Re: Key length security (calculations!)

[wcs@anchor.ho.att.com (bill.stewart@pleasantonca.ncr.com +1-510-484-6204)]

> solman@mit.edu writes (quoting someone else initially):
> >What about Shamir's triple pass key exchange protocol (explained briefly
> >below). Its the perfect key exchange algorithm. 
> >It obsoletes Public key systems entirely as long as you only need to 
> >exchange keys and not authenticate.

[ A: Ea(K)   B: Eb(Ea(K))  A: Da(Eb(Ea(K))) == Da(Ea(Eb(K))) == Eb(K) ]

Some problems - commutative symmetric encryption algorithms aren't very 
common; the most obvious one ( M xor K ) doesn't work here,
since you have known plaintext, making it trivial to find Ka and Kb,
while the usable M**K mod p looks suspiciously close to Diffie-Hellman
from a patent perspective, though Diffie-Hellman makes useful simplifications,
as Hal points out.

Also, an active eavesdropper can break it (maybe at the cost of reencrypting the
entire conversation), which is the main weakness of Diffie-Hellman;
authentication is still necessary in environments where this matters.

Hal writes:
> Also, this does not obsolete PK since it requires several exchanges 
> before communication can occur, making it inappropriate for high-latency
> communications, such as for most email.

Lots of email these days flows over SMTP connections, where
a couple of extra messages at handshake time isn't a major problem,
though it's really more applicable for link encryption than end-to-end,
given the amount of mail that really goes through MX records or
other mail gateways.  It wouldn't be that hard, now that RSAREF includes
Diffie-Hellman, to do a DH-SMTP and DH-POP, though the <censored> export
regulations make it a bit annoying to use internationally,
and you could just as well use Shamir's 3-way handshake if there's no 
patent problem.

				Bill Stewart