Justin Mason wrote:
SBL's effectiveness is 8.01% for us; for Chris it's 7.63%. Pretty
close. But our effectiveness for OPM is much lower, at 6.3% instead
of 51%.
Two reasons for the OPM issue. One is that OPM's coverage has had a
huge jump over the past month or two. Second is "I'd have to kill you"
(which is also related to reason #1) :-)
Too bad you didn't have MONKEYS in there too. That'd have answered the
question of whether the "I'd have to kill you" factor predominates or not.
Chris notes an accuracy rate of 20% for SPEWS. We get 53% (as
relays.osirusoft.com), but with 11% FPs.
My numbers for SPEWS and yours for relays.osirusoft.com aren't
comparable. relays.osirusoft.com is a combination of several things in
addition to SPEWS. OSIRUS relays (inputs.relays.osirusoft.com) and
Socks (socks.relays.osirusoft.com) are very large (OSsocks larger than
SPEWS), and together they skew the relays.osirusoft.com values far away
from SPEWS alone. However, I'd suggest that SPEWS was the main
contributor to the "nonspam" value (tho, "spamsites" may have
contributed too). It'd have been more useful if you differentiated the
contributing lists to OSIRUS.
Strangely enough, we did try DSBL, and weren't very happy with it. No,
we weren't using "unconfirmed".
Ditto NJABL. It may be better now.
Be interesting to try SpamAssassin on our spamtrap. Maybe one day.
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg