That's a little odd (if I'm allowed to say so). You're essentially saying that the amount of evidence required to support a claim with low probability should actually depend upon the impact of a claim on a society/individual? So, essentially, the amount of evidence required is completely...