ietf-822
[Top] [All Lists]

Re: MTS transparency and anonymity

2005-02-27 17:10:13

On Feb 27 2005, Arnt Gulbrandsen wrote:

I wonder how many spam filters reject blah(_at_)[](_dot_) Does anyone on the 
list know?

Arnt

This is not an easy question to answer, because modern spam filtering
technology is very dynamic. There was a presentation by Ken Schneider
of Brightmail at the 2003 MIT spam conference (unfortunately his
presentation seems to be a dead link now, here's the abstract

http://spamconference.org/abstracts2003.txt

IIRC, his overview of Brightmail described a complex feedback system
built on rule specifications which are created around the clock 
automatically and have to compete with thousands of other rules. The
best rules are sent to their filtering systems around the world
every few minutes.

So you could say a rule such as From:(_dot_)*Arnt(_at_)[] might not trigger
anything while From:(_dot_)*Fred(_at_)[] might do so at one time or another. I
doubt that spam filter rules look very closely at RFC 2822 semantics
in general. Some rules may well pick up illegal RFC constructs, but
other rules may just pick up particular substrings of the message
viewed as an ASCII string, and all rules have variable lifetimes.

In summary: I'm not sure that your question makes a lot of sense for modern
spam filters.

-- 
Laird Breyer.