Keith Moore wrote:
spammers interpret filters as damage and route around them.
Or, learn how to pierce through them...
Others have used the sport analogy, still others refer to it as
war. In any case, spam filters that rely on analysis of messages
contents are simply not the solution -- no matter how strongly marketed
they may be and no matter how interesting the construction of filtering
systems may be to college students.
The best we can expect from spam filters that rely on textual
analysis is an ever escalating "arms race" where email recipients deploy
ever more complex and sophisticated spam filters to defend against
spammers whose methods grow continuously more complex and sophisticated.
Even the "learning" filter systems, based on Bayesian logic, etc., which
are so greatly touted these days, cannot hope to defend effectively
against spammers who use "learning" techniques to discover what the
filters have learned. When the offense and defense both have the same
tools available to them, an arms race is the most expected result. And,
an arms race does little but cripple all involved by forcing them to
spend more and more of their resources to maintain a stable
relationship.
I was very pleased to see that the charter of this group focuses
on the problem of "consent-based" communication and appears to direct
attention away from filtering solutions based on textual analysis. I
must admit though that I'm somewhat worried about the discussion of
"approaches that determine the nature of a message based on its
characteristics" and sincerely hope that that this phrase does not
become a trap-door which permits endless discussions of methods by which
the useless textual analysis arms race can be furthered.
Filters based on textual analysis cannot succeed. Let's focus on
fundamental solutions to the problem of spam -- i.e. how to build
systems for "consent-based" messaging.
bob wyman
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg