Science: the reality versus the ideal

There’s an interesting article at litigation and trial about the use of science in the courtroom.
Too often courts and judges treat scientists, unless they have been completely dismissed as cranks, as oracles of unbound wisdom. this is far from the case, and there is a large gray area. Discussion in the comments, apart from wishing aloud that more people read Feynman, raises issues such as voice recognition software which is probabilistic – yet is sometimes treated as essentially error free.