Hi John,
At 13:07 12-06-2008, John Hardin wrote:
>That's not the "what" that you're measuring with the probability
>factor, that's the "how" you're measuring it.

Right. It's been a long day. I'll refrain from elaborating or else
I'll say another blunder. :-)

>Was the OP _not_ talking about "trust" in the only context that SA
>uses it, then? If so, that's what prompted my comment.

I was responding to the subject.

>I don't see how "does/does not forge headers" can be anything but
>binary, thus rendering

That's binary.

>a discussion of levels of trust meaningless in that context.
>"Probability of spamming" may be a useful metric to judge a host
>(witness the effectiveness of DNSBLs), but when discussing SA we
>should not use the term "trust" to refer to that concept.