>> Right, they are almost always worse than first-hand information,
You're probably right about this, in a sense, but it beckons a pretty interesting question: provide an example when "signatures" are better than first-hand information.
Counter example: If your job is to find enemy subs, then sonar feedback off the hull of a Akula Class nuclear attack sub may be better than direct visual observation of the sub
Nobody complains too loudly if you fire a torpedo at a blue whale, but there would be dire consequences if you hit the wrong sub. I wouldn't call sonar contacts statistical evidence though, even if it is partially correct.
The counter example doesn't fit. In you example the sub is hostile because it is Russian. Here the people are convicted to death by circumstantial evidence. Very, very bad circumstantial evidence. It is just "worth" it because there are little consequences for the murderers.
You're probably right about this, in a sense, but it beckons a pretty interesting question: provide an example when "signatures" are better than first-hand information.
Counter example: If your job is to find enemy subs, then sonar feedback off the hull of a Akula Class nuclear attack sub may be better than direct visual observation of the sub