It's clear to me that ad hominem considerations are sometimes appropriate when evaluating the evidence for a particular assertion.
For example, say a clever grad student posts to her blog a weak statistical argument for some yet to be proven mathematical hypothesis. We know that she is clever, so we read and understand the argument, concluding that it is indeed evidence in support of the hypothesis.
Now suppose Terry Tao, arguably the greatest living mathematical genius, posts the exact same argument to
his blog. We might imagine that Tao has, in his head, multiple converging lines of evidence in support of the hypothesis, of which this is just one. We might also suspect that it is part of some deeper framework, yet to be elucidated.
We should consider the second situation far stronger evidence in support of the hypothesis, solely based on who originated the argument.