Originally Posted by

**laserlight**
It might not be a fluke, but I wonder if you're solving the right problem:

I get the impression that you've come up with your own cryptographic hash function, except that it is intended specifically for password hashing, to address "the inherent weaknesses (and flaws) of the very algorithms that we rely on". The thing is, when we're talking about hashing a password for storage, we're talking about preimage resistance, but (as far as we non-NSA people know) the algorithms commonly used for password hashing these days are preimage resistant, even if it is just plain MD5 applied once with no salt.

Consequently, the algorithms that build on these for better password hashing don't address preimage resistance since attackers are unlikely to approach the problem from that angle anyway. From what I understand, they just assume that if the password is random and sufficiently long, it really is computationally infeasible to find the password given the hash and the algorithm, thus the problem to solve is how to protect passwords that are not so random and/or not sufficiently long.

Do you disagree with this assessment? What are the "inherent weaknesses (and flaws)" that your algorithm fixes?