Re: MD5 Alive?

Unruh wrote:


Running the Digest on itself say upwards of 1000 times is -- in the
circles that I've been in kind of an accepted thing to do.

Really bad idea. On each level, you loose about as he says, 1/e outputs.


This would mean you loose about 0.66 bits per iteration, so after 194 iterations you would have created a unique collision...

Ie, aftr 1000 times,. you will have only something like 1/e^1000 unique
outputs.


You assume that the iterations are independent. However, for MD5 you're very likely not not stumble upon an additional collision after the first iteration. This is a property inherited by the design.

(I am sure it is more than that, but lets stick with the random
mapping assumption).


Be careful about your assumptions.
.