Pardon me - I misread your memo. They use adding rather than multiplying.
Adding is more appropriate for scores related to spam than to ham. (The
proper would be to somehow invert the probability of being spam score,
multiply them together, and then reinvert to get a spaminess score. The
additions approximate this fairly well in the long run.)
{^_^} Sorry about being somewhat misinformative here.
----- Original Message -----
From: "Chris Hastie" <[EMAIL PROTECTED]>
> On Thu, 16 Jun 2005, jdow <[EMAIL PROTECTED]> wrote
> >From: "Chris Hastie" <[EMAIL PROTECTED]>
> >> Thus if a piece of mail has failed all three of these tests, the
> >probability of
> >> it being ham is 0.05 * 0.2 * 0.4 = 0.004, or 1/250. Or put another way,
we
> >can
> >> be 99.6% sure it is spam.
> >>
> >
> >They got there before you did, Chris. That is how the scoring system
> >within SpamAssassin itself works. It is also how the scores on rules are
> >set - nominally.
>
> Cool. I must admit I've never poked about in the internals of SA - I'd
> just somehow got the impression that scores were summed. But then I
> suppose if the scores are somehow logarithmically related to the
> probability of a test picking up spam that would amount to the same
> thing.
>
> --
> Chris Hastie
