Matus UHLAR - fantomas schrieb:
> it just proves that the mail was sent through sane server, but there could
> be spambod behind it.
> -0.1 and -0.2 is very small numbers. Do you encounter any case where that
> would help?

There are many tiny score rules that tend to also hit on some ham -
thats not a problem because their score is low but it will prevent the
message from being autolearned in bayes.
So it would be nice to have just as many tiny score negative rules that
compensate for that.