Re: Weakness of Feistel ciphers

Kim G. S. Øyhus wrote:
what happened in my tests, done while developing my ciphers,
was that in those Feistel ciphers of mine, the codes remained clustered.
Similar codes remained similar, and they did not become less similar with
increasing number of rounds. They even increased clustering.

It seems most likely that there is something wrong in your choice
of round function or in your implementation of the cipher. There's
nothing wrong with the Feistel structure, if you choose a strong enough
round function and adequate key schedule (this can be proven; see Luby
and Rackoff's result).

Care to share the definition of your Feistel cipher design? Perhaps
we can point out the problem in your choice of round function, key
schedule, or implementation.