I am under Mojave and the version of SpamSieve installed is 2.9.41. I have been using SS for over a decade. The question I have is NOT about the current release, this has been a question for months
My Spam folder (SpamSieve) usually has about 5 items per day that Apple Mail did not put into the Mail Junk folder. At least 2 of the Junk folder messages are ones I want to see
I use the SS train as Good or NOT JUNK options. BUT SpamSieve continues to put these messages into its’ Spam folder
I am discouraged that I am unable to train my SS to consider these message sources as good
I don’t know if this might be the reason why you’re having this issue, but you might want to check that Enable Junk Mail Filtering in Apple Mail is disabled (as per instructions when installing and setting Spam Sieve).
I notice on my Apple Mail / SpamSieve setup that Apple Mail “Junk” is different than SpamSieve “Spam”. Apple Mail Junk is trained by dragging/moving the item to the inbox. SpamSieve Spam is trained via the menu. If you have both Apple Mail Junk and SpamSieive Spam going the Junk mailbox, it will be hard to tell which one is correct. Apple Mail Junk seems to keep doing its thing even after I say “No” in the checkbox.
It appears that Apple puts its “bad mail” into a folder called “Junk” and SS sends messages to a folder called “Spam”.
My issue is with messages that Apple didn’t flag but SS is putting into the Spam folder
Since SS owns the Spam folder I question why the “Train” isn’t working
Surprising for working with Apple I have had success /luck with informing Apple Mail that a message is good. It takes more than one “hint”
The hints are to click on the “not Junk” icon or drag the message to the Inbox
SS ignores “not Junk” clicks WRT future message assigbmebg
I am going to assume you checked the log (Filter menu in SpamSieve) to see if the training is recorded. I have found the SpamSieve’s documentation on their website very extensive and email sent to them promptly replied to.
You should only use the SpamSieve - Train as Good command. Using the Not Junk command trains Mail’s own filter. If SpamSieve classifies a good message as spam and you don’t train it as good, that’s (almost) like training it as spam. So using Not Junk instead can lead to more good messages classified as spam.
Apple Mail should obey the checkbox, but it’s possible that you have a server junk filter that’s moving the messages.
There’s some information here about how you can figure out if SpamSieve is what’s moving those messages and, if so, why. You could also send in a diagnostic report and I can help you directly.
Apple Mail’s junk filter should be off when using SpamSieve. If you have a server junk filter that’s moving messages to the Junk mailbox, it may help it if you move good messages back to the inbox. There’s no additional benefit to using the Not Junk button.
I must have missed the guidance to NOT use Apple Mail’s junk filter. I just went to Mail preferences and turned it off. Will see if that changes the situation
I do use some rule / filtering on the Mail server but it is only to see messages that I want to show up in my iDevices
I think “Mistake: False Positive” is what you want to see. It is telling itself that SpamSieve’s assignment of “Spam” was a false positive. It is learning.
The Trained: Good (Manual) means that the training took. The Mistake: False Positive means that SpamSieve recognizes that it had incorrectly predicted this message to be spam (as opposed to your training a message that had been moved to the Spam mailbox for some other reason). So this is good in the sense that SpamSieve knows it made a mistake and is learning.
You could also look backwards and find the Predicted: Spam log entry for that message to see why SpamSieve originally thought it was spam. Sometimes there is an important clue there about something that needs to be adjusted in the settings.
It’s also possible that, over time, using Not Junk instead of Train as Good has built up a lot of incorrect information in SpamSieve’s corpus. In other words, there were other messages that looked like this that SpamSieve thought were spam, and since you didn’t tell it otherwise, it learned that they were spam, making future mistakes like this more likely. In that case, the remedy is to choose Reset Corpus from the Filter menu, then re-train SpamSieve.