Venona not all decrypted?

[Travis H.]

When they re-used the codebooks, you can assume that they did so until
they got new codebooks, so I don't think such places should be so hard
to identify.

I think the problem is that once you've detected pad re-use, you have
two plaintexts added (XORed) together.  I don't know of an algorithm
that recovers the comingled plaintexts in a completely automated
fashion; it is my understanding that this still requires guesswork,
trial and error, and assumptions about plaintext that may transcend
what computers can do.  I read the book "The Venona Secrets" and I
seem to recall this as being an immensely tedious process.

Think about it; you have the sum of two letters stored in one, and
there are no crypto dependencies with other symbols to leverage. 
Basically you can think of one plaintext as a key and the other as a
conventional plaintext, and the key is just as long as the plaintext. 
In this case the key isn't chosen randomly, but is natural language,
and there's the weakness.

I wonder if this decryption involves increasing the Shannon entropy,
and by how much.  The comingled streams should have the same entropy
as an autocorrelated string in Russian, so you can subtract that from
2 times the entropy of typical Russian plaintext, and if that's
greater than zero, you're working magic.
--
http://www.lightconsulting.com/~travis/  -><-
"We already have enough fast, insecure systems." -- Schneier & Ferguson
GPG fingerprint: 50A1 15C5 A9DE 23B9 ED98 C93E 38E9 204A 94C2 641B

---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at metzdowd.com