[Spambayes] Re: Slice o' life

[Tim Peters]

[Tim]
> I scored my own "spam" folder, and discovered 5 spam with scores of 0.0.
They
> all have one thing in common:  they're spam that SpamAssassin didn't
catch, and
> came to me via a python.org mailing list.

[Harald Koch]
> This is why I don't usually bother spam-filtering my email lists. I
> don't get much spam that way to begin with; most of my email lists have
> their own spam filters in place already. In the olden days, filtering
> lists resulted in too many fps; now it confuses the classifier.

Unclear.  I retrained my home-mail classifier to go back to the "ignore most
header lines" defaults, and these low-scoring spam scored high again.
Regular list traffic continued to score low, presumably because it had
genuine hammish content.  What suffered some was personal email, which is
sometimes very brief, and where sucking up header clues about who sent it is
a real help.  Some solicited commercial email also suffered.  The system was
still highly accurate, although this is not a controlled experiment, and the
database has only one week of non-random email (so I won't draw any
conclusions based on this).