Re: A few questions about C programming.

From: Arthur J. O'Dwyer (ajo_at_nospam.andrew.cmu.edu)
Date: 10/08/04 

Date: Fri, 8 Oct 2004 11:07:09 -0400 (EDT)

On Fri, 8 Oct 2004, goose wrote:
>
> "Arthur J. O'Dwyer" <ajo@nospam.andrew.cmu.edu> wrote:
>> On Thu, 7 Oct 2004, DEMAINE Benoit-Pierre wrote:
>> [Someone wrote. Please don't snip attributions.]
>>> | If you can't keep the key away from the cracker, you can't keep the
>>> | algorithm away from him either. If he has both the key and the algorithm,
>>> | you are hosed no matter what algorithm you used.
>>>
>>> I am 100% aware of that.
>>>
>>> I don not want a specific algo to be able to stand agaist a cracker who
>>> [has] both algo and key, but I want a generated algo not to look like an
>>> other algo generated by the same generator.
>>>
>>> I need a generator to be able to generate at least 10 000 algos. within
>>> 6 months. I might need 100 000 after 6 months.
>>
>> That's what people who know crypto call "the key." Seriously. You
>> do not seem to understand the basics of encryption and decryption here.
>> You seem to be asking for "an algorithm that generates unique encryption
>> algorithms [given some input from the user]." That's what a cryptologist
>> would call "an encryption algorithm [with a bunch of random keys]."
[...]
> I suspect that DBP wants a "genetic alorithm" way of generating
> algorithms. He does not just want the algorithm being used to be
> secret, what he wants is that even if someone were to reverse-engineer
> a specific alg. out of the 10000 that he generates, they must not
> be able to use that to *guess* the next one in the sequence.
>
> Essentially, he wants an alg. which can generate AES type alg.

   As I said, this is what a cryptologist would call "AES with a random
key." Use a good PRNG to generate the random keys, and then even if
someone were to reverse-engineer one of the random keys, he wouldn't be
able to guess the next one in the sequence. No genetics involved.

> Or, I'm totally wrong on this :-)

   I think you're right about what the OP says he wants, but the OP still
doesn't know what he /needs/.

   (I don't know if anyone's pointed this out yet, but "I don't want anyone
to be able to crack it in a month" and "I want to write it myself but I
don't know much about C programming" generally don't mix.)

-Arthur

Relevant Pages

  • Re: A few questions about C programming.
    ... > algorithms. ... what he wants is that even if someone were to reverse-engineer ... > Essentially, he wants an alg. ... someone were to reverse-engineer one of the random keys, ...
    (comp.programming)
  • Re: A few questions about C programming.
    ... | Essentially, he wants an alg. ... which can generate AES type alg. ... Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org ...
    (comp.programming)
  • Re: A few questions about C programming.
    ... | Essentially, he wants an alg. ... which can generate AES type alg. ... Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org ...
    (sci.crypt)