mailing list archives
Re: Month of Random Hashes: DAY THREE
From: Valdis.Kletnieks () vt edu
Date: Fri, 15 Jun 2007 16:56:52 -0400
On Fri, 15 Jun 2007 16:59:01 -0300, "M.B.Jr." said:
but only one string can produce that md5 hash signature,
that sha1 hash signature, fucking that sha256 hash signature, fucking that
<any_other> hash signature, etc...
Nope. There's an infinite number of strings that would produce the same
MD5/sha1/sha256/whatever hash. The interesting point about such hashes is
that although given a particular string A, we can *easily* compute the hash H.
However, knowing H, we don't have a good way to recover A, nor do we have any
easy way to compute a *second* string B that hashes to H.
So, given a hash H, we know one of 3 things is true:
1) The person we got H from has A, and easily computed H.
2) The person doesn't have A, but does have either a way to use several million
CPU-years or a crypto breakthrough to compute some string B that also hashes to H
3) The person just pulled a pseudo-random string of bits out of their ass,
called it H, and has as little clue about A and B as we do.
At the current time, (2) is believed to be impractical, and (3) fails the
instant the person actually has to produce A itself. As a result, we can
usually presume that if they have a hash H, they've got the A it hashed from.
This becomes interesting if you want to prove that you have a prior claim on
something, without revealing the something (for instance, an advisory or PoC
for something while you're still working with a vendor about fixing it) - you
can (for instance) post the hash of it on May 1, release the announcement on
July 1, and when others dispute your claim you knew about it on May 1, you can
point to the hash from May 1, and show it's the same as the hash of your July 1
announcement, and thus prove you knew about it back on that date.
Full-Disclosure - We believe in it.
Hosted and sponsored by Secunia - http://secunia.com/