CDR: RE: New OLD cryptograph patent for NSA

Ray Dillinger bear at sonic.net
Wed Oct 11 15:31:56 PDT 2000



>On Wednesday, October 11, Bo Elkjaer Wrote:
>>Yesterday oct. 10 NSA was granted another patent for a >cryptographic device invented by William Friedman. The >application for the >patent was filed oct. 23 1936 -- 64 years >ago.
>


On 11 Oct 2000, raze wrote:

>My question is this; why would they patent something that is 64 year old technology?  This is like the Enigma machine no?!
>

Um, actually, no.  The attacks we know on Rotor machines assume 
that the rotors rotate at predictable, constant intervals.  This 
was true of the Engigma, but not true of some later Rotor machines.  
The papertape variation of this system, with every cipher wheel 
rotating by some varying amount between each letter, won't fall 
to any rotor attacks we know of until the papertapes have repeated 
at least twice each.  Even then, it takes some fancy mathematics 
to figure out *how* to apply the rotor cryptanalysis to the system.

After reading this newer patent, I think it's actually *LESS* 
secure than the system it purports to replace. I could be wrong 
here -- I'd like to actually see the diagrams and drawings and 
my browser doesn't support 'em -- but it looks like the rotations 
are constant per keystroke with this system, which would make it
fall to rotor cryptanalysis.  The crucial question, the one I 
can't make out without looking at the diagrams, is whether the 
mapping of rotations to rotors is different each time.

Here's what I bet: department of the army didn't like the paper 
tape idea -- too fragile, too vulnerable to wet, required delicate 
machinery to read that had to be maintained -- and they wanted 
something a lot more rugged.  So he designed something that ditched 
the paper tape idea for them. It wasn't as secure, but it was still 
better than the Hebern-style machines that were likely under 
consideration as an alternative. 

				Bear








More information about the cypherpunks-legacy mailing list