Re: manual cryptography

From: Jim Gillogly (jim_at_acm.org)
Date: 10/31/03 

Date: Fri, 31 Oct 2003 14:15:49 GMT

John Savard wrote:
> On Thu, 30 Oct 2003 06:31:31 -0500, "John E. Hadstate"
> <jh113355@hotmail.com> wrote, in part:
>
>
>>Hmmm. You're right. Suppose one were to generate the key material by
>>autokeying the secret key and the alphabet (or any other pre-agreed text).
>>Then just encrypt the plaintext by adding the resulting key material mod 26.
>>It would take twice as much time, but would pretty well close the "probable
>>word" loophole.
>
> I'd be more inclined to follow the kind of thing Douglas Gwyn pointed
> out - Playfair followed by a transposition.

Where did he recommend this? Doing the transposition would work better,
because you can use the Playfair identification characteristics (no XX
digraph, high digraphic index of coincidence) as criteria to unwind the
transposition, allowing you to recognize the Playfair without having to
solve it yet. Are you sure he didn't recommend doing the transposition
first, followed by Playfair? I didn't find his recommendation with Google.

> But let's _really_ get fancy: use a straddling checkerboard, then
> "Giant Playfair" using a 10 x 10 square of two-digit pairs, then
> double transposition.

Again, I think after doing the straddling checkerboard (aka monome-dinome),
it would be better to do transposition first, and finally Playfair. Your
idea here seems sound, though: I'd hate to have to solve this with the
fate of mankind resting on a quick plaintext recovery.

>>Now, do you know of a way that is practical to compute manually to
>>authenticate the ciphtertext?
>
> With manual encryption, if it decrypts and makes sense, it's
> authentic... as long as you _don't_ use something like the one-time
> pad. Authentication is too prone to error to do manually, although I
> suppose autokey might work as an intermediate layer.

I think autokey is a dangerous thing to use in manual systems, because
once you get screwed up you stay screwed up. If the encipherer makes
a mistake (e.g. trying to code it up fast while the gendarmes are on
your tail), it can be a real pain for the recipient to unravel it. I
agree with your previous thought: a coherent decipherment verifies itself.
That's also the watchword of the cryptanalyst: a coherent decryption
verifies itself. 

-- 
	Jim Gillogly

Relevant Pages

  • Re: Double transposition cryptanalysis
    ... Here are larger examples with key and plaintext. ... Examples 1 and 3 avoid the severe weakness of a ciphertext several times ... is not, IMO, in the spirit of double incomplete columnar transposition. ...
    (sci.crypt)
  • Re: ciphertext
    ... > I wanna ask if yo know the ciphertext is encrypted using columnar ... > transposition of a plaintext ... > Assume that the width of the transposition N is an integer in the ... If columnar transposition is used I would not guess at words, ...
    (comp.security.misc)
  • Re: manual cryptography
    ... > Doug said Playfair followed by double transposition would be hard to crack ... Playfair encryption more effectively than a legitimate double transposition ...
    (sci.crypt)
  • Re: manual cryptography
    ... >> Doug said Playfair followed by double transposition would be hard to crack ... The difference is that you can recognize a legitimate ... > Playfair encryption more effectively than a legitimate double transposition ...
    (sci.crypt)
  • Re: manual cryptography
    ... I wouldn't feel at all confident trying to ... >>do the same thing if the double transposition were first and the ... >>Playfair encryption more effectively than a legitimate double ... Double Playfair isn't what I had in mind when I said "standard ...
    (sci.crypt)
Quantcast