[Cryptography] Hypothetical WWII cipher machine.

Ray Dillinger bear at sonic.net
Sat Jul 18 17:31:11 EDT 2015



On 07/18/2015 02:39 AM, Dave Horsfall wrote:
> On Fri, 17 Jul 2015, Ray Dillinger wrote:
> 
>> And it does have a cool diagram and one-page basic explanation.
> 
> I'm having trouble in visualising this (especially when I can't get the 
> page to print at anything other than 46% scale).
> 
> Would I be right in thinking that those segments rotate around themselves 
> in the manner of a Rubik's Cube?

Yah, on consideration it's sort of like a weird, flattened,
three-cornered, hexagonal-faced version of a Rubik's cube
with no immobile central faces.

The three circular regions rotate in chaotic sequence, and
because any two of the circular regions have six parts in
common and all three of them have four parts in common,
they swap parts depending on which one's rotating, very
much in the style of the way edge and corner cubes get
swapped around on a Rubik's cube. Within a few turns, any
part can move from anywhere to anywhere else or be
reoriented in any direction or both.

Sequences of rotations continually swap and reorient parts,
and have chaotic feedback because which rotations happen
depend on which 'boats' get lined up with the driving wheel
and what direction they're pointed in when they do.

The inputs are a positive and negative electrical impulse
routed independently through the maze.  It's very difficult
to even predict how long the paths they'll take are other
than in statistical terms, and unlikely that the path lengths
will be equal, so a cyclometric attack or anything like it
doesn't have any obvious way to work.  And at least as far
as I can tell there is no obvious similarity to the
alphabetic mapping produced by adjacent states to hang
a related-key attack on the way there is with a single-
wheel step of a rotor machine.

			Bear

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20150718/dc8291d7/attachment.sig>


More information about the cryptography mailing list