"David Gerard" dgerard@gmail.com wrote:
On 23/01/2008, Bernhard Reiter reiter@fsfeurope.org wrote:
On Wednesday 23 January 2008 10:50, Alan Pope wrote:
..or anti-spam on the list server.
We also do this, but it does not catch everything.
It's a very blunt instrument in my experience with Wikimedia lists. Our mail server assigns a spam score to everything that comes through, but it's amazing how low some spam scores and how high some human emails score. I tend to set high scoring mail to "discard" and a somewhat lower threshold to "hold for moderation". On lists where it's particularly important every human mail be read (e.g. Oversight-l, the alert address for deleting sensitive personal information from the wiki), I don't discard anything automatically.
I suggest to simply use a good spam filter and enqueue marked e-mails in a separate queue which is maintained by a human, so "learning data" can be collected and the false-positive rate can be reduced. I think discarding e-mails is not useful.
But when using spam filters keep in mind that people who send spam also use them for testing purposes and optimise their e-mails to pass the filter. But generally speaking I think the spam rate on the FSF Europe mailing lists is quite low and if someone cleans the archive, I don't care about less than 1% spam.
-- Matthias-Christian