The paper tackles a long-standing problem in algebraic geometry about the structure of a special class of geometric shapes called Fano varieties. These are smooth geometric spaces with a certain built-in curvature property that makes them, in a loose sense, positively curved. The conjecture in question, proposed by Fujita, predicts that Fano varieties can be classified into a specific short list of types once you measure their geometric complexity using a quantity called the pseudoindex, which tracks the smallest possible "degree" of curves that can be drawn on the space.
The proof strategy begins with a tool from a different area of geometry called Gromov-Witten theory, which counts how many rational curves (the simplest kind of curves, topologically shaped like a sphere) pass through pairs of points on the variety. The authors use this counting machinery to establish that there are only finitely many basic families of these rational curves that behave in a controlled, "unsplit" way, meaning the curves cannot be broken into simpler pieces. This finite collection of curve families becomes the backbone of the entire argument.
With that foundation in place, the authors run a step-by-step construction, controlling the geometry at each stage using the degree bounds those curve families provide. Because every family encountered during the construction stays unsplit, classical results about how Fano varieties can be contracted or fibered (due to Fujita and others) apply cleanly at each step. The inductive argument eventually closes, confirming that every smooth complex Fano variety satisfying the conjecture's numerical condition must belong to the predicted short list of geometric types.