Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
Re: Collisions for Hash Functions

From: Artur Grabowski (artblahonga.org)
Date: Thu Aug 19 2004 - 17:54:13 CDT

Erik Norgaard <norgaardlocolomo.org> writes:

> The fact that you can may find a colliding message isn't that
> dangerous, the distance between meaningfull messages is huge, and
> the distance between meaningfull and usefull messages even
> greater, it would be a problem if you could make a message of
> choice collide.

Read the recent publications, especially about the problems in md5.
IIRC in one of the published collissions had less than 10 bits
differing between two plaintext that generate the same hash.

10 bits is more than enough to change:
"Release not, send to Gulag."
"Release, not send to Gulag."

I think the original message was in polish, and yes, someone edited a
message like this a 50 years ago or so when the guards weren't looking
and got away. Don't remember the full story.