With N key bits, there are 2^N different subsets of key bits.  If you
fix a plaintext, then each ciphertext bit is an N-to-1 boolean
function.  Is there any way to show that there is no subset of key bits
whose parity is a good linear approximation of the function?
-- 
Mike Stay
Cryptographer / Programmer
AccessData Corp.
mailto:[EMAIL PROTECTED]

Reply via email to