You also are committing a fallacy of "if X is true, then I must hold an immoral position Y, so X cannot be true". But logically either Y does not follow from X, or Y is not really immoral. (X = "poor people make bad decisions", Y = "the poor deserve their poverty") This logical fallacy has got you rationalizing everything that poor people do.
A fourth possibility is that patio11 and idlewords have priors which are not absolutely continuous w.r.t. each other. I.e., idlewords' prior for patio11's hypothesis is 0, and no amount of evidence will change his mind even if he is a completely rational. (It is by this mechanism that a rational Bayesian can also be religious. )