[Cryptography] Solving Matt's hash problem

[Kyle Butt]

On 11/16/22 15:51, Viktor Dukhovni wrote:

> On Wed, Nov 16, 2022 at 02:56:58PM +0000, Weger, Benne de via cryptography wrote:
> <snipped>
>
> This is also different from quines, because the article is not some
> animated GIF, or program, ... that "prints" its owne hash, but rather
> just "natural" (given) text that is not specifically cooked to contain
> variants of colliding blocks.

Small note on terminology: a "Hash quine" is not properly a quine, but
a document X that contains a representation of H(X) for some H. The
name is borrowed.

> I took the problem as asking given a fixed given enclosing message (an
> article on the subject of hashes, ...), with a variable slot just wide
> enough for a hash, to find a hash that would then be the hash of the
> whole article *including* that hash.

The article was in HTML. It makes sense to ask if you can solve the
problem for a given format. For plain text it involves a fixed point,
but so many of the formats that we consume are not plain text. For
binary formats where the hash is checked, the bytes of the hash are
almost always contiguous.

The web not being plain text shows up in copy-paste vulnerabilities,
where there are hidden characters that get copied when you copy some
command from a webpage.