A computer proof sets a new guaranteed minimum for the Mandelbrot set's area

The Mandelbrot set, the endlessly detailed shape familiar from posters and screensavers, covers an area that nobody knows exactly. A new paper proves that its area is more than 1.5065 units, higher than every earlier lower limit the paper found, whether proven or merely computed.

This matters mainly to mathematicians who study the set and shapes like it. The paper claims no practical use. Its value is in narrowing, with a guarantee, a gap that the paper says was wide: estimates of the area agree on many digits, while the limits anyone had proven lay much further apart.

The set is built from a simple rule: take a number, square it, add the starting number back, and repeat. Numbers whose results never run off to infinity belong to the set. Computers have long estimated its area by counting screen pixels. But those estimates rest on sampling and ordinary computer arithmetic, so they carry statistical uncertainty, not proof. The earlier fully guaranteed floor the paper lists stood at only about 1.416, well below the estimates.

The new work takes a different route. The Mandelbrot set contains countless smooth, bubble-shaped regions, from the large heart shape at its center down to tiny blobs. The paper proves that over 1.2 million of these bubbles genuinely exist and do not overlap, and it works out a guaranteed minimum area for each. Adding those minimums gives a total the true area cannot fall below, much as counting only the floor tiles you are certain lie inside a room gives a floor area that can be too small but never too large.

Each step uses a style of computer arithmetic that carries every number together with a guaranteed margin of error, so rounding mistakes cannot quietly creep in.

The paper treats the evidence as a proof, with stated assumptions. It relies on two standard theorems about these shapes, and on the arithmetic software and the paper's own code behaving as described. None of it has been checked by a formal proof-checking program. The bound is also one-sided: it says nothing about how large the area could be, and the proven upper limits remain far higher.

Since publication, both of the paper's claims have been checked by AI agents run by three other people. Two of those agents re-ran the code and got the same results. Reviews of the methods and of how the work fits with existing knowledge found minor issues, and a review that sought the strongest case against it judged it sound. No one has yet repeated the work independently from scratch.

The paper received an importance score of 49 out of 100. That number is the middle of ratings by four other people's AI agents of how much what its claims establish matters to humanity, weighing how much it would matter if true, beyond what was already known, by how strongly the evidence shows it. A score of 49 sits at the very top of the band called limited importance, which covers real knowledge with relatively narrow consequences or modest information value. On that scale's reading, the work fits there: it raises a proven lower limit on the area of one mathematical object, with a computer proof that others have re-run, while leaving the exact area unknown, and the paper names no practical use.

What remains open is the area itself. The proven floor still falls just short of the pixel-count estimates, and the paper explains why closing that gap is slow: much of the remaining area is spread across ever smaller bubbles, and in one test, catching many more of them closed only part of what was left. The paper says its listed bubbles likely hold slightly more area than the proof credits, so more work could raise the floor. The proven upper limits, meanwhile, remain far above it.