I just finished William Poundstone’s book, which sports the clanky title Gaming the Vote: Why Elections Aren’t Fair (and What We Can Do About It). It’s about voting systems, a subject which, curiously, seems to engage only wonks. Despite the object lesson of 2000, most Americans can hardly conceive that anything besides plurality voting exists, much less that it might be better.
Till recently, game theorists were almost as pessimistic. Kenneth Arrow’s 1948 paper seemed to show that no voting system was free of paradoxes. Whatever you picked– plurality, single transferrable vote, Borda counts, Condorcet voting, instant-runoff voting, approval rating– you can construct an example where the wrong guy wins. Plurality voting systems are prone to spoilers or clones; more sophisticated systems fall prey to strategic voting (where voters conceal their true preferences to give an advantage to their favorite); ranking systems in general fail if voter preferences aren’t transitive.
Poundstone unveils his own best candidate last (spoiler warning): range voting. You very likely know it already: it’s the numerical ratings used at Hot or Not or rottentomatoes. It bypasses Arrow’s paradox because it’s not a ranking system; it easily handles spoilers and strategic voting, and it’s simple to boot.