Belief Samples Are Sufficient for Social Learning with Probability One
Learning the true state of the world occurs with probability one in a social network setting where agents only communicate samples from their beliefs, rather than full belief distributions.
