Adam Funk wrote:
Why would you configure that way? It doesn't reduce incoming traffic,
which I thought was the goal of blacklisting.
I certainly wouldn't. The fact is, there's a lot of not-so-bright
admins out on the internet. Some of them might be running machines that
you need ot send mail to.
However, I can think of one legitimate reason to accept-and-drop known
spam. If the spam is sent to a forwarding account on a third machine,
which you don't control, you really have to accept-and-drop; if you
reject at SMTP time, the forwarding machine will generate a bounce,
which will lead to collateral spam. The general consensus is that this
collateral spam is not _your_ fault, but rather the fault of the admin
of the forwarding machine for accepting the spam in the first place, but
still, it's nice to do what you can to reduce collateral spam.
That seems like a strange way to train a Bayesian filter. Shouldn't a
Bayesian analyser be trained on manually tagged data (i.e. mails that a
human has classified as good and bad)?
Spamassassin is configured by default to autolearn as spam any message
with a score of 12 or higher. We could debate whether or not that's the
_right_ thing to do, but the fact is it's being done.
- Marc
--
## List details at http://www.exim.org/mailman/listinfo/exim-users
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://www.exim.org/eximwiki/
