Suppose that would depend on circumstance but in that case, the you've got authoritative evidence - ie the fingerprints and powder burns. You have no such thing to stand on in saying "a planet exploded and created x rings and asteroid feilds etc." Your evidence is - well these things are there - therefore something caused them to be there, so we'll infer a planet exploded. So a comparable instance re your murder example would be, the person is dead, your brother knew them, Your brother had as much reason to kill the person as anyone, therefore your brother did it. You don't have finger prints or powder burns. You have no prime suspect or evidence pointing to a prime suspect. You have a result looking for a cause and are blindly stating one cause as being the instigator of the situation. That is not science. And there is no confidence in your result. I'd refer you to first year logic, among other things.
Science is not really about first year logic. The logic employed by science is speculative, and produces levels of confidence, not proof. The science employed by biologists studying evolution is identical to forensic science. It is forensic science.
So I am asking you for a simple answer to a simple question: which is more reliable, forensic evidence or human witnesses? I understand that it depends on the specifics, but which is generally more reliable, solid forensic evidence, or the testimony of someone you love?