Hi, >> I disagree with that. In my opinion, only two criteria are needed >> to define spam: >> >> 1) An objective criterion: Was the message unsolicited? > > Unfortunately, that can be difficult to determine. People frequently put themselves on mailing lists as a consequence of creating a free account on a website or some such and then forget about it. A large, well known, reputable company is unlikely to be sending spam. So if you're on their list, you probably added yourself somehow. Also, their unsubscribe links tend to work, so it is much less work to simply unsubscribe yourself than to figure out how to get the emails marked as spam (which might affect someone else who actually does want the emails).
And you'd never get the truth from either the sender or recipient, even if you asked them. In my experience, end-users subscribe intentionally to very little, and unsubscribe from even less. They may sign up for something from a company, say, Kraft, not realizing they'll receive mail from all of Kraft's companies and products, not necessarily just the one they authorized. >> 2) A subjective criterion: Is the message unwanted? >> >> How the message gets to you is beside the point. >> >>> Two of the three messages, although unwanted, weren't necessarily >>> unsolicited. Ideally I'd like to stop these messages before the users >>> sees them, but how are we to know whether an individual user wants a >>> legitimate email or not? >> >> You ask them. You use an anti-spam system that allows per-user decisions >> about spaminess. > > But you still have to consider point 1. If a user starts complaining that he's getting spam from Amazon, I'm not going to mess with SA, I'm going to tell him to click the unsubscribe link at the bottom of the email. (Assuming that it actually is from Amazon, of course) I don't really like the per-user control. The challenge is to build a system that requires as little maintenance as possible - that's what we're supposed to be doing, IMHO. I'd like to be able to see each message a user considers to be spam. I'm afraid they will just dump everything into that per-user folder, instead of addressing each message individually, and have the false impression that we're not doing their job, when really they don't realize it's something they actually requested. Hopefully that sentence is clear, heh. I've had clients set up a "Place Spam Here" common folder on their Exchange system, then accessed it via IMAP regularly to analyze it for spam patterns, but Exchange 2010 apparently no longer supports IMAP with public folders. Ideas as a replacement would be greatly appreciated. Thanks, Alex