Recently I set up a new mailserver and the spamassassin's bayes filter running on that machine has not been feeded with enough spam yet to be working properly. So how would one go about getting that little bugger some usable information?

The way I chose was setting up a special sub domain having both a web and mailserver running. The webserver does nothing but present a robots.txt which basically disallows all access from good web crawlers and for all other requests it presents the requester with an email address. That address is randomly generated and information about the requester stored in a text-file. An email to that address will cause the email to be stored in a special mailbox. A small script in turn issue 'spamc -L spam' for all new messages in that mailbox and so feeds the bayes filter.

I am curious about how well that is going to work, but actually it should attract quite some spam