C-Command Software Forum

What exactly does "Train as Good" do?

I’m using SpamSieve 2.6 with Apple Mail.

I’ll get mail from an address which I don’t want to add to the Address Book nor do I want to create a rule for the mail. The address is the same every time. I “Train as Good” and the mail is marked unread and moved to my Inbox. The next day, I’ll get a new e-mail from the same address and, typically, with the same subject but SpamSieve still considers it spam.

Is this normal? Do I HAVE to add them to the Address Book or create a rule in order for theis to work?


“Train as Good” adds the words in the message to SpamSieve’s corpus, so that they can be used by its statistical classifier. It also creates rules on the whitelist that match the sender name and sender address (and mailing list headers, if present).

That’s not typical. There are two possibilities that I can think of:

  1. You’ve trained a message as spam that came from that address, so although SpamSieve created a whitelist rule, that rule is disabled. One possibly remedy would be to find the rule in the whitelist and enable it (i.e. make sure it’s checked).
  2. SpamSieve is not classifying the message as spam, and it’s being moved to the Spam mailbox because of some other rule.

In cases like this, you should use the Open Log command to see what SpamSieve did. For example, if it thought the message was spam, then there would be a “Predicted: Spam” entry in the log, and the “Reason” would give some clue about why it thought the message was spam. If there is no “Predicted: Spam” entry for that message, then SpamSieve did not think that the message was spam.