“RE: The Incredible Non-Uniqueness of an MD5 Hash”

Your math assumes that the hashes the MD5 algorithm generates are equally distributed across the entire space possible with the characters that are being used. I don't know enough about the MD5 algorithm to say whether that's true or not, but it's an interesting implicit assumption in your reasoning.

I devised an algorithm to generate serial numbers for my applications. Since the serial numbers are 20 numbers long, there are theoretically 10^20 possible serial numbers. However, due to the way the algorithm works internally, I know for a fact that there are really only 10^10 possible serial numbers that it can spit out. Likewise, the size of a hash doesn't necessarily reflect the number of possibilities the algorithm will generate. That depends on the algorithm.

But you're probably still safe. I'm just being a pedant. :)

Sean