sorry with all respect but Expensiveness is related with how many zeros have the target because you have too loop more.
That is why you waste more energy (more expensive) now than years ago.
That is why you waste more energy (more expensive) now than years ago.
If the problem was adjusted where it wasn't using a double hash, and had some simple two-way function than it wouldn't be considered computationally expensive. That is irrespective of the length of the nonce. E.G if it was just a how much do I add to get a nonce of length x.
The complexity here is caused by the double hash.
The length of the nonce is just an even further scaling of an already difficult problem.