On Tue, 23 Jul 2002, John Rudd wrote: > > So what brilliant metric do you have in mind to measure unsolicited vs. > > solicited commercial email? > > Personally, I'd rather not see them differentiated. (and, I'm going to > assume that "commercial" means "advertisement", and not things like tech > support information, and transaction updates, and things like that) > > For one, the difference is entirely subjective and personal taste. For > example, for me, there is no such thing as solicited commercial email.
There's no "personal taste" in the definition of (un)solicited: Either you asked for it, or you didn't. You may, as a matter of personal taste, never ask to receive an advertisement, but SA (as it comes out of the box) is supposed to be in the business of deciding how likely it is that one asked for something, not how likely one is to want what one asked for. Yes, someone who accepts your address under the pretense of sending you soemthing you wanted, and then uses it to send you advertising instead, can be said to be spamming you, and you're perfectly within your rights to configure your own copy of SA to dump their messages. (Though in those cases I would think there's usually a working unsubscribe mechanism that would be more effective.) That doesn't mean that the characteristics of those messages are good indicators of spam in general. Thinking about what I just wrote, I'm again going to plug the idea of categorizing rules as to what they're recognizing. The more_spam_to and all_spam_to configurations should not be based on the score value, but on what caused the scoring -- John should be able to ask that advertising fingerprints be considered, while others might prefer to filter only on forged headers or pornography indicators, etc. ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
