the error is you dismiss an omniscient perfectly objective being, while accepting an immortal Invisible Hand
The error you make is humanist. That man could defeat the inviolable Invisible Hand is the impossible goal of all humanist religions that have names such as Communism, Socialism, Fascism, Totalitarianism, etc.. Enjoy the futility (and mega-death of the deviation error from fitness due to top-down control).
Man? You described a being not a Man. No man can hold those properties, but an AI is a whole different matter.
While AI-Central Planning is within the realm of possibilities, The invisible Hand is fiction, and true, a fiction can never be defeated.
This has already
been refuted (by Lindsey Lamport and other Byzantine fault tolerance researchers) because the speed-of-life is not infinite, thus no perspective can be a total ordering. Or stated another way, due to the delay of propagation of information there will exist a plurality of arbitrary perspectives none of which are a total ordering.
Sorry. It is impossible to argue with that truth.
But my pragmatism is, damn the torpedoes and cover thy eyes, ears, and logic. Buy the dips with your student lunch allowance!
The Invisible Hand and Free market is simply an algorithm to walk towards an equilibrium, A Central Planner algorithm simply tries to compute where that equilibrium is. As communication and analytical tech improve those 2 algorithms will converge making the whole debate useless.
Do you think that in a dawning vast network of economic actors, they and not decision bots will make the calls? Will they not exchange info? in effect creating a vast decentralized Central Planner?
Problem is both are wrong seeking an equilibrium that is dynamic and ever changing. One should imo seek the point with minimal distance (cost) from wherever the equilibrium might appear.
The ordering problem and the relativistic reality are no problem just cluster the damn perceptions and work with statistics, no need to be realtime or be precise, if you are off by an hour or a day or miss a few perceptions it wouldn't be that bad.
A K-server problem over a K-means clustering of perceived equilibria, as long as the clusters trajectory (maybe use a Kalman filter) can be anticipated we can solve K-server optimaly. If clusters teleport all bets are off and nothing can help there, that cost will be unavoidable no matter your strategy.
Hell I don't see the reason for all that IT and AI and cryptos if the
goal is not to solve the problem of Optimal Resource Allocation at least partially or even supervised