On Thu, Apr 10, 2003 at 09:01:13PM +0900, Jason Voegele wrote:
> > As an aside not only would the origonal spam be removed by your rules, 
> > but your comment on the spam and Brians' comment on your comment. For 
> > for one spam we would have two false positives.
> > 
> > This is why killing spam is so hard.
> 
> If you use the latest version of Mozilla the mail client includes a 
> Bayesian spam filter.  I've only been using it for about a week, but 
> already it catches over 95% of the spam I receive, and there was only 
> one false positive on the day I first started using it.  For instance, 
> the filter marked the first message in this thread as junk, but was 
> smart enough to realize that the replies were not junk.

And a Bayesian spam filter requires a body of marked 'non-spam' as well as
'spam', so it only works at the client side; another good reason to do the
filtering yourself, not on the list.

(That is, unless someone is going to volunteer to mark each ruby-talk
message as 'spam' or 'non-spam' - a sort of post-mortem moderator. But their
efforts would give such a minimal benefit to the community that I don't
think it's worth it. The amount of spam I receive from sources other than
ruby-talk is at least 100 times higher)

I would not like to see any static rules-based filters on ruby-talk: they
have too many false positives, and every time a false negative occurs it
will start a thread saying "why don't we update the filter set to match XXXX
..." which IMO is worse than the spam itself :-)

Anyone who hates spam can apply their own filters, and they will have the
benefit of trapping spam from other sources as well.

Actually, since I thought ruby-talk is gatewayed to/from Usenet, it seems
strange that we get so little spam here?

Regards,

Brian.