I initially took issue with "first asymmetric proof of work", thinking that in practice any NP-complete problem would surely be fine, even if we can’t prove that it can’t be solved quickly. But is it difficult to find NP-complete problems which can't also be quickly approximated? And I suppose it might also be hard to find NP-complete problems in a "Goldilocks zone" of difficulty to make for practical proofs of work, since their complexity increases so sharply.