You wrote:
Some words about DES - I spoke with one cryptoanalisyst from KGB and he sow, that for number crypto algotitm c(key, text) (key is keyLength tall) present f(key, text), that for all key1 and key2 present key with length keyLength, that c(key2, c(key1, text))==f(key, text).
He also say, that now present f() for c()=des(), more f() wery like des().
That`s why for decrypting of des(k1, des(k2, ... des(kN, text) ... )) we must try 2^56 keys with spetial function.
I suspect that you will get several comments about the proof that DES isn't a group. That doesn't apply here since you have not claimed that f() is DES -- only that it's DES-like. I would expect the strength of triple-DES to show up in Diff.Cryp. attacks, but maybe not. Meanwhile, I stick to my sci.crypt suggestion: compress|des|tran|des|tran|des If you could get more details from your KGB friend about that attack, there are one or two people here who would be very interested. :-) - Carl