Question on spam/ham
Say one of my users puts a "good" mail into her junk folder. It auto-forwards to the spam account for training. I notice the move, select the message in the user's junk folder and label it as "not junk", thereby forwarding the message to the ham training account.
When those identical messages process, do they cancel each other out since one says "spam" and one says "ham"?
Yes, as the bayes should re-train itself when it sees the HAM.
I figured they would, but you never know sometimes!