Angelo Bertolli wrote: > When I did, I remember thinking that without bayesian filtering I had > to set the threshold pretty low to ensure that there were no false > positives. I think that should be "pretty high" ... I think I kept it around 8 (not 5)