[ssivark]

There’s the fascinating example of “Nooshare”, which is an interesting proposal for embedding arbitrary Monte-Carlo simulations as the proof challenge! http://web.mit.edu/alex_c/www/nooshare.pdf

It’s an elegant exploitation of the design reasons why the work must involve something like random search (for an easy to validate target).

On squinting, it’s kinda like a smart contract for a compute job, where the compute job itself constitutes proof of work to validate the settlement of the contract!

[koheripbal]

This was written in 2012. It does not appear that this idea went anywhere - not sure why.

It could be the fact that it's not trivially easy to verify the work completed, unlike a hash.