of natural talent and holds that expertise just reflects extensive practice. Top
performers have no greater natural talent, they just got more practice, and any “normal,
healthy individual” can reach elite performance levels by starting early enough and practicing
extensively. “Deliberate practice” is seen as particularly important. It is defined as performing
“... appropriately challenging tasks that are chosen with the goal of improving a particular
skill”. In chess, deliberate practice might involve receiving coaching, reading chess books,
studying opening variations and endgame positions, studying games of top players, and
playing a very strong computer to identify one’s weaknesses. This view acknowledges that people do hit performance ceilings but holds that these ceilings can be overcome with more of and the right kind of practice.

The idea is appealing and fits in with the notion that hard work can overcome all
barriers. Indeed, in East Asian cultures, the view is widely held that school success depends only on
hard work. In the West, many school children today are told that they can be
whatever they want to be. And, some studies indeed suggest that top performers often report
having had a lot more practice than lesser performers.

However, the idea that only practice is important actually is extremely hard to test. The problem is that motivation, amount of practice, initial success, opportunity, and possible natural talent are all interwoven. We tend to like what we are good at and successful at and persist in such activities, and dislike and avoid activites that we are no good at. If I wanted to be a champion boxer but lacked the strength or speed and got knocked out in the first round every time, soon I would abandon boxing for other interests. The boxers who do persist may have succeeded early because they have a lot more natural talent. So it is with chess. People skilled enough to get an FIDE rating might simply have much more natural talent than others who start chess but soon drop out because they always lose. Those who become grandmasters might just be those with much talent who persisted because they succeeded early. Someone who shows initial success also might be more likely to get coaching, more opportunities to play, perhaps turn professional, and secure a cumulative advantage.

Chess as a useful test domain

Chess is a very useful test domain to study expertise development and test the natural
talent idea because playing is an obviously intellectual task and rating systems measure
expertise levels over many years. Chess has no glass ceiling. Unlike the case in most other
fields, it is unnecessary to have the right social background and to network and to tirelessly
self-promote to succeed. Furthermore, chess experts really do have to be skilled. In many areas, a person
can be judged an expert, even an eminent one (e.g. in wine tasting or economic forecasting,)
just by getting others to agree that he or she is eminent, regardless of actual skill and
performance. Securing such agreement is much easier than might be expected.

There is no doubt that a lot of practice is needed to develop chess skill and knowledge is important. A club player
with many years experience would easily beat a future world champion who had just
learned the moves. Some studies suggest that chess experts report having had more practice
than non-experts. Furthermore, the case of the Polgar sisters supports the importance of and
power of extensive practice. All three were home schooled and were exposed to much chess
study from an early age, reportedly studying chess for five to ten hours a day. Two became
grandmasters and one (Judit) reached the top ten list. However, perhaps they had a lot of natural
talent to start with, coming from an intellectual background. Furthermore, the sisters
themselves apparently believe in natural talent for chess, stating that they thought Sofia had
the greatest natural talent for chess of the three. As Zsusza put it in a 2005 Psychology Today
interview, “Everything came easiest to her [Sofia]”.

Some research studies suggest some link between IQ and chess skill, but samples in such studies
usually are small and have few or no master level players. However, in 1987 German
newsmagazine Der Spiegel arranged for a team of psychologists to test Garry Kasparov’s
abilities. He indeed had a high but not spectacularly high IQ score (well over 120) and an excellent
memory, needed no doubt to memorise countless opening variations.

Aims of the survey

The survey had several aims. The first was to look at the link between amount of practice and rating development with a larger and more skilled sample. Previous studies usually have small samples and few or no master players. The second aim was to get more data on career histories; why people drop out for example and whether players actually do hit
and try to get beyond performance peaks. I also have a computerised database of all FIDE ratings from 1970, which will help in analysis. The third aim was to get players’s views on some of these issues.

Survey results

The sample to date consists of 581 players, including 103 with a master title (five
grandmasters, 25 international masters, 67 FIDE masters, two woman’s grandmasters, two
woman’s international masters, and two woman’s FIDE masters). The entire sample had an
average peak FIDE rating of 2153, and for the 103 with a master title (g, m, f, wg, wm or wf)
it was 2375. Their median age of learning the moves was 8 years old, and 6 for those with a
master title. The median age of starting serious play and/or playing in a rated tournament was
14 (12 for the masters). Most had received coaching at some stage and almost all had
belonged to a chess club. Only about 21.5% still were playing frequently in FIDE rated
tournaments. Very few said this was because they had lost interest or found the competition
too severe or because they had hit a performance ceiling. The most common reason (43% of
581 players) was that they had too many other commitments. Another common reason was a
lack of nearby rated tournaments. Other reasons were “Stress”, “I was not able to focus on the
game”, “Might be able to improve but its too much work”. “Cost”, “Get less competitive with
age”, and perhaps a perceived lack of natural talent; “I’m rubbish”.

The median number of hours of study of chess in an average week was 4.5 hours (6
hours for the masters) but this varied widely; from 0 to 50 hours. There was a correlation
between amount of deliberate practice from the age of serious play and peak rating (and with
current rating). However, the correlation was quite small, suggesting that other factors are
more important. Another question was whether players had hit rating performance
asymptotes and had undertaken special practice to try to get beyond it. Only 195 of 581 players
said that they had and 129 of these had tried to get beyond it. For those that had, about half
said that they were successful in overcoming the apparent ceiling.

Most respondents are firm believers in the existence of natural talent for chess and believe
that the top ten players have a special combination of traits. Many believe that study can
propel a player a long way but that few can reach grandmaster or top ten level.

Table 1

Percentage selecting each alternative.

Believe top 10 have more
natural talent

Yes 83.99

No 6.37

Unsure 8.09

No answer 1.55

Believe that anyone can
reach GM or top 10 level

Yes 32.7

No 51.63

Unsure 13.94

No answer 1.72

Some comments were as follows:

“I think there is a ceiling given by your natural talent.”

“Yes if they start at a very young age. Top ten is much harder .... you need a lot of talent and
luck besides the hard work. “

“Grandmaster yes, but top ten no.”

“Chess can be a nightmare if you are not gifted and want to improve.”

“Chess is 90% hard work and 10% talent.”

“I have two sons. I’ve both [taught] them chess exactly the same way. One has inherited my
talent (a good amateur) the other is extremely talented and a possible master.”

“There is always an element of natural ability in the top players.”

“Perhaps most could reach GM with a Polgar upbringing, but very few could reach the top
ten.”

“Everyone can reach about FM title.”

‘Definitely not... Top players clearly possess attributes that can’t be taught or learned,
however early one starts.’

‘Top ten is much harder... you need a lot of talent and luck besides the hard work.’

‘Study is necessary but not sufficient... most ... people can get to around 2100 with study but I
see a lot of players (like me) who get stuck there.’

‘Not GM level but most players can reach 2300 ... if they study and work a lot.’

‘Absolutely out of the question. I have seen many players working like professionals but not
reaching the same heights as a more lazy but talented player.”

Most respondents perceived that natural talent for chess consisted of a mixture of
ability and personality factors.

Table 2

Percentage agreeing that natural talent consists of each trait.

High motivation 70.7
Ability to concentrate 69.2
Extreme competitiveness 68.8
Creativity 64
Mental speed 59.7
Spatial ability 48.9
High IQ 37
Other 24.6

In the “other” category, respondents cited pattern recognition, logic, intuition, ability
to control one’s emotions, a sense of danger, and psychological hardiness. The most cited
“other” characteristic (cited by 50 respondents) was a good memory, needed to learn and
update vast amounts of chess information. As one respondent put it, “Memory, above all!”

Some further comments about the perceived traits of the top ten were as follows:

“They often have a great memory and high IQ.”

“They have a better memory, calculating ability and visualisation skills.... the top ten players
are able to develop these skills better and more efficiently.”

“Chess mastery demands a high natural skill in at least 10-15 different and distinct fields
(calculation, spatial skills, competition, concentration, patience, etc). Top ten players are
highly talented in almost each of these skills.”

“They have ... the killer instinct.”

“These players have the ability to create.”

“They are above all hard workers.”

“Most of the top chess players I’ve met have not struck me as having great intelligence. .. Most are quite other worldly.”

“Talent, psychological prowess, character are necessary to make it to the top.”

“... decision-making and organising your priorities.”

The top ten have a much better understanding of chess than others.”

“Certainly becomes clear from analysing with them after a game. It is virtually impossible for
an intelligent and decent IM to keep up with them at times.

“They seem to have an innate feel for position and what might or might not work.”

“Plenty of GMs and IMs work very hard at their chess but never get to top ten level.”

And on how skill is best developed, most believed that both study and playing rated
games are equally important.

Table 3

Percent agreeing which is more important

Rated games more important 14.29

Study of chess material more important 14.29
Both are equally important 70.23
No answer 1.2

Some comments were:

“Need to start early, play and study frequently, analyse games, and have a mentor to
determine your mistakes.”

“You can reach a certain basic level with studying but at some point you really need to play to
get better.”

“Not enough to have the theoretical knowledge, you have to have practical experience with it
also.”

“The most important is effortful study - going beyond one’s limits.”

“At master level , studying and preparing for specific opponents have become increasingly
important.”

“... serious analysis of one’s own games and opening repertoire is also critical.”

“Chess improvement is only possible with an intense study of chess material.”

“Not enough to have theoretical knowledge, you have to have practice experience with it
also.”

“...chess is very much about absorbing bits of information and being able to process it
naturally - players learn more by playing and doing post-mortems.”

“I find I need to play regularly to maintain form.”

“Players will tend to stagnate without study.”

“I played in FIDE tournaments without study for years and my strength slowly declined.
Chess improvement is only possible with an intense study of chess material..”.

“Playing motivates you to study more.”

“Playing is more important than studying. Not every strong player (2400+) studies the game.”

“You have to analyse your performances ... you learning from doing it. The off-board training
augments this.”

“For me, studying the art of calculation raised my level about 100 elo points.”

“Personal trainer [who] shows one’s errors. Otherwise its very difficult to proceed, at least
above certain levels.”

“The right mentor ... is of paramount importance. Reviewing one’s games with that person is
very important.”

“One quality is to be easily amused... a lot of practice ... is quite mechanical and boring.”

“Lots of ... blitz...makes you a better player.”

“It is fundamental to start playing serious chess when you are under 10.”

FIDE rating data study

Two recent studies of mine using a database of all FIDE ratings supports the widespread views of most players
above. The studies suggest that top players have greater natural talent, the effects of which are
evident early, and that natural talent constrains ultimate performance level.

The first FIDE rating list came out in 1970 but FIDE only has computerised rating
lists from 1990 onwards. I spent three years computerising the 1970 to 1989 lists from dusty
issues of Chess Informant and trying to resolve countless inconsistencies in all the lists. The
resulting database is not perfect but is a treasure trove of information about expertise
development.

In the first study, I selected the 3471 players first on the list from July 1985 (when
FIDE first recorded numbers of game in each rating period) to July 1989 and looked at their
rating development up until July 2006. Players first on the list in this period were very
selected, males needing a minimum rating of 2200 to get and stay on the list. For all 591
players who entered the list from July 1985 onwards and who had gained the grandmaster
title by July 2006, the median number of FIDE rated games needed to become a grandmaster
was 390. About 90% of grandmasters had gained the title after about 740 rated games. The
median total number of games of all players first on the list from July 1985 to July 1989 was only
109 games and so most players do not play anywhere near enough to have a realistic chance
of becoming a grandmaster. Of the players who had played at least 740 games, about two
thirds had gained the grandmaster title. Of the remaining 89 players, about 83% had taken the
international master title but on average could not seem to get above 2400 level. But the finding is hard to interpret. It may be that practice eventually will overcome all obstacles or just that only the every talented play more than a few hundred games.

Diagram 1 shows the rating development of all players first on the list between July 1985
and July 1989 who had played at least 1050 FIDE rated games. Ratings in 50-game categories
are given, up until the July 2006 list. There were five eventual top ten players, 73 other grandmasters, and 30 non-
grandmasters. Because everyone sometimes plays above or below his or her rating, only
averages are given to smooth out variability. All these players are highly active and seem to
have gotten near their performance limits.

Diagram 1

Several points are evident. First, players destined for the top ten are identifiable from
list entry. They got on the list younger on average than the others, at a median age of 16, and
at a higher level, and got the grandmaster title faster and with fewer games than the other
grandmasters. Second, the diagram suggests that most players strike a barrier after around 750
games which seems insurmountable. Even after nearly 20 years and extensive practice, the
non-grandmasters on average were stuck. They had not reached the level at which the eventual top ten players
had started. Now one could argue that they were happy to stay at these performance levels,
and did not undertake the difficult deliberate practice needed to ascend higher. But there is no
evidence for this. Third, the overall pattern of differences between the groups is the same. There is a
small initial difference which progressively widens and then stays very large and constant.
This pattern has been found in a few other domains when huge amounts of practice are
involved, when people get to their performance limits.

The second study involved looking at the five most active players from July 1985 to
October 2007, those who had played the most rated games in this period. All are grandmasters and all came
on the list before July 1985. One was playing in international tournaments in the 1960s. All
played a minimum of 2300 FIDE rated games from 1985, which is a lot. Garry Kasparov
played only 761 games in the same period.

Diagram 2

Diagram 2 shows that these five players on average seemed stuck at a performance
ceiling from about 850 post-1985 games, which covers nearly 20 years. Their ratings
fluctuated within a 100 point range but none could get into the top ten. Now again one could
argue that they just did not undertake the difficult deliberate practice needed to ascend
further, but again there is no evidence of this. Furthermore, players so active would be
passionate about chess, and would spend much time in chess activities. If they believed that
becoming a candidate or world champion was just a matter of more practice, surely they
would have undertaken it.

Conclusion

The study suggests that natural talent for chess almost certainly exists and limits
ultimate performance level. It probably consists of a number of characteristics, both ability
and personality. Players do get to performance ceilings after extensive practice. A special
kind of practice might help overcome such a limit but ultimately natural talent probably sets a
peak level. However, in chess and in other domains, most people probably do not get
anywhere near the limits of their talent because they do not get enough practice. A great deal
of study and practice can propel many players further than one might have thought, but not
necessarily to grandmaster level. And, as one respondent suggested, perhaps “There is a lot
of untapped natural talent in the world. If everyone ... took up chess ... the top players would
be closer to FIDE 3000 than 2800.”