Absence of evidence is necessarily evidence of absence. More specifically, if you could get some evidence E that would support a hypothesis H, but you get not-E instead, and you’re a coherent Bayesian, then you must become at least somewhat less confident in H.

The true claim in the neighborhood of what you said is that absence of evidence isn’t necessarily strong evidence of absence–absence of evidence for a hypothesis needn’t make you very confident that the hypothesis is false.

It seems that for some readers I need to clarify…

So some readers think you mean what you actually write instead of the disclaimed version you obviously meant? Silly readers!

Oops, I mean Robin is not necessarily being blunt.

Absence of evidence is sometimes only weak evidence of absence, but it’s legitimate evidence nevertheless.

Good point. All evidence needs to be considered. As I have argued, Karl Popper’s writing led many to see evidence exclusively in terms of either falsifying a hypothesis or failing to do so. We see this in nonsensical language like, “There was no evidence of X (p=0.2).” Sure, it’s shorthand for “There was no statistically significant evidence,” but it’s quite misleading.

It’s important to note that confidence intervals (or credible intervals) allow for shades of gray, rather than just black or white.

1. If nobody’s looking, it suggests a consensus that there’s nothing to find, which Aumann compels us to consider.

2. Publication bias means we tend not to hear about people looking unless they get positive results.

…is a bit of a mouthful.”

It’s just not catchy enough, that’s all. How about: “Absence of evidence is evidence of absence. Absence of looking for evidence is not.”