Hacker News Reader

Comment by smallnamespace | Hacker News Reader

In the limit, a central plan that works well is going to contain a detailed simulation of something that looks like a market economy, with rational actors modeled out at every step.

Turns out that at our current state of technology, doing that isn't feasible yet, but that may change at some point.

View on HN

Eh, not necessarily. Markets don't find the optimal solutions for every problem. Look up "Price of Anarchy" in game theory. I don't claim that it's feasible to come up with central plans that beat markets, but there are cases where central planning can produce objectively better results.

Except that once you plan hits reality, the rational actors in the real world can look at the plan and change their behaviour to take advantage of it to serve their best interests.

It's like creating a fixed plan for each move in a game of chess where you can't change your game plan once the game starts, but your opponent can and also gets to see your plan at the beginning of the game.

No, it isn't like that. What GP is most likely suggesting is central planning done second-by-second, not year-by-year. Think less "government decrees" and more "closed-loop feedback control".

I suppose there exists a Nash equilibrium, where every actor can't act in a way that will decrease the effectiveness of the plan. Problem is that it's way too computationally expensive to calculate one for a 10 player Hold'em poker game, let alone an economy with millions of decisions and actors.

Reply on news.ycombinator.com