On Mon, Oct 16, 2006 at 09:36:52AM +0100, Nigel Metheringham wrote:
[...]
> I am not at present planning to move to the far more drastic method
> of just bouncing mail from non-subscribers - which would break our
> long standing policy of being open to non-member queries.
a couple of comments. Firstly, the amount of effort
required to subscribe to the list is pretty small, and
many potential users will be used to temporarily
subscribing to lists (with many open-source projects it's
necessary to subscribe, either because their mailing lists
do have a reject policy or because the moderators are so
overloaded that legitimate messages don't get through or
are significantly delayed). It's not like the traffic on
the list is so great that it'll overload people to
subscribe for a little while.
Secondly, has anyone experience of using bayesian filters
to distinguish spam from nonsubscriber mail to mailing
lists? On the face of it this should work pretty well for
a list like exim-users where the subject matter is
well-defined.
--
`Evans boldly put 50 atm. of Ethylene in a cell with 25 atm. of oxygen. The
apparatus subsequently blew up, but luckily not before he obtained the spectra
shown in fig. 8.' A J Merer and R S Mulliken, Chem. Rev. 69, 645 (1969)
