> Knowing the SPAMassassin score might not help ... why do you expect to
find any
> correlation bwetween score and either false positives and false
negativews?
>
> I have some messages from the SPAMassassin lists that got analyzed
> (somehow, my whitelist settings got ignored sometimes) -- and of those
> (28 messages out of ~1600) some got scores as high as 13.9 ... OTOH I
> have seen unflagged SPAM with scores as low as 2~3.
>
> Try this ... sort by subject and eyeball through the list
> Sort by sender and eyeball through the list
Good suggestions. I used to do this before I started sticking the score on
the front of the spam subject line, allowing me to sort (roughly) by score.
It turns out that while you are quite right, there can be ringers at just
about any score, and FPs (for me) at around 13 are not all that unusual (I'm
rather liberal in my scoring), the FPs do for the most part hit in the lower
scores.
The rare "real" mail that makes into FP-land typically will score around 7.
Most of that will be do to me scoring anything from rr.com with 3 points to
start with, since a vast amount of my spam and very little ham comes from
RoadRuiner.
The scores around 13 tend to be ads and commercial mailing lists that I
didn't actually sign up for, but the submitters figured that "prior business
relationship" was sufficient reason to send them to me. (Arguably these
actually are spam; but I know the company is legit enough to honor an
unsubscribe should I send it - and when they originally stuck me on these
lists promenantly sent me a thing saying they were doing it, and if I didn't
like it, here was the "no" button. I actually read these fairly often,
which is the main reason I haven't unsubscribed. OTOH, I'm insufficiently
motivated to whitelist them.)
Loren
