Re: Historical Ciphers

From: John A. Malley (
Date: 06/25/03

  • Next message: Tom St Denis: "Heads up on the LibTomMath Book"
    Date: Tue, 24 Jun 2003 20:35:56 -0700

    Lurker wrote:

    > It gets pretty muddy trying to examine 19th century perspectives of
    > 16th century codes. How much of the 19th century analysis was
    > subjective creation on the part of the analyzers?

    Astoundingly, very little, according to the anecdotes in Kahn's "The

    The 19th Century holds the roots of many modern cryptology concepts.

    We see the first steps toward "codification" of cryptanalytic "truths"
    as cryptanalysis books appear in print. European military cryptanalysts
    as well civilian cryptanalysts with strong connections to the military,
    published basic cryptologic facts in the 19th Century - Kirchoff's
    Principles, the Method of Kasiski. By the 19th Century, military
    cryptanalysts knew substitution ciphers (and codes) preserved the
    frequencies and dependencies of the "plaintext" in the "ciphertext" or
    encoded messages. They knew of the probable word attack. The knew of
    stereotyped beginnings and endings as ways into systems. They understood
    the importance of messages in depth for substitution cipher systems, and
    the importance of multiple same-length messages from transposition
    cipher systems. They knew to exploit characteristics of the code books.
    Code numbers assigned to plaintext tended to reflect the alphabetical
    order of the plaintext - so an unknown code group between two known code
    groups corresponded to plaintext alphabetically between that of those
    two code groups. Here we see the root of "cryptosystem characteristic
    exploitation", the idea of taking advantage of characteristics of the
    mapping from plaintext to ciphertext to break into the system without
    the key, or to reconstruct the key. This idea grew in the cryptanalysis
    of rotor based cipher machines and flourishes today as in the
    mathematical assessment of group characteristics, differential and
    linear cryptanalysis, fast factoring algorithms, related-key attacks,
    and more.

    > Look at some of the analysis of the Voynich manuscript for example.
    > What gets pulled out of the water in this case depends to a great
    > extent on what the analyzer brings to the table.

    Yes. Kahn covers this phenomenon well in his book in a chapter on
    pathological cryptanalysis (IIRC). I tend to think of it as an
    affliction of the amateur cryptanalyst. Consider it a blessing, though,
    because had we no pathological cryptanalysis, William Friedman may had
    never taken to cryptanalysis! (See Kahn's book for the story of his work
    at Riverbank and the Quest to Decipher Shakespeare. )


    John A. Malley

  • Next message: Tom St Denis: "Heads up on the LibTomMath Book"