[Python-Dev] The first trustworthy <wink> GBayes results

Neil Schemenauer nas@python.ca
Tue, 3 Sep 2002 14:52:51 -0700


Tim Peters wrote:
> Since deciding on the largest acceptable false positive rate is far
> more a social than a technical issue, a group of nerds will do
> anything rather than face it <wink>.

I think we pretty much ran out of things to do. :-)  Still, I think the
acceptable rate depends heavily on what happens to the rejects.  If they
go to /dev/null then it would have to be very low.  If there are bounces
and a way for the innocent victims to bypass the filter then I consider
0.5% good enough for most situations.  The major remaining problem would
be handing legitimate automated email.  For mailing lists that probably
isn't an issue.

I'm probably not the guy to listen to about acceptable rates, though.  I
currently use TMDA and therefore am a heartless bastard. :-)

  Neil