A story by Shreya Raman
I love Chess! Okay, that is an understatement. I am obsessed. My dozen-games-a-day track record is proof enough. So, when Soma said we need to start thinking of ideas for our first projects, it was a no-brainer. It was time to merge (I literally tried doing that with my datasets for this project but failed miserably. Totally ignorable Python joke.) my obsession of Chess with my passion for reporting on gender.
Now, before we start, I do want to tell you all that nothing that I will be talking about here is ground-breaking really. Many have explored these questions that I am exploring and have written about it in much more detail than I could. And most of this report will be me stealing all these people’s work to make a point. (So, please don’t accuse me of plagiarism).
My personal purpose for doing this project was simple. I just wanted to get myself–who had nightmares of a 26k row data file until three weeks ago–to be brave enough to tackle a dataset with 1.1 million rows! And I must say, I did succeed (with lots of help from super kind teaching assistants and mentors), as you can see from all the charts that are embedded in this story. However, to make it easier for me, I have only looked at data for Indian chess players (It was still a file with 100k rows!)
Anyway, enough about me. This story is not about me. It is about players who are actually good at Chess. More than that, it is about the data on these extraordinary people. And most importantly, it is about asking this data, why do you tell us that women are bad at chess?
Do you not believe me? Take a look at this first chart. The International Chess Federation, also known as FIDE, awards titles to chess players based on their performance. While the actual rules and regulations are long (14 pages long, to be precise) and complicated, the titles in order of importance and difficulty are–Grandmaster (GM), International Master (IM), Fide Master (FM) and Candidate Master (CM).
Now back to the chart. This chart shows all Indian players who have these open titles and I have colour-coded them as per sex (Yes. The Chess data world is heteronormative).Clearly, it seems like there are very few female chess players at the top of the game. And I must confess that I have not been completely fair. In addition to these open titles, the Chess Federation introduced women’s titles in 1950. These titles–Woman Grandmaster, Woman International Master, Woman Fide Master and Woman Candidate Master–have lower requirements than the open titles. Also, female players are not forced to compete only in these titles. I tried to calculate including all the women’s titles and the proportions still seem off.
One of the most prominent reasons that all articles (like this one, this one and to some extent this one) state is that it has to do with the proportions of overall players. Because, a larger group leads to higher chances of extreme values (i.e. higher chances of exceptionally good players). Looking at the data, that makes complete sense. Only 16% of the total players that are associated with the Federation are female. And professor Wei Ji Ma, who is also a Fide Master explains this in more detail and in a better way here.
But, many researchers, like David Smerdon (who is also the author of this piece) believes there is more to the story. Before I tell you about more interesting hypotheses, I will add that I also tried to find answers for this question within the data. I knew FIDE had more data than they were sharing in the downloadable form. For example, look at this FIDE profile for the reigning champion Magnus Carlsen. Wow! They have time series data on his performance since 2001. I believed that unlocking this dataset will give me all the wisdom that I need to make confident and accurate postulates.
After receiving no hints or help from the FIDE website, I started looking for an API. And I found one! This beautiful one by a computer engineering student! This one seemed pretty simple. You just make requests based on FIDE IDs and you will magically get all the historic data. I fumbled a bit but thanks to Jessie and Ilena, I figured it out. And now that I have the magical data, I thought I would finally be able to solve this crucial problem that plagues modern chess.
Alas, other than the fact that Indian Grandmasters on average peak around the age of 13, I did not find much.
With that sad story out of the way, we can focus on what real researchers and more intelligent people have to say about this issue. Since you all know David Smerdon, let’s start with his opinions and his work. Smerdon, who is also a Grandmaster(!), says that stereotypes and something called ‘Stereotype Threat’ impacts performance. Stereotype Threat is basically “where minorities underperform solely because they’re aware of a stereotype that people of their group do worse”, in his own words. He says the performance drop is “roughly equivalent to a woman giving her male opponent the advantage of the first move in every single game”. (Here is his paper on the impact of stereotype threat)
Smerdon also cited a very interesting study from 2007 where male and female players were made to play against each other with some players knowing the sex of the opponent and others unaware. The male and female players played equally well when they did not know the sex of the opponents and when they knew, male players performed better than female players.
In addition to patriarchy and structural constraints, this piece by Omkar Khandekar, cites biological or neurological capacities as reasons. But none of these have been proven. While the reasons could be multiple and complex, one thing’s certain that data may not always tell the whole truth. And more than that, it is better to be vary and curious. I will end this story with another chart. This one highlights the individual trajectories for each of the Indian Grandmasters. Not really sure what it is trying to say but I let it stay because my always helpful and sweet mentor, Veronica Penney said it looks interesting! So, please, go ahead! Search for your favourite Grandmasters or sort them by sex, rating difference or their activity status.