[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: A spam fake From: example
I agree with you about the best being the enemy of the good; I think we
also agree that any heuristic method for protecting IETF WG mailing lists
should be backed up by a human review.
We do indeed agree.
(I scanned my last 2 weeks of SpamAssassin-filtered mail today; 3
misclassifications, one of which was deliberately taunting SpamAssassin,
the two others were from the SpamAssassin mailing list.... I have tweaked
some of the metrics, though..)
I keep tabs on this as well. Here are my tallies so far this year:
1/1/2003 20F-
1/2/2003 9F-
1/3/2003 19F-
1/4/2003 14F-
1/5/2003 1572+ 22F-
1/6/2003 214+ 15F-
1/7/2003 17F-
1/8/2003 5F-
1/9/2003 304+ 27F-
1/10/2003 14F-
1/11/2003 501+ 27F- 2F+ (tls)
1/12/2003 536+ 4F- 3F+ (vpim, ttj, tls)
The notation should be obvious, but in case it isn't, + is correctly detected
spam, F- is a false negative -- spam that should have been caught but wasn't,
F+ is a false positive -- legit email caught as spam. I don't track legit email
not caught as spam.
I obviously don't check the spam folder every day and of course it always has a
backlog...
Ned