I once split a lunch bill with a Ph.D. statistician. By the time he finished his calculations, 50-50 had added up to three different sets of tips and two plus two no longer consistently equaled four.