[ih] Re: spam...

Joe Touch touch at ISI.EDU
Tue Jan 21 09:04:05 PST 2003


Perry E. Metzger wrote:
 > Joe Touch <touch at ISI.EDU> writes:
 >
 >>We will not block HTML. Even if the numbers above were correct, it
 >>would not solve the problem (foreign language isn't caught even in
 >>ASCII) and it would raise the number of false positives that require
 >>manual attention.
 >
 >
 > I respect your decision, but experience shows your claims are not 
correct.

We've discussed this before on E2E.

But let's check your claims, first...

 > You (essentially) contend that blocking HTML and other similar things
 > doesn't improve the situation and just makes for more work.  Although
 > nothing "solves" the problem, actual experience running large numbers
 > of lists shows that filters that generally catch spam and not real
 > traffic are a valuable technique and do not pose a burden.

I'm using SpamAssasin. This has nothing to do with HTML. SA kills that, 
but only in English as far as I know. It doesn't catch spam in every 
obscure language, especially when written in  English character sets.

Spam which got by so far is because it is in a foreign language, not 
because it's in HTML. E2E has examples of similar posts in ASCII which 
are not caught either.

If you have a better solution that is equally packaged, let us all know.

 > By the way, the NetBSD lists are unusual in that they allow postings
 > from non-members for technical reasons, so anti-spam filters are
 > important. Most lists I am involved with eliminate 100% of spam by
 > blocking postings from non-members. (And I know you claim your list
 > software would then make it difficult for people to post from multiple
 > accounts, but the documentation for your list software says how to do
 > it.)

Saying "how to do it" is consistent with it being equally "difficult". 
We chose simple, given that no solution is 100% effective anyway.

I will address further questions about the management of the list 
off-line only.

Joe




More information about the Internet-history mailing list