Algebraic geometry studies geometric objects defined by polynomial equations, and one of its central goals is to classify and relate different geometric spaces. Del Pezzo surfaces are a particularly nice class of two-dimensional geometric objects, like the ordinary sphere or more complicated curved surfaces, defined over a field (think of the real or complex numbers, though the paper works in greater generality). The condition of "Picard rank one" is a technical way of saying these surfaces have a relatively simple structure in terms of how curves sit inside them.
Birational maps are a core tool in this classification program. Two surfaces are called birationally equivalent if you can pass between them using rational functions, even if the map is not perfectly well-defined everywhere. Sarkisov links are the fundamental building blocks of such maps, analogous to elementary moves or atomic operations. A major program in the field, the Sarkisov program, says any birational map between suitable surfaces can be broken down into a sequence of these basic links, but it does not say how many you need.
The main result of this paper is a sharp and elegant bound: if two del Pezzo surfaces of Picard rank one are birationally equivalent, you never need more than two Sarkisov links to get from one to the other. This is a strong structural statement, saying the birational complexity between such surfaces is tightly controlled. It streamlines and deepens our understanding of how these surfaces relate to one another, and the result holds over any perfect field, making it broadly applicable across different number-theoretic and geometric settings.