Three's company when seeking unanimity
Abstract
Random searches are routinely used in many algorithms, for instance, when looking for the shortest path connecting 100 cities or for ordering genomes in a most-parsimonious descendency tree. In this issue of PNAS, Matsen and Nowak use a random search for another type of problem, that of finding coherence rather than optimality. The task thus consists of reaching unanimity. The proposed solution is outrageously simple: Keep switching until you agree with two others, then stop. Under a wide set of conditions, this slapdash recipe works.