[exim] Looking for a volunteer to test a new spam blocking t…

Looking for a volunteer or maybe a few to help test a new spam filtering
technique that I've been testing with good success. I want to see how
this scales up.

Requirements, someone who is fairly savvy with exim, and running Spam
Assassin.

What I'm doing is providing DNS lists where an host IP can be classified
as black, white, or yellow.

Blacklisted of course means you can just block it. No testing needed.
White lists means that you can just pass it without any spam checking.
Yellow means the host sends some good email and should never be blacklisted.

The purpose of these lists isn't as much to catch new spam, but to
prevent false positives, not only in my black lists, but others as well.
This list is more for actively identifing nonspam.

Here's how it works. You first read the white lists. If it's white you
pass it on. If it's yellow then you skip all other blacklist tests. If
it's black you bounce it. Anything else you process with Spamassassin.

After processing it with SA if the result is nonspam/autolearn then you
run a script that will tell my database that the IP was a ham. If the
result is 15+ points you run a script telling my system that the message
was a spam. All you will pass to me is the IP address.

What I'm doing is counting HAM/SPAMS in a mysql database. Every 5
minutes I recalculate the lists. If the host IP has 30+ hams and is 99%
ham it is whitelisted. If the host has 2 or more hams it is
yellowlisted. If the host is 99% spam it is blacklisted. On the front
end of this database if a MyDNS server that provides DNS off of the
MySQL data.

I'd like someone who has say 100,000 hams a day and 2,000,000 spams more
or less. That should give good load testing. And - your data will
combine with mine to produce more accurate results.

Looking for someone who likes innovative solutions and understands and
likes this idea.

If interested - email me privately.