How do you account for “let the world burn”-type actors in this scenario though? It seems this pulls at the key misunderstanding around crypto and blockchain in general — it can’t sanitize the endpoints.

If any human party has the power to defraud, manipulate, or commit any other nefarious act (as you imply China does), the system is vulnerable.

This is why proof of work provides some level of resilience—you have to be nefarious AND have access to massive compute resources to do harm. China has the compute resources, but not the incentive. A proof of stake power broker may have the incentive and not need the compute.