Consider an “isolation paradox” game with many identical players. By definition, conforming to a rule which maximizes average utility is individually a strictly dominated strategy. Suppose, however, that some players think “quasi-magically” in accordance with evidential (but not causal) decision theory. That is, they act as if others’ disposition to conform, or not, is affected by their own behavior, even though they do not actually believe there is a causal link. Standard game theory excludes this. Yet such “rational folly” can sustain “rule utilitarian” cooperative behavior. Comparisons are made with Newcomb’s problem, and with related attempts to resolve prisoner’s dilemma.