Discussion about this post

User's avatar
Thomas Hvizdos's avatar

Penrose:

I think the core conceit of this article works extremely well rhetorically. It’s attention-grabbing, funny, and thought-provoking. And I think your primary question of: “when do planned economies outperform market economies?” is interesting and worth exploring.

That said, the logic here didn’t work for me. I couldn’t quite tell if the AoE case study was meant to be tongue in cheek or not, but, assuming it was serious, I think there are some serious flaws in the reasoning.

First, Age of Empires doesn’t have an economy. It has a game system themed as an economy, and the underlying structure of the game system isn’t at all like the underlying structure of a country’s economy. The game system is an interesting optimization problem designed to be solved by a single person. There is a single meaningful output: victory in the game. Every piece of the system is an extremely crude abstraction meant to provide flavor to the underlying optimization problem.

An actual economy arguably has no single output to optimize for. It was not created and tuned by a small team of software developers, it is a word used to describe an emergent system.

Age of Empires is a toy optimization problem, not an economy, and the two are different enough that I don’t think you can discover anything meaningful by analyzing it as though it was.

As an analogy: imagine a jigsaw puzzle where each piece represented some discrete part of a country’s economy. When assembled, it created a pictorial representation of an economy. Presumably this would be easier to assemble in a “planned” fashion than by trying to create a model whereby pieces would fall into their correct place according to some algorithm. But this doesn’t tell us anything about economies.

A note about the US healthcare system:

I agree with you that healthcare is a mess in the United States, and would probably function better if it were more similar to the NHS. But it’s not really recognizable as a free market–it’s more of an oligopoly with some government price controls thrown in. Some specifics:

The majority of people get health insurance through their employer, which offers a subsidized plan. You can buy insurance on the free market, but its not subsidized, so it's vastly more expensive than whatever your employer offers. You don’t really have the ability to choose your insurance provider.

Then, when you go get healthcare, you can only go to certain “in network” providers. These providers won’t tell you how much your care will cost, even if you ask, so you can’t shop around and find a good deal. It’s also difficult to get information on care quality, so you’re effectively buying blind.

Medicare and Medicaid are about 40% of healthcare spending in the country, and they do have some price controls. IIRC hospitals lose money on medicare/medicaid patients, and then make up the difference by charging private insurance companies more.

Ironically, I think this market is pretty “efficient” in the sense that it makes people a lot of money, and charges exorbitant prices for something people hold dear. It’s just not efficient in the sense of equitably distributing care. Probably someone could rip apart the system and come up with free market version that worked much better, but at that point you may as well just go single payer.

Answering your specific questions:

Did the introduction drag on too long? I was thinking about cutting down the section on the calculation problem to stream line things, but I wasn't sure if it was necessary context.

I enjoyed the explanation of the calculation problem. I think you went into too much detail re: houses and pencils, but the core concept was good. I did sort of feel like you were strawmanning Misses in the following sections (surely he couldn’t have meant that planning is never useful for economic agents?), so making the case that he did mean that, or cutting the section entirely and relying on most people’s priors about planning being inefficient might’ve been a better tactic.

Is this just too broad a subject for a blog post to shift your priors?

No, but you’d need to provide a detailed real world example of central planning being effective, and I don’t feel you did that here.

Did any of the digressions feel pointless?

The giant diamond thing didn’t really make sense to me, and didn’t feel that important to your core argument. Also, the section on housing didn't feel rigorous enough to convince me that it would be better planned.

Just any thoughts or opinions people have on the topic.

I shared some above, but to close on a positive note: I thought the point about central planning being able to benefit from vastly improved computing power was interesting, and kind of exciting. The analogy of economy as neural net was thought-provoking. That point was probably what shifted my opinion the most about the benefits of planning.

Expand full comment
K. Liam Smith's avatar

> RTS games were my first exposure to economics and a large part of what got me interested in it, I’m sure there are many people of roughly the same age in a similar position.

It made me interested in being a dictator.

> AI strikes me as having absolutely staggering future potential for uses in planning. The technical process of running a plan comes down mapping the inputs and outputs for every industry in a giant matrix, then running a linear programming function to optimise the weights, a process that’s extremely similar to training a neural net.

Could you give me a specific example for the inputs and outputs? I haven’t actually put this on a whiteboard, but my gut tells me that supply chain management is a Traveling Salesman problem, which is NP-Hard. Of course, that means it’s NP-Hard for humans too. I bet that if you measure the complexity of the supply chain, as the complexity rises then the loss surface is going to become more and more non-convex, and then distributed approaches (ie market) economies will optimize it better. If the loss surface is smoother, then a central approach would be better. Again, just a gut feeling.

As for the intro, I would’ve liked to see a concrete example. We started with abstraction and then went to example, and I think this would’ve been more engaging the other way around. In “I, Pencil” the concrete example made something abstract feel very vivid. I want to see something at stake and that thing should seem like a matter of life and death. And this case it is. The Great Chinese Famine is considered one of the worst human-made disasters in history with some provinces losing 1 in 5 people to starvation. But America’s healthcare system is market based and enormously inefficient, and obviously lives are at stake there as well.

Expand full comment
21 more comments...

No posts