🚀 go-pugleaf

Once again I've seen a reference to the fact that protocols relying on
MACs (message authentication codes) are preferable to those using
private key encryption mechanisms like DES.  The claim is that
libraries containing APIs to MAC mechanisms, like MD5, are exportable
because they don't fall under the munition regulations.

Isn't it a simple matter to use an algorithm like MD5 to produce a
very strong cipher?  Although there are lots of possibilities,
something like the following should work well

        Y[0] = MD5(K^R)
        Y[n] = X[n] ^ MD5(K^Y[n-1])

where X[1..n] are the blocks of plaintext and Y[0..n] are the
resulting ciphertext blocks, ^ is exclusive or, and R is just some
random value.  MD5 can be replaced by other strong hashing functions
or one-way functions.  It has been conjectured that MD5 is strong,
i.e. that for a given value V, finding a K such that V = MD5(K) 
takes on the order of 2**128 operations.

If the conjecture about MD5 is true then the resulting encryption
should be stronger than DES; furthermore, MD5 is much faster than DES.

So why is MD5 exportable and why couldn't it be used as a replacement
for DES?  (Of course there is the obvious reason that there hasn't
been as much time to study it.)








--

Todd
-- 
Todd Smith                                     TIVOLI Systems, Inc.
todd@tivoli.com                                6034 West Courtyard Dr.
                                               Suite 210
(512) 794-9070 [794-0623 fax]                  Austin, TX   78730