In message <v04210102b54c9601e1fc@[193.173.137.42]>, Jacob Palme writes:
>Should IETF do anything to fight the increasing incidences
>of net criminality? Can we do anything? Can the protocols,
>which IETF manages, be modified so as to make it easier
>to fight virus distribution, mail bombing, ping attacks
>and the other ways in which people are harassing the
>Internet?
>
>Or would such changes to the Internet protocols mean
>more invasion of privacy, in a way which is even less
>acceptable than letting the criminals continue? It seems
>that there is a strong group in IETF which likes the
>freedom of the net and believes that changes to stop
>criminal usage would also remove this freedom? Is this
>true, or can we fight net criminaltiy without risking
>the freedom of the net?
Harald made some very good points; let me just add a few more.
First -- as a computer security professional, I'm certainly concrned
about "cybercrime". But the changes you suggest cut both ways. I
don't know how to weaken a security protocol selectively, so that only
"good guys" can read the traffic. Apart from varying definitions of
what is good and bad -- and even the major democracies on this planet
have very different standards on that -- a hole is a weakness, and
requires much more complexity to protect that hole. Complexity is, in
itself, a security problem; in my opinion, and in the opinion of most
(though, of course, not all) of the security folks in the IETF, adding
any sort of back door to our cryptographic protocols would create far
more security problems than it would solve. (For more on this line of
reasoning, see http://www.cdt.org/crypto/risks98.)
There is also a major limit to the utility of strong authentication for
tracing back attacks: just as the bad guys use other folks' computers
to launch their attacks, they will also use others' cryptographic
credentials. I'm far from convinced, for example, that the LOVEBUG
virus would have been prevented were all mail digitally signed, because
I strongly suspect that the attack would have invoked a digital
signature API to generate digitally-signed copies of itself. The real
effect of such a policy would be a vast decrease in privacy on the net.
(I note that the EU has very strong privacy protection laws, and U.S.
law strongly protects anonymity as necessary for free political speech.)
In short -- yes, there are problems, but the best approach for the IETF
is to design, build, and deploy stronger systems.
--Steve Bellovin
