> I stopped using CBL on Jan-05-2004, though, because the > SpamHaus XBL is a superset of CBL, e.g.:
Looking at the individual results on my server CBL has had 32% of correct "votes" (Between 04/01/04 and 04/11/04) In the same timerange XBL has had 30% of correct votes. (really a superset?) The absolute number of false positives (for the individual tests) was with 204 for CBL compared to 43 for XBL more then 4 times higher. In relative numbers this are false positive rates of 0,25% for CBL and 0,05% for XBL. There are other ip4r tests like IPWHOIS, NOBASUE and FIVETEN-SRC having a FP-rate of 3, 4 and 5% (in this order) Another report shows that both tests have a congruity of 90% for positive results. XBL has positive results - without a result from CBL - in 19% of all cases. (values from 24 hours from begin of february) So in my opinion both tests are very good. However looking at the subject line of this thread I would really love to see a new test type that allows us to give extra points if certain combinations of tests has failed. Markus --- [This E-mail was scanned for viruses by Declude Virus (http://www.declude.com)] --- This E-mail came from the Declude.JunkMail mailing list. To unsubscribe, just send an E-mail to [EMAIL PROTECTED], and type "unsubscribe Declude.JunkMail". The archives can be found at http://www.mail-archive.com.
