Or, in similarly simplified headlinese, “Math finds the best doughnut”. A little bit more precisely, Fernando C. Marques and André Neves claim in a preprint on the arXiv to have proved the Willmore conjecture, that the minimum achievable mean curvature of a torus is $\frac{2}{\pi^2}$.
The article I linked to is some surprisingly non-stupid coverage from the Huffington Post. It seems they have a maths professor writing a column. I will never understand that site. I don’t know if there’s a Serious Business way of framing this, but the result is nice to know.
Richard Elwes has written a very short post on Google+ with some more real-maths information about what’s going on.