Abstract (eng)
This work surveys work on the core of voting rules when the issue space can be identified with some finite dimensional Euclidean space. It turns out that the core is almost always empty. Making this precise has proven to be mathematically
challenging. The most commonly used notion of "almost always" is
the topological notion of being residual. This notion is hard to interpret and depends strongly on the topology one imposes on the space of profiles. We show that the major results obtained this way can be reformulated using a more natural notion of "almost always", the notion of prevalence. Prevalence was introduced by Hunt, Sauer and Yorke in 1992 and is easily interpreted in terms of random distortions to a system. These techniques matter only for large classes of preferences and utility function. In the case in which preferences are Euclidean and can be specified by a point in issue space, things are much easier. A new theorem on the generic emptiness of the core of majority and some supermajority rules is obtained in that case. The theorem, albeit weak, has an easy proof, drawing on intuitions from elementary highschool geometry, and is instructive in how one can prove related results in the case where profiles lie in an infinite-dimensional space.