Re: Are natural languages secure ciphers?

From: George Ou (533george_ou234_at_netzero234.com)
Date: 10/05/03 

Date: Sun, 05 Oct 2003 17:25:46 GMT

On Sun, 05 Oct 2003 13:57:42 +0200, Mxsmanic <mxsmanic@hotmail.com>
wrote:

>George Ou writes:
>
>> Just like natural languages provide a low level of protection.
>
>I don't think the level of protection is low at all. If you don't have
>the key (that is, the rules of the language and its vocabulary), you're
>not likely to crack it any time soon.
>
>> I hope you're not comparing the DLP, ECDLP, or IFP to the difficulty
>> of reverse engineering the ZIP algorithm, or the secret of a natural
>> language where 1.3 billion people know the same secret.
>
>Yes, I am.

You're comparing a shared static secret key to the break in the DLP,
ECDLP, and IFP. The only way I can put it is that is insanely stupid.

>> The question is "are natural languages good for cryptography".
>
>Yes. And thus far, it seems that they are, as long as you can keep them
>secret. That is, natural languages do not readily yield to brute-force
>cryptanalysis.

Not brute-force, but careful study.

>That kind of oversight is what allows adversaries to come up with nearly
>unbreakable encryption systems. You're thinking so much about Chinese
>that you've forgotten all the other languages that could be used.

It doesn't matter what language you use. Let's assume you can't find
a turn coat, and you have to learn a new language from scratch. The
biggest problem which John mentioned is that "attack at dawn" would
always be written the same way no matter what language you used. The
more contexts you inevitably hear through out the course of a war, the
more you can deduce a meaning.

Essentially, a natural language is like a super sized static key, kind
of like a book cipher but a little more complex. Unfortunately, it
can't with stand any kind of known text attack.

You keep saying, "but it's unbreakable without context". How in the
world to you propose to avoid revealing context? The more you use it
the more you reveal.

Take an AES 128-bit key. No one has ever managed to even come close
to brute forcing it, and context doesn't help you either if it's used
in as a session key.

A natural language is laughably weak.

George Ou
http://www.LANArchitect.net

Relevant Pages

  • Quantum Computational Linguistic Model (and then some)
    ... I have developed a quantum computational linguistic model but ... is far too clean to be a model for natural language. ... Natural languages can be models as the solutions in a game theory. ... By this, I mean the functor from the data set, the data set being the ...
    (sci.math)
  • Re: Notation
    ... natural languages, and the former can well be used in a computer program. ... natural grammar ambiguity goes beyond the ... These are the techniques I seek to avoid. ... a language would suffice for a start. ...
    (sci.lang)
  • Why flet wins, was Re: scheme seems neater
    ... But natural languages are quite cluttered. ... Common Lisp is more like a natural language than Scheme is. ... Note that programmers often create namespaces by convention. ...
    (comp.lang.lisp)
  • Re: teaching a child - console or GUI
    ... I don't agree that this is substantively different from natural languages. ... The evolution of natural language follows similar paths. ... I think there is a good argument to made for OOP actually simplifying ... is centered on strict procedural programming. ...
    (comp.lang.pascal.delphi.misc)
  • Re: Are natural languages secure ciphers?
    ... > language where 1.3 billion people know the same secret. ... natural languages do not readily yield to brute-force ... You're thinking so much about Chinese ...
    (sci.crypt)