I told it many times already that the freedom of speech includes
not only right to speak but also right not to listen.
Internet is very good at the first part; and woefully indaequate
at the second.
Can it be fixed? Of course. But the first step in filtering out
those who are trying to push their unwanted speech on us is to
make sure they won't pretend to be somebody whose words we'd
want to listen to.
So -- the problem has two aspects: the first is authentication,
the second is defense against flooding attacks. They are
closely related, but not identical.
The source address verification is powerful enough to get flood
attacks stopped. It is still not enough to get rid of unwanted
messages.
The second line of defense should be digital signatures on messages,
certified by some authorities (what is "authority" depends on your
personal point of view -- you're free to choose whom to believe)
which to a some extent make sure that signatures correspond to
physical people. Then you can just stop accepting any unsigned
mail (note that a reputable anonymous remailer would also check
signatures on incoming messages; and substitute them with its own).
There's no magic technology involved; this is just the problem of
how to actually implement it. Until we do that we all live in
danger of having our name smeared if some jerk decides he's pissed
and posts some nazi propaganda, or threats, with a reputable person's
e-mail address. I already was an antisemite, and an agent of KGB,
thank you very much.
Now, how about doing the right thing: make the NANOG list the
first one to require signed messages? Somebody has to start.
- --vadim
Can it be fixed? Of course. But the first step in filtering out
those who are trying to push their unwanted speech on us is to
make sure they won't pretend to be somebody whose words we'd
want to listen to.
[ ...]
Now, how about doing the right thing: make the NANOG list the
first one to require signed messages? Somebody has to start.
So in order to post to nanog you would have to have your PGP
key signed by NANOG or the list operator or another entity trusted
by all. How do you establish trust for that signing?
The Usenix key signing is a good model for techie types, but
we are the minority. How would the great unwashed have their
keys signed? Besides, I like being able to post without having to
attend a NANOG meeting though I could live with the restriction.
I agree that both goals, authentication and flooding defense, are
desirable. Source address verification is important and doable as
long as everyone in a position to do so wants to. We have our filters
in place, does everyone? There are 3000+ of us little guys and
we are the ones who need to do it. The NSPs and regionals could
possibly filter at customer gateways but with multi-homing how much
human/CPU load would that present?
Authentication is worse when looking at ubiquitous verification.
Unless we give up on PGP and switch to PEM and RSA certs (using RSA
as a trusted authority). And even then, for personal certs anyway,
they don't seem very secure to me. I can get one for free and all
they require is a valid e-mail address to send it to. Once I have it
I can forge my e-mail address and use the cert to sign messages
originating from another account. Of course the original "valid"
e-mail address has long since disappeared. I haven't actually tried
this, but I don't see how a signed message injected into an SMTP port
could be distinguished from a "real" one. Sure, the IP address will
be in the headers, but they aren't signed.
Now the threat of RSA pursuing me for violation of their personal
use restrictions *might* slow down some spammers, but probably not
until a few have been caght and hung.
This might be more on topic on cryptography@c2.net.
Majordomo list. Low volume.
Dan
- --
Dan Busarow 714 443 4172
DPC Systems / Beach.Net dan@dpcsys.com
Dana Point, California 83 09 EF 59 E0 11 89 B4 8D 09 DB FD E1 DD 0C 82