Current Bitcoin PoW algorithm leads to a block time distribution like this

with is quite sparse. If instead it requires a chain of lower-difficulty hashes (e.g. instead of one hash of difficulty 12 it would be enough to provide chain of 4 hashes of difficulty 10), the distribution would be more dense

This way block mining time (and therefore transaction finalization time) will be more predictable. Is this simply a historical occasion that is difficult to alter, or are there significant drawbacks in this mining algorithm?