tightened up my spam-processing loop. here's how it works: i use eudora, have three
ax@ accounts and 100 filters for sorting inbound traffic.
first line of defense is the "X-From_" not contained in "From" check. that works out to one filter per account. this one catches a lot of ambitious to/cc and bcc spam.
second line is to tag all unexpected bcc's as spam. this introduces a margin of error in the detection, but hang on, we just moved a maybe else where. this takes one filter per account to route explicit to/cc's elsewhere and one final filter to route everything left to spam.
whenever something now gets tagged as spam i have eudora notify a short application which takes a temp file containing the spam-candidate and queues it forward to my spamcop account.
spamcop responds to these forwards with a message that i filter into a dedicated box. a manual only filter is the triggered on whatever is in the spamcop bin. it opens one browser-window per pending spam-candidate.
from the spam-cop windows contining the candidates i send reports with a single button-press -- or i catch an overlooked list or ligitimate bcc.
at the same time that the manual filter opened browser windows, it queues the spam for my isp's spam processing system. if i find a message that was not spam in the process of clearing spam-candidates, i can switch back to eudora and remove the isp notification from the queue by subject.
it's never been that easy to protest that spam. the code's in delphi. there's not a lot of it.