The problem is that I can author "foo bar" and "moo bar" constructed to hash the same, and then assert retroactively which one I meant. While collisions are likely to have random binary garbage in them, it may not matter - I am on phone now, but IIRC I have two PDFs on my desktop that have the same MD5, one of which viewed is an airbus pamphlet, the other being a Boeing one. (created by Dan Kaminski, IIRC)
Comments
That's not why it would be a problem.
The problem is that I can author "foo bar" and "moo bar" constructed to hash the same, and then assert retroactively which one I meant. While collisions are likely to have random binary garbage in them, it may not matter - I am on phone now, but IIRC I have two PDFs on my desktop that have the same MD5, one of which viewed is an airbus pamphlet, the other being a Boeing one. (created by Dan Kaminski, IIRC)
Can we see them?
I guess they are on a backup or misremembered - but examples are really easy to find on the net if you want them:
http://th.informatik.uni-mannheim.de/people/lucks/HashCollis...
http://www.win.tue.nl/hashclash/ChosenPrefixCollisions/ has a multicollision: 12 PDF files with different content and the same MD5 hash.