[liberationtech] Peer-review required: SwaTwt and TweedleDH

[Uncle "The Dod" Zzzen]

On Tue, 2010-10-05 at 16:33 -0500, Brandon Wiley wrote:
> 
> I seem to have missed your original about SwaTwt and TweedleDH as my
> mail server was down for a few days. So this is the first I've heard
> of them. I just wanted to say, very cool!
By now - I got convinced by the reviewers here that it wasn inherently
insecure, but thanks. It was a cool ride while it lasted :)

> 
> Also I noticed that you're using Python, so I wanted to let you know
> about the Elliptic Curve Diffie-Helman (ECDH) library for Python,
> specifically Curve25519: http://github.com/warner/curve25519-donna
If I understand correctly, the main advantage of ECDH over "plain DH" is
speed. I don't think DH is be a computational bottle neck in TweedleDH
(or a possible "next generation" system).
On the other hand [at least at the moment] I'm trying to stick to
"inspectable code" (i.e. algorithms simple enough to enable visual
inspection of the code to rule out possible trapdoors).
I still need to see whether people here believe "inspectable crypto" is
still possible in 2010, but if it is - I guess curve25591-donna is too
complex to be included in such a system.
If people here convince me that the best course is to abandanon my
"inspectable fantasies" and use tried and tested 3rd party libraries
(e.g. m2crypto or pycryptopp), curve25519-donna may also be a candidate
(depending on advice from reviewers here - of course).

Cheers, and thanks for writing curve25519-donna,
The Dod