Chinese economists say Big Data can replace markets in planned economies

Originally published at:


a hybrid of managed, two-sided “platform” markets (modeled on Ebay, Alibaba and various app stores) and central planning informed by machine learning and big data to fairly and efficiently regulate production in a system in which all substantial assets are owned by the state.

What could go wrong?


Late stage communism?


(modeled on Ebay, Alibaba and various app stores)

Damn near spit my coffee on to the screen when I read that.


The utopian version of this is something like the Australia Project from

Notably, the Chinese attempt differs in having humans in charge. Really, that’s a fundamental flaw in any political system humans are likely to come up with. Unfortunately, all available non-human alternatives are not known to be human-compatible.


I have this visual of a train wreck in progress, lasting decades, and destroying millions if not billions of lives.


This idea intrigues me because I’m curious how it would be different than capitalism. Market-focused economics seems to say that what the market does is “right”, that we can extrapolate from what is to what should be. A learning algorithm will work differently than a large group of people acting independently for their own interests, but it won’t necessarily come up with different results.

Maybe this is the system we already have,


Sort of like when they decided Sparrows were pests, and 50 million starved? (Including thousands of recorded instances of cannibalism)


Potentially. The results aren’t always viable; but one thing about trying to run a planned economy is that it requires your economists to hammer out how to model all the things, since the assumption that the invisible hand will deal with the details is even less reliable than usual(and even though it didn’t actually produce a working economy, things like the 30s-40s Soviet work on linear optimization are worth noting).

Now that “big data” is the new hotness, it’s no surprise to see it being pressed into service. It’ll probably be ‘AI’ and ‘deep learning’ tomorrow.

(All that said, none of this is exclusive to communism, or even necessarily led by it: somebody trying to obtain an advantage by modeling a market run into some of the same problems; and even in market economies, markets rarely extend into economic actors: households don’t budget that way; and ERP systems look a lot more like command and control economies, albeit at the scale of a corporation, than they do like mini-markets. (Though, anyone who has had the pleasure of a nice quiet ERP implementation might consider “well, a planned economy is basically just Enterprise resource planning on the scale of all the economic activities of an entire nation…” to be a rather chilling prospect…)


Isn’t that what we’re living through right now?


Bogdanov wrote about this in 1908, before he was kicked out of the Bolsheviks.

(Here’s a PDF of the book)

Chile even tried to do it in the '70’s

(Here’s a New Yorker article about it: The Planning Machine–Project Cybersyn and the origins of the Big Data nation)


Not impressed. To quote Heinlein, “The justification for free enterprise is not that it’s more efficient, but that it’s free.”

Free - if one’s notion of free is to spend most of our lives hustling for resources.


I like the idea of explicitly valuing and weighing freedom against other good things and coming to a reasonable balance. The Chinese government doesn’t seem to weigh “individual freedom”* very highly and the American government seems to weigh it very highly. That’s a question of values. If we’re willing to leave it at a question of values and not try to layer on an objective-sounding pseudo-scientific argument about the market being efficient then I’d say that it’s true that big data can’t replace markets, but it’s also true that we don’t have a great position from which to say that anyone aspire to have markets or a market substitute.

* That is, the sort of thing that people talk about in America when they talk about “individual freedom” which I think @popobawa4u rightly points out might not mean actual freedom to choose what to do with your life.


The problem with this isn’t the requirement that they have a pervasive surveillance state to provide the input data for the model, it’s the fact that the model can’t be trusted to provide good results, even if you could guarantee perfect input data (which you can’t do either).

1 Like

I think the surveillance question is certainly a problem worth considering.

But, yeah, the fact that such a system could exist is utterly beyond belief is certainly another problem.

1 Like

Everything old is new again


oh, of course. and the idea that China might be thinking of implementing it anyway with no idea whether it would actually work or not is a pretty frightening thought.

but it’s important to note that the problem isn’t just with how you get the data, it’s whether the idea of accurately modeling stuff as complex as a human economy is even possible in principle (some people think it’s been proven impossible in fact, but even if that wasn’t true, nobody has yet established that it’s possible).

It’s not a popular perspective, but I think that human economies have always been based upon wishful thinking, and as such they have never been possible in their details. They are just contradictory and obfuscated enough to keep most from complaining about it. Honestly, most deep human systems such as governmental election, resource distribution, family structure, etc are based upon so much BS that they would be more fair as straight-up random lotteries.

Algorithms are the closest humans have come to implementing systems which can avoid self-serving bias. The obvious challenge, of course, being to prevent such biases from being baked into the algorithm.