There are several obvious Spams that persistently are ignored by SpamSieve (e.g. Toenail Fungus…) - I checked the log and found out, that these were ignored because they somehow appear on the whitelist: Even after training these multiple times as Spam, the Toenail Fungus mails still somehow seep through. The only way to actually get rid of them was to manually delete the entries from the whitelist. I am pretty sure, I never trained these kind of mails as Good, as they are so obviously Spam. I might have deleted some of them manually on my iphone - does this trick SpamSieve into thinking that these are Good?!
I also found countless list ids from unsubscribe-messages on the whitelist - pretty much every unsubscribe link from a mailing-list (these are unique through the attached unscubscribe token) is stored here. There are also a lot of unsubscribe-messages from spam-mails there. Why are these unsubscribe-links needed on the white-list, and does this cause problemes, as the white list database grows with every mail I get from a mailing list?!
This means that either there were older spam messages that got through to the inbox that you did not train as spam, or else you trained them as spam after the new messages had already been received and filtered. I recommend correcting all the mistakes and doing so as promptly as possible.
You should not delete spammy whitelist rules. It is better to just train the spam messages in your inbox as spam, and then SpamSieve will automatically disable the corresponding whitelist rules.
Yes. SpamSieve assumes that it’s correct about a message being good or spam unless you tell it otherwise. So if you get a spam message in the inbox and delete it without training it as spam, that is almost as if you had trained it as good. If it’s not possible for you to train all the mistakes, you can avoid this problem by turning off the Auto-train with incoming mail feature.
If many such messages have gone uncorrected, you will need to reset the corpus in order to remove the incorrect training data.
The intent here is that the unsubscribe link is a way to identify legitimate mailing list messages to help prevent them from being classified as spam even though they may be sent from people you have not previously received mail from. Some spam messages also include unsubscribe headers, in which case SpamSieve disables the whitelist rule when it figures out that the message is spam. Keeping those disabled rules doesn’t really cause problems other than slightly increasing memory and disk usage, but you can delete them if you want.