[Cryptography] Cryptanalyzing a whole-message cipher and a double-tree hash function

[Pierre Abbat]

On Wednesday, December 27, 2023 7:14:08 PM EST Jon Callas wrote:
> In other considerations, I looked at your code and would really like to see
> a paper that describes what you're doing, what problems you want to solve,
> what mechanisms you're using to construct things, and so on. When the code
> is the documentation, then there's no difference between a trivial bug and
> your thing being broken, for example.

I'm better at writing code than specifications, so I wrote the code in two 
languages, with a shell script that runs both and compares them.

> There are a number of questions I have that I know less about for having
> looked at the code than I did before. As an example, you said in your first
> message that you want to encrypt a whole message. Yet, what does that mean?
> Are you building a very-large-block cipher where any one-bit change causes
> the whole thing to decrypt differently? Or are you doing a chaining mode or
> something like it?

It encrypts an arbitrarily long message (as long as calculations involving the 
length don't result in arithmetic overflow), and flipping one bit results in 
flipping half the bits throughout. It can be used as a block cipher with 
arbitrarily long blocks, such as the payload of an MTU, and in any of the 
block cipher modes.

> I freely admit that I spent probably a whole five minutes (okay, maybe ten)
> looking at your code and I couldn't really see what the organization of the
> cipher, its scaffolding and test functions. I'm lazy and really want it all
> spoon-fed to me. I also think that if you wrote a paper explaining
> everything, it would force you to think about it in a different way than
> just the code.

I'm sending you off-list a draft of a paper I'm planning to submit to a 
journal. It's only a draft so far, and I don't want to dump the PDF file, or 
even the LyX source, in lots of people's inboxes.

Pierre
-- 
When a barnacle settles down, its brain disintegrates.
Já não percebe nada, já não percebe nada.