The test was uniformily unfair. The author trained the Bayesian spam detector programs with something like 70 messages of spam and ham each. Thus, even if SA had been run with Bayes enabled, it might've yielded atypical results that fell short of how SA might perform in a production environment. As others have mentioned, the network tests were disabled as well.
Arguably, SA is a hybrid since it uses builtin feature detection rules, Bayes, and third party network resources including Razor, and various DNSRBL's (DNS Realtime Black Lists). Therefore, if properly tuned, SA really should outperform any single specific approach. > > Slashdot www.slashdot.org has a story on rating spam filters. > The link is to freshmeat at www.freshmeat.net/articles/view/964/ . > Spamassassin is rated without the Bayesian filters turned on, > which is disappointing. > > ------------------------------------------------------- This SF.net email is sponsored by: VM Ware With VMware you can run multiple operating systems on a single machine. WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk