🚀 go-pugleaf

In <5906@transfer.stratus.com> cme@ellisun.sw.stratus.com (Carl Ellison) writes:

>In article <1992Aug26.021824.930@decuac.dec.com> mjr@hussar.dco.dec.com (Marcus J. Ranum) writes:
>>cme@ellisun.sw.stratus.com (Carl Ellison) writes:
>>
>>>If I give you an RSA key, that's all you need.  I'm the only one in the world
>>>who knows the two primes in N.  That defines *me*.
>>
>>	With the usual caveat that he has to meet you or something so you
>>can give him the RSA key. Otherwise I can give him *my* RSA key and ask
>>you nicely for yours and just unwrap and rewrap everything that goes past.


>That doesn't work.  You could just as easily meet him personally, present
>an ID card saying you're me and hand him your RSA key.  To combat the
>impersonation attack, we need to make sure that we have the equivalent of a
>true broadcast which both the intended recipient and the sender can read.

The problem seems to me the authentification of the couple or relation:
(name (or person) and key (or signature)).

In some cases you are only interested in knowing that you have been
talking to the same person from the beginning of the conversation (like
in the problem of impersonating netnews posters), whereas in some cases
you want to make sure that you are indeed talking with John Doe, of
SmallTown (32 W, 16 S). In the first case you don't need the name-key
coupling, you are just "talking to a key". In the second, you need it
as you want to talk to a "name". (Thus the suggestion, lets give up our
names, and just use keys :-)

But rather than the One True Broadcast, or certifying agency, I was
thinking, couldn't you use the method suggested by the idea "you know
many friends, they know many friends, etc, so transitively you know
everybody in the world (6 times removed)"

This is a method studied for operating systems/networking security. See
for example:

%A Butler Lampson
%A Martin Abadi
%A Michael Burrows
%A Edward Wobber
%T authentication in distributed systems: theory and practice
%J Operating Systems Review (ACM SIGOPS Review)
%J Proceedings of the 13th ACM symposium on operating systems principles
%C Pacific Grove, CA
%D Oct. 13-16 1991
%V 25
%N 5
%P 165-182
%K transitive authentication, operating systems, DES, RSA, security,
channel, RPC, remote procedure calls, public key encryption, name
lookup, groups, access control, delegation, revocation, principals

This may be applicable for a "secure netnews" mechanism, as everybody
knows everybody on the net even more than eveybody knows everybody in
the, ahem, "physical" world.

Pierre.
-- 
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
       Pierre Uszynski, Philips Research Palo Alto, CA, USA
            USA-415-354-0328, pierre@PRPA.Philips.COM