Not without a causal link, the absence of which is conspicuous.
Not necessarily. Causation might not be present, true, but causation is not necessary for correlation, and statistical correlation is what Bayes is all about. Correlation often implies causation, and even when it doesn't, it should still be respected as a real statistical phenomenon. All Jiro's update would require is that P(success|genius) > P(success|~genius), which I don't think is too hard to grant. It might not update enough to make the hypothesis the dominant hypothesis, true, but the update definitely occurs.
Another month, another rationality quotes thread. The rules are: