[Mailman-Users] Spam to list-owner

[Brad Knowles]

on 12/20/08 1:54 PM, Lindsay Haisley said:

> Well I'm probably not doing as effective a job of pre-filtering as you
> are at UT.

And what we're doing for python.org and ntp.org can't match those 
numbers, either.

For ntp.org, we do pretty much the same as you -- use a few RBLs 
up-front, and then SpamAssassin on the back-end.  What I want to do for 
the next version of the mail system there is to pull the RBLs into 
policyd-weight under postfix (so that we can apply SpamAssassin-like 
scoring into the RBL process), and not do any direct RBLs.  I'd also 
like to make sure that SpamAssassin is set up to run interactively.

For python.org, we're already running policyd-weight plus SpamBayes.  In 
the future, I'd like to see us pull SpamBayes into interactive mode.

For both sites, I want to look at pulling in reputation scoring systems 
like SenderBase, and get those incorporated into the process before 
handing off the remainder to the content-based scanning tools.

> As always, your advice and concerns are well-considered, Brad.  I do
> need to accept a certain amount of this stuff, consistent with the
> requirement that 100% of legitimate email be delivered (and SA is far
> from perfect), and because I'm a SOHO business, and a small one at that,
> I can't afford dedicated analytical appliances and proprietary software
> for this, most of which is outside my budget.  There's doubtless more I
> can do.

You have a lot of options available to you, and you need to decide what 
you want to do and how you want to do it.  We can help you with that to 
a degree, but there are limits to what we can do.  And right now, 
Mailman doesn't have the best anti-spam features.  That is something we 
want to address, but we're not there yet.

-- 
Brad Knowles
<brad at shub-internet.org>        If you like Jazz/R&B guitar, check out
LinkedIn Profile:                 my friend bigsbytracks on YouTube at
<http://tinyurl.com/y8kpxu>    http://preview.tinyurl.com/bigsbytracks