Quote:
Originally Posted by BruceZ
Except the only relevant probabilities are > epsilon and < 1 - epsilon. Nobody actually knows what epsilon is, but presumably there exists some epsilon which would provide a correlation. I changed the wording of the verdicts btw.
Not sure I understand. (I assume you mean something small by epsilon)
in some cases its might be 'more likely than not' when there are huge ranges of scenarios which clearly fall on one side or the other without any great idea of how likely
or jurors might have to consider something like how likely it is that the fingerprint evidence is wrong compared to how likely it is the defendant is lying.