I get an enormous amount of spam, and wading through it all to find false positives takes too much time. As a solution, I want to move “very spammy” email into the trash instead of the spam folder. However, I’m worried about false positives. Looking at my log, most of my false positives had a score of 99 with Classifier: Blocklist. I’m thinking I should disable the blocklist and let the Bayesian filter do the best it can. I’d throw 95+ scores into the trash and look through my spam folder now and then for false positives. I’m thinking this would give about the same error rate but with a much smaller spam folder for me to filter manually. Am I crazy?
I don’t think that’s typical. Unless you were creating your own overly general blocklist rules, the blocklist would mostly be e-mail addresses and sender names that you had previously trained as spam. Would you be willing to share some examples of the problematic blacklist rules so that I can see what we’re dealing with here?