Is there any research on cryptomoney with central authorities, but also with reduced attack surfaces on a whole system? E.g. authority may be cryptographically bound in some way to only store the database and emit new tokens, but cannot spend them because they get freeze-signed by a receiver to their wallet. Then when you get a payment you check the path of money and algorithmically accept that path only. Anyone who accepts a similar subpath is on their own, because it is double-spending. Subpaths within few minutes self-cancel to prevent instant double-spend.

This is just a vague example, not a working idea. The point of it is that the level of security and trustlessness is not always required to be absolute. E.g. even with fully-secure pow crypto we still have to trust non-crypto claims about usdt, [non?]shitcoins, “hot” wallets, and other maybe-not-ponzis.