The post is about a confusion between the question that p-values address ("how likely is this data to have come from the null hypothesis?") and another very different question ("given several findings that were unlikely to have come from various different hypotheses, how many are likely to be spurious?").