At 12:40 PM 2003-07-23 -0700, you wrote: >This may not be the best test because Bayesian filtering is based on the >spam/ham that YOU get and not what other people get. So although you get >a certain kind of spam that doesn't guarantee another person will. And >when you go to test out your newsletters it might be considered spam >when run through your system but not someone elses, leaving you with >misleading results. And the reverse is possible as well. Your spam >filter won't recognize it as spam but your recipient will because of the >difference in spam/ham between the two of you. Sure. I'm writing a web-mail style app right now that got caught in Yahoo's filters. I'm guessing that they are using Bayesian filters, and the causes of the catching were probably that I included HTML and inline images. That alone weighs the factor against the most legitimate email. The point, however is that it's a useful tool for figuring out things that one might not have considered. And as the author or an application designed for use by technologically semi-literate people, it's in my best interests to factor these things in up front. (Rather than being blamed that my app doesn't work because joe at yahoo.com didn't get my email.) -- Frank Marion lists at frankmarion.com Keep the signal high. If with Yoda so strong is the force , why correct grammar can he not use?