I'm experimenting with subject only rules rather than applying rules to
the subject and the entire body. I have found that rules applied to the
subject seem to by more accurate and return far less false positives
while still getting a high number of matches. Because of the higher
accuracy I can use broader rules than I could get away with in the
message body.
For example - the phrase "overnight shipping" if applied to only the
subject is very accurate but when applied to the body returns a lot of
false positives. Same with "save money".
If you broaden the rules you can increase the number of hits without
increasing the number of false positives significantly. You can also use
".*" without worry because the subject is short. And the tests are
faster when testing only the subject.
I know this is a hard concept to grasp - but think of the subject as hot
water and the body as warm water. Mixing the subject and the body
dilutes the spammyness of the subject and leaves warm water. There are a
lot of tests in the subject that you would never get away with in the
body because the subject is so hot.
I'm going to post more on this but I want to stir interest and debate in
the subject.
