Three in Four: What FIDE's Survey Records About Women in Chess, and the Detail the Report Left Out

August 2026, the Chennai Olympiad. The women's hall, third hour of the afternoon, and the same old habit of mine — watch the hands first, the clock second. A player rested her finger on a knight, held it eleven seconds, and pulled the piece back. The notebook filled faster than the scoreboard. That day India took bronze in both the open and women's sections, lifted the Gaprindashvili Cup, D. Gukesh scored 9/11 for India-2, and the hall roared for eight seconds. I filed 4,200 words and kept 600.

One thing about that hall never made it onto a scoreboard. Three years later, a FIDE survey reports that three in four female chess players have faced gender bias or harassment during their careers. The number is large. There is no room behind it — no name, no date, no sample size.

The portion of the survey that reached me was a secondary report, not the primary document. It carries no sample size, no fielding period, no geographic scope, no questionnaire design, and no link to FIDE's official publication. Among the items listed as key points sat a website's privacy and cookie notice. It is a small signal that says a great deal: this page was not written at a desk that read the document, it was scraped by a machine. The number does not become false, but the scaffolding around it stops being trustworthy.

The survey has to be placed in the structure that produced it. The top women's ratings sit around 2550–2600 today, while the world No. 1 is above 2830 — a gap of roughly 230 to 280 Elo. Fewer than 50 women have ever held the full Grandmaster title, under two percent of all GMs. Judit Polgár's peak reached 2735, Koneru Humpy played above 2600, and still this remains a profession where women are few in number and those who listen to them are fewer.

Inside India the picture splits in two. Tamil Nadu's Velammal school pipeline, the Tal Chess Club lineage in Chennai, and generous state cash awards do produce titles — that machinery genuinely works. In the same state, AICF election cycles are fought over posts every few years, and in the same period another champion quietly appears. After ChessBase India launched in 2026, India acquired the world's largest chess media operation; what it never acquired was a written record. A sport without its own written archive will keep the survey's number, but it will not have a way to attach that number to a name.

This figure measures prevalence, not performance. It tells you how many players experienced something; it does not tell you what that experience did to the board. In international sports journalism that distinction matters, because a policymaker needs both — how many are affected, and through which channel.

The report's biggest methodological weakness is not the sample size but the merging of categories. "Gender bias or harassment" — two very different things folded into one percentage. Bias is a culture problem: being underestimated, having advice imposed, being softened in commentary. Harassment is a safety problem, and its route can be criminal. The remedies differ — training and appointment rules for one, a complaints mechanism and accountability for the other. A merged figure cannot tell an organiser which problem is standing in front of them.

Every sample has a statistical range. If roughly 400 players responded, the margin of error on a 75 percent estimate is about four percentage points; if only 100, it widens to about eight and a half. Without a disclosed sample size, the number indicates direction, not precision.

And selection bias is the largest risk here. A survey distributed through federation and platform channels reaches first the players who are most active — who play tournaments, who are present online, and who are therefore most exposed to both bias and harassment. Prevalence estimates from such self-selected samples typically sit above population estimates. The standard way to test this is to ask male players the same question, so it becomes clear whether the problem is gendered or occupational. That baseline question is not mentioned.

There is no stratification either. Elite and amateur, over-the-board and online, one continent and another — all dissolved into the same 75 percent. For policy, that absence is a bigger problem than the sample size, because without strata you cannot aim.

There is another knot. Much harassment in chess now travels through platform chat, direct messages and social media — territory outside FIDE's jurisdiction. The federation is publishing data on conduct it has limited tools to police. That is not a weakness; it could be an honest acknowledgement of limits. The report does not appear to show that acknowledgement.

In a small, tightly networked profession, reporting also runs in reverse. Where coaches, arbiters and organisers move in the same circle, the most serious incidents get recorded the least — because the person complaining must accept next season's invitation from the same people. In nine years on this beat I have seen that pattern repeatedly: what gets recorded is usually the mildest version.

The channel to performance exists, but this survey does not measure it. Attention breaking during preparation, a rising error rate under time pressure, and above all event selection. Declining invitations, avoiding a city or a hall, withdrawing from mixed opens. This is an inference, not a measured fact, and I will label it as an inference. Every indicator that could turn it into fact is missing: withdrawal rates by gender, participation continuity across a cycle, error rates under time pressure, drop-out by age band.

The real question here is not the number but the emptiness around it. The three-in-four figure will run in newspapers for years, while its fielding date, its sample and its methodology never become public. The cleaner the number, the longer it will be quoted as a static fact — while both prevalence and reporting norms keep shifting. The screen showed the moves and hid the room.

The second thing is that the rhetorical force of the figure makes the section after it invisible. Condensed secondary reporting usually drops the paragraph where the federation states its next steps — complaints mechanism, investigation process, sanctions, training. That is exactly where the actionable information sits. Nobody clapped, and that was the loudest thing in the hall.

For India the next question is direct. The Tamil Nadu pipeline will keep producing a new title every two or three years, and Chennai recovered in December 2026 through Gukesh the crown it lost to Carlsen in 2026 — the loop closed. But the same system that made an 18-year-old capable of carrying a federation's expectations — has it built a complaints route that girls of the same age can trust? Over the next cycle I will watch three things: whether the survey's methodology is published, whether male players' responses surface, and whether withdrawal rates appear in any table. After three days on one board, I had already learned that the real story sits where the camera turns away.