A random number generator with enough computational hardness

In the web 3.0, all security so far is based on being deterministic while still being too difficult to break. Public-private key cryptography is deterministic, but is still regarded as "zero knowledge", it clearly isn't "zero knowledge" but it's too difficult to obtain the knowledge that is there, it has high computational hardness. The maze hash function (mhf) I described a while back, could be used as a random number generator. It views the state itself as that it from any given position is an almost infinitely long source of random number, and recursively traverses the state based on input (entropy) to get a random hash, a random number.