I understand that the desire for a Proof of Work to be useful is strong and that's what is the appeal for these kinds of things. You should know that everyone would be totally onboard if there were a way to keep critical cryptocoin properties AND make the computation useful. The naysayers here don't dislike folding or any other useful computations. They just have experience with the cryptocoin design elements.
There's some subtleties to how a cryptocoin works that aren't obvious and they're absolutely critical to making it work. Bitcoin can have competing chains of blocks that eventually get orphaned. And it's because there's this race condition while block solutions are propagated across the network. Eventually, after enough blocks the network agrees on which chain wins. But imagine if you could know how the next block would start - what its inputs would be. You could start solving that block now.
A penalty mechanism is possible that disincentives such behavior and makes it impractical. Filecoin has such a system. But a powerful enough mining pool (China has 65% of mining power) can also do this for Bitcoin.