Spammers catching on

Tue Dec 17 15:38:39 CET 2002

> Suzanne Skinner <tril at igs.net> writes:
>
>> Here's another case of sneaky, bayesian-filter-evading spammers. This
>> one, alas, was successful: it scored around 0.5 with Robinson-Fisher.
>> Fortunately, my SpamAssassin backup (threshold 10.0) caught it with a
>> whopping 29 point score.
>>
>> *********
>>
>> <HTML><FONT  COLOR=3D"#ffffff" BACK=3D"#ffffff">the story is sad and
>> downtr=
>
> Looks like we need to kill these sections when tokenizing -- or reject
> mail with text/html parts right away.
>
> Thanks for warning us, I'll put SpamAssassin back in place.
>

This presents a problem wherein statistics can be made to lie?

I think bogofilter has a role, but it's not absolute.
As a strategy it has a part, but it may not be 100% of the solution.