Three in Four Female Chess Players Faced Bias or Harassment — and the Gap Behind FIDE's Survey

In 2026, inside a tournament hall in Da Nang, I sat behind a thirteen-year-old girl for four hours. It was the last round of a junior event, a rook-versus-knight endgame, and she won with a mating move I had to replay twice before I understood it. Four years later I asked after her through her old coach. She had left the board at seventeen. Not because she was losing. She told him something I wrote down word for word: "I am tired of having to explain, at every tournament, why I am here."

That sentence came back to me while reading the report on the International Chess Federation survey: three in four female chess players have faced gender bias or harassment. I read it three times, then realised I was searching for something the article never contained — the number of respondents.

The text carried exactly one substantive fact: the three-quarters figure, attributed to a survey conducted by FIDE. The rest of the page was advertising copy and a privacy notice — the fingerprint of an aggregator, where the original piece has been stripped of the one thing that matters: method.

Surveys of this kind are usually distributed online, in several languages, through FIDE and national federation channels, on an opt-in basis. Those three features combine into a familiar outcome: people who have lived through something unpleasant are more likely to answer than people who have not. The measured rate therefore tends to run above the true rate across the whole population. That is a technical point, not an accusation.

One more clarification: the headline folds "gender bias" and "harassment" into a single division. The two differ sharply in nature and in consequence. The full report almost certainly separates them. The headline does not.

I do not have the primary report. The publication date is unverified. That is why I write this carefully rather than angrily.

What this survey measures is treatment, not playing strength. That boundary matters and it is the easiest one to cross. A high rate of bias says nothing about the level of women players, and anyone using it to infer that women play worse is misreading a sociological instrument as if it were a rating list.

The evidence sits elsewhere. Judit Polgar, on the July 2026 rating list, reached 2735 — the highest mark ever recorded by a woman, and she was once among the eight strongest players in the world. If a biological ceiling existed, 2735 would have no place to stand. What exists is a leaking pipeline.

The gap between the women's number one and the open number one has generally sat between 150 and 250 Elo points in recent years. I have followed women's events and open events with women's sections for more than thirty years, and what I see is not girls playing worse at the same age. What I see is fewer of them with each cohort. A smaller pool produces fewer elite players — simple arithmetic, and it hurts more than any explanation.

Based on my experience covering matches at national junior events, the sharpest drop sits between twelve and eighteen. That is the age when a young woman must choose between staying in an environment that requires her to justify her presence, and leaving. People talk a great deal about role models. The problem is the middle layer: the 2200 to 2400 players, unknown, uncontracted, who hold up the entire system.

In Vietnam the picture is starker because the pool was always thin. For years the women's game here rested on one or two names carrying almost the whole international front, most visibly Pham Le Thao Nguyen. Behind them, silence. Hoang Thanh Trang, born in Hanoi, earned the grandmaster title and played for Hungary — a case showing that talent was never scarce, only the road was narrow.

There is also a question of reach. FIDE governs the events it licenses and the officials it certifies. Most incidents plausibly occur in clubs, in amateur events, and on online chess platforms — places where FIDE holds no jurisdiction. A FIDE survey therefore documents a problem FIDE only partly controls.

And the separation must be absolute: a prevalence rate is an estimate of how widespread something is. It is not a finding of misconduct against any identifiable person. Merging the two is the single largest legal and ethical error in this whole story.

The blind spot lies in how we frame the question. The chess world is arguing about three in four, when the heavier question is how many left, and at what age. A survey measures the experience of those who stayed. Those who quit do not answer federation questionnaires.

So I suspect three in four is a floor, not a ceiling. Whistleblower protection is thin in most national federations, and without confidential intake and anti-retaliation clauses, people stay quiet. Silence always pushes the measured rate below the real one.

A backlash cycle will follow. Critics will seize on the missing sample size, the missing date, the missing definitions — and they will be technically right, because the article handed them too easy a target. When that happens, the people who pay are the ones the survey set out to protect.

What I want to see in the next twelve months is not another percentage, but a document with a sample size, a date, a separate breakdown for bias and for harassment, and a commitment to measure again in two years. One measurement is news. Repeat measurement is governance.

And for that girl in Da Nang, I keep an odd conviction: the chessboard is the one place where a child can stand opposite an adult and not ask permission. Until we can hold that open for them, every rating list remains half a story.