The paper tackles a fundamental question in algebraic geometry about the complexity of certain geometric spaces called conic bundle threefolds. A conic bundle over the plane is essentially a three-dimensional space where, above almost every point in a flat plane, there sits a conic curve (something like a circle or hyperbola). The "discriminant" refers to a curve in the base plane above which the conic degenerates in some way. The central question is whether these spaces are "unirational," meaning roughly whether you can parameterize almost all of their points using simple rational functions from ordinary Euclidean space.
The authors prove that when the discriminant curve is sufficiently complicated (degree at least 18 and chosen generically), the conic bundle cannot have a "rational multisection of degree two." A rational multisection of degree two is essentially a way of selecting two points from each fiber (each conic) in a manner controlled by rational functions. Ruling out such selections is a concrete obstruction to the space being unirational, because any unirational parameterization would eventually produce such rational multisections.
This result is a first step toward resolving a longstanding open conjecture by the mathematician Iskovskikh, who predicted that some conic bundle threefolds are not unirational. This conjecture matters because rationally connected varieties, a broad class of geometric spaces where any two points can be connected by a rational curve, are all suspected but not proven to sometimes fail to be unirational. Confirming Iskovskikh's conjecture would provide the first proven example of a rationally connected variety that resists rational parameterization, revealing a fundamental gap between two important notions of geometric simplicity.