If you have kids, you've probably thought hard about how to name your child. Should you choose a 'special' kind of name, or rather a very trendy or well-known one? Popularity is an important aspect when it comes to choosing virtually anything. The same goes for chess openings: do you want to go for popular main lines or for 'off beat' variations? Both options come with their own advantages and disadvantages. A trendy baby name may suggest a popular baby - and trendy parents - but a special name will suggest a special child. Likewise, playing a trendy chess opening will imply theoretical knowledge and opting for an obscure variation will suggest more independence and creativity. A lot has been written about the popularity of baby names (see, for example, the chapter 'What's in a name' in Steven Pinker's recent book
The Stuff of Thought and the chapter 'Would a Roshanda by any other name smell as sweet' in the book
Freakonomics by Levitt and Dubner).
But why do baby names (and chess openings) start to
lose their popularity at some point? Is it merely that more and more people become 'fed up' with it, or is there more going on?
Recent research in the
Proceedings of the National Academy of Sciences (May 2009) on baby names suggests a solution: the
rate at which a cultural trend becomes popular, is also indicative of its decline. As one
Scienceblogs article on the research put it: "the faster the rise to prominence, the steeper the fall from grace." The researchers, Jonah Berger and Gael Le Mens, looked at the changing popularities of first names in France and the USA over the last 100 years. They found that parents were less inclined to give their children names that had become very popular very fast, regardless of the overall popularity of these names. The names Tricia and Krisi, for example, became very popular very fast in the 60s, and lost their appeal in the 70s equally quickly, whereas the name Charlene slowly gained in popularity and also declined much slower than Tricia and Kristi. The reason? "Fads are perceived negatively, so people avoid identity-relevant items with sharply increasing popularity because they believe that they will be short lived."Of course, chess openings are much more than just a cultural trend. In chess, unlike fashion, popularity is not merely subjective - it's also highly
objective: a losing line won't be played much. That's why nobody plays the Damiano Defence (1.e4 e5 2.Nf3 f6?) these days. A 'hard refutation' can simply kill a chess line's popularity instantly. But what about lines that are not known to be refuted, or even known to be inferior? Might their popularity also be dependent on the idea that they might be 'short lived'? I have always wondered whether adopting a particular variation isn't also part of one's
image. Apart from any objectivity - isn't it
cool to play popular lines? I realize this is a different point of view that most chess players will think of. They will, of course, accept that chess lines are subject to change in popularity, but my guess is they will try to analyse this change in terms of chess theoretical developments: novelties, strong players starting to experiment with the line, matches being played with it, etc. Indeed, in a
recent discussion on this site, many posters gave reasons for why certain openings fluctuated in popularity over time. But these were all reasons to do with
chess. In this article, I want to see if there are
other reasons for adopoting a particular line or not. I want to raise the possibility that chess openings, like baby names, can be a
cultural trend, subject to sociological and psychological forces. Take, for instance, the 7...Qc7 line of the Winawer variation of the French Defence, a pretty cool and exciting line which I have analysed myself a lot of times:
1.e4 e6 2.d4 d5 3.Nc3 Bb4 4.e5 c5 5.a3 Bxc3 6.bxc3 Ne7 7.Qg4 Qc7In a recent issue of
ChessVibes Openings, Merijn van Delft and Robert Ris noted that this line became increasingly inpopular in the 90s, even though nobody seemed to know why. In top grandmaster games, Black was usually fine after the opening, and Black even seemed to score better than usual with this line. But all of sudden, Black players just switched to 7...0-0 without an apparent reason. (Now, by the way, 7...Qc7 seems to be back on the scene.) What was going on? If the reasons for the change couldn't be related to particular objective developments such as games or novelties, maybe there were other, more subjective different reasons. I found myself wondering whether Berger and Le Mens' conclusions also played a role in the popularity decline of this particular variation. Could the mere
speed with which the variation became popular be of influence to its eventual decline? To test this, I decided to do a little research of my own. In ChessBase MegaBase 2008, I counted all games in which the diagrammed position occurred between 1980 and 2004 over a reperiod of 5 years. For instance, in the period 1980-1984, the position occurred in 47 out of 96230 games, i.e. 0,048%. In 2000-2004, the position occurred in 737 out of 1173157 games, i.e. 0,062%, and so on. To make sure I hadn't overlooked some kind of 'refutation' of the line, I also checked the score results of these games, but I didn't find any strange results compared to the overall results of chess games in this period. (The average was around 54% for White.) This was the easy part. Chess databases are a real gold mine for this kind of research, actually, but the problems only start here. One important question is: how to measure the popularity of a particular opening variation? First, I decided that it would be best to take
all Qc7-Winawer games played - not only the ones played by grandmasters or titled players. This may be surprising to some readers, but I believe I had good reasons for it: 1. the number (N) of games would be bigger, and thus more likely to say anything significant; 2. there is no reason to assume that the popularity of a line doesn't go beyond the top level of chess players; 3. Berger and Le Mens research also didn't exclude particular groups or classes of people; and, most importantly, 4. because including all games in the sample rules out the possibility that a certain small group of players (namely, the elite) simply became bored with it or couldn't surprise their (also elite) opponents with it anymore. Thus, the popularity of the line (P) in a particular period would simply be defined as the percentage of games played with that line of
all the games played. By comparing periods, something could be said about the rate of decline and increase of popularity over time. (One possible problem with this method is that comparing different time periods may not be such a straightforward task in chess databases, since these databases contain vastly more games, and by weaker players on average, for later periods due to the rise of digital game storage and the internet.)I must confess I was pretty excited when I starting counting and found that in the period 1980-1990 (a period for which I recalled the Winawer becoming more and more popular), the popularity of the Winawer variation indeed seemed to increase slowly but steadily until around 1986. In 1980, P was 0,006%, in 1986 it had gone up to 0,019%. But in 1987 it suddenly took flight: P for 1987 was 0,028%, for 1988 it was 0,038% and in 1989 it had grown to 0,042%. Within a three year's period, the popularity of the Winawer had more than doubled! And then, in 1990, the popularity of the Winawer suddenly sunk back to 0,017%. Here are the results in a graph:
Sure enough, these data seem to support the conclusions of the baby name research. Here, too, we see a sharp decline in popularity
right after a sharp
increase in the preceding period. After that, the variation has never again been as popular as it was then. But as usual, doubt soon crept in. Even assuming the
results of the opening (on top level) didn't affect the overall popularity of the variation (if anything, the variation scored
better in the 5 year period right after it was abandoned), what did the data tell me, exactly? In the baby name research, the popularity of a name was measured in comparison to the popularity of
other baby names. But my little research didn't say anything about other opening lines (such as the Rubinstein French, or indeed the Sicilian). To be able to draw any conclusions, I had to include other opening lines in my data as well. As co-editor Merijn van Delft pointed out to me recently, you can't measure the popularity of the 6.Be3 Najdorf without taking into account the status of the Poisoned Pawn variation, which arises after 6.Bg5! It soon became clear that this would be an enormous amount of work - interesting work which, regrettably, has to wait until further notice.
But even so, I think it's definitely possible that in the Winawer's case, something similar like baby naming has occurred. Of course, my data does not say anything about the
reason of the Winawer's sudden decline, but what the data does show, in my opinion, is that the 'sharp peak' phenomenon can also occur in the choice of chess openings
in principle. I'm sure it can also be shown that there are variations that gain in popularity slowly, and then again decline slowly as well, but that's not the point. Further research is necessary to validate the hypothesis that the baby name effect is present in chess, too. Berger and Le Mens speculate that marketing and technology play possibly play important roles in cultural extinctions, and the same could be true of the Winawer, especially with the rise of computer chess around 1990. But as the author of the hyperlinked Scienceblog article remarks:
[T]heir work suggests that regardless of these external influences, newly popular items are swayed by internal forces that limit their own stay at the top. They predict that these effects should be much stronger in areas like names, where cultural tastes are used to communicate our identity. The clothes we wear, the cars we drive and the gadgets we flaunt would fall into this category too, while our choice of refrigerator or bathroom tiles might not.
But does our personal choice of chess openings 'communicate our identity'? I think it does - on all levels. I've always wondered whether Morozevich's or Nakamura's choices to play obscure openings weren't partly a way to create some kind of 'rebel image' that could serve them well - both in terms of more invitations and in terms of popularity among amateurs. On a more down-to-earth level, I myself remember how hip I thought I was when I started playing the King's Indian, following in the footsteps of my hero Kasparov. Other chess players that I know think it's particularly cool to play lines that nobody else plays. In my local chess club, there are even people who
dismiss other players merely for the openings they play.I used to be a fan of the hardcore thrash metal band
Slayer. Everybody else I knew hated their music - for some reason, it felt great. In the same way, there may be more to the popularity of chess openings than just their correctness.
Jonah Berger and Ga?´l Le Mens (2009). How adoption speed affects the abandonment of cultural tastes Proceedings of the National Acedemy of Sciences
If you have kids, you've probably thought hard about how to name your child. Should you choose a 'special' kind of name, or rather a very trendy or well-known one? Popularity is an important aspect when it comes to choosing virtually anything. The same goes for chess openings: do you want to go for popular main lines or for 'off beat' variations? Both options come with their own advantages and disadvantages. A trendy baby name may suggest a popular baby - and trendy parents - but a special name will suggest a special child. Likewise, playing a trendy chess opening will imply theoretical knowledge and opting for an obscure variation will suggest more independence and creativity. A lot has been written about the popularity of baby names (see, for example, the chapter 'What's in a name' in Steven Pinker's recent book The Stuff of Thought and the chapter 'Would a Roshanda by any other name smell as sweet' in the book Freakonomics by Levitt and Dubner). 
But why do baby names (and chess openings) start to lose their popularity at some point? Is it merely that more and more people become 'fed up' with it, or is there more going on? Recent research in the Proceedings of the National Academy of Sciences (May 2009) on baby names suggests a solution: the rate at which a cultural trend becomes popular, is also indicative of its decline. As one Scienceblogs article on the research put it: "the faster the rise to prominence, the steeper the fall from grace." The researchers, Jonah Berger and Gael Le Mens, looked at the changing popularities of first names in France and the USA over the last 100 years. They found that parents were less inclined to give their children names that had become very popular very fast, regardless of the overall popularity of these names. The names Tricia and Krisi, for example, became very popular very fast in the 60s, and lost their appeal in the 70s equally quickly, whereas the name Charlene slowly gained in popularity and also declined much slower than Tricia and Kristi. The reason? "Fads are perceived negatively, so people avoid identity-relevant items with sharply increasing popularity because they believe that they will be short lived."Of course, chess openings are much more than just a cultural trend. In chess, unlike fashion, popularity is not merely subjective - it's also highly objective: a losing line won't be played much. That's why nobody plays the Damiano Defence (1.e4 e5 2.Nf3 f6?) these days. A 'hard refutation' can simply kill a chess line's popularity instantly. But what about lines that are not known to be refuted, or even known to be inferior? Might their popularity also be dependent on the idea that they might be 'short lived'? I have always wondered whether adopting a particular variation isn't also part of one's image. Apart from any objectivity - isn't it cool to play popular lines? I realize this is a different point of view that most chess players will think of. They will, of course, accept that chess lines are subject to change in popularity, but my guess is they will try to analyse this change in terms of chess theoretical developments: novelties, strong players starting to experiment with the line, matches being played with it, etc. Indeed, in a recent discussion on this site, many posters gave reasons for why certain openings fluctuated in popularity over time. But these were all reasons to do with chess. In this article, I want to see if there are other reasons for adopoting a particular line or not. I want to raise the possibility that chess openings, like baby names, can be a cultural trend, subject to sociological and psychological forces. Take, for instance, the 7...Qc7 line of the Winawer variation of the French Defence, a pretty cool and exciting line which I have analysed myself a lot of times:1.e4 e6 2.d4 d5 3.Nc3 Bb4 4.e5 c5 5.a3 Bxc3 6.bxc3 Ne7 7.Qg4 Qc7
In a recent issue of ChessVibes Openings, Merijn van Delft and Robert Ris noted that this line became increasingly inpopular in the 90s, even though nobody seemed to know why. In top grandmaster games, Black was usually fine after the opening, and Black even seemed to score better than usual with this line. But all of sudden, Black players just switched to 7...0-0 without an apparent reason. (Now, by the way, 7...Qc7 seems to be back on the scene.) What was going on? If the reasons for the change couldn't be related to particular objective developments such as games or novelties, maybe there were other, more subjective different reasons. I found myself wondering whether Berger and Le Mens' conclusions also played a role in the popularity decline of this particular variation. Could the mere speed with which the variation became popular be of influence to its eventual decline? To test this, I decided to do a little research of my own. In ChessBase MegaBase 2008, I counted all games in which the diagrammed position occurred between 1980 and 2004 over a reperiod of 5 years. For instance, in the period 1980-1984, the position occurred in 47 out of 96230 games, i.e. 0,048%. In 2000-2004, the position occurred in 737 out of 1173157 games, i.e. 0,062%, and so on. To make sure I hadn't overlooked some kind of 'refutation' of the line, I also checked the score results of these games, but I didn't find any strange results compared to the overall results of chess games in this period. (The average was around 54% for White.) This was the easy part. Chess databases are a real gold mine for this kind of research, actually, but the problems only start here. One important question is: how to measure the popularity of a particular opening variation? First, I decided that it would be best to take all Qc7-Winawer games played - not only the ones played by grandmasters or titled players. This may be surprising to some readers, but I believe I had good reasons for it: 1. the number (N) of games would be bigger, and thus more likely to say anything significant; 2. there is no reason to assume that the popularity of a line doesn't go beyond the top level of chess players; 3. Berger and Le Mens research also didn't exclude particular groups or classes of people; and, most importantly, 4. because including all games in the sample rules out the possibility that a certain small group of players (namely, the elite) simply became bored with it or couldn't surprise their (also elite) opponents with it anymore. Thus, the popularity of the line (P) in a particular period would simply be defined as the percentage of games played with that line of all the games played. By comparing periods, something could be said about the rate of decline and increase of popularity over time. (One possible problem with this method is that comparing different time periods may not be such a straightforward task in chess databases, since these databases contain vastly more games, and by weaker players on average, for later periods due to the rise of digital game storage and the internet.)I must confess I was pretty excited when I starting counting and found that in the period 1980-1990 (a period for which I recalled the Winawer becoming more and more popular), the popularity of the Winawer variation indeed seemed to increase slowly but steadily until around 1986. In 1980, P was 0,006%, in 1986 it had gone up to 0,019%. But in 1987 it suddenly took flight: P for 1987 was 0,028%, for 1988 it was 0,038% and in 1989 it had grown to 0,042%. Within a three year's period, the popularity of the Winawer had more than doubled! And then, in 1990, the popularity of the Winawer suddenly sunk back to 0,017%. Here are the results in a graph:
Sure enough, these data seem to support the conclusions of the baby name research. Here, too, we see a sharp decline in popularity right after a sharp increase in the preceding period. After that, the variation has never again been as popular as it was then. But as usual, doubt soon crept in. Even assuming the results of the opening (on top level) didn't affect the overall popularity of the variation (if anything, the variation scored better in the 5 year period right after it was abandoned), what did the data tell me, exactly? In the baby name research, the popularity of a name was measured in comparison to the popularity of other baby names. But my little research didn't say anything about other opening lines (such as the Rubinstein French, or indeed the Sicilian). To be able to draw any conclusions, I had to include other opening lines in my data as well. As co-editor Merijn van Delft pointed out to me recently, you can't measure the popularity of the 6.Be3 Najdorf without taking into account the status of the Poisoned Pawn variation, which arises after 6.Bg5! It soon became clear that this would be an enormous amount of work - interesting work which, regrettably, has to wait until further notice.
But even so, I think it's definitely possible that in the Winawer's case, something similar like baby naming has occurred. Of course, my data does not say anything about the reason of the Winawer's sudden decline, but what the data does show, in my opinion, is that the 'sharp peak' phenomenon can also occur in the choice of chess openings in principle. I'm sure it can also be shown that there are variations that gain in popularity slowly, and then again decline slowly as well, but that's not the point. Further research is necessary to validate the hypothesis that the baby name effect is present in chess, too. Berger and Le Mens speculate that marketing and technology play possibly play important roles in cultural extinctions, and the same could be true of the Winawer, especially with the rise of computer chess around 1990. But as the author of the hyperlinked Scienceblog article remarks: 