Stopping spam by attracting it?

[Chris Shenton <cshenton@Outbounder.com>]

Perhaps this is painfully obvious and folks are doing it already; if
so, does it work? If not, here's what I'm thinking. 

Insert a bogus, spam-attracting email address into all the email I
send, like victim@mydomain.com; this address should never send mail,
so any mail it receives will be from address-harvesting spammers.
Alias that address to a small program which examines the headers and
extracts source IPs, From: lines, Subject: lines, etc and installs
these into some filter config. When your MTA receives mail, it
consults the filter looking for match and rejects any messages
matching the filter; alternatively, have the MUA do this and put it in
a folder/group for suspicious mail.

The trick would be building the program to find "significant" headers,
avoiding generation of overly loose matches which would filter too
much legitimate mail.

Anyone doing anything like this? Any thoughts how I might do something
like this with Gnus? 

Not quite Gnus-related, but if one collected lots of spammer/relay
host addresses, they could automatically be added to an RBL-like DNS
server, operating for an individual, an enterprise, or the public at
large. Would also be nifty if such fingerprints could be shared to
automatically build a list of suspicious senders.