Research Review: The Use of Knowledge in Society (Hayek 1945)

By | November 23, 2021

Title: The Use of Knowledge in Society

Author: F.A. Hayek

Year: 1945

Difficulty: Moderate, a conceptual understanding of economics required.

TLDR: Must read. A mind blowing prose on information, prices, and the development of civilization!

The central economic question asks: what is the best use of the available means? And branches of economics suggest that we can use ‘economic calculus’ to solve this question. Hayek’s point is that reality is not so simple as the necessary information initially required by a centralized decision making mechanism to initially perform the ‘economic calculus’ may be decentralized and unknowable. In short, it is:

“a problem of the utilization of knowledge not given to anyone in its totality”

F.A. Hayek, 1945

So, with this in mind, Hayek prompts the question: to the extent information is centralized versus decentralized, should the economic decision making mechanism be centralized or decentralized? His answer is that the more efficient system of decision making will be the one that makes the fuller use of existing knowledge.

Centralized information example. If a decision is to be made about something only a particular handful of experts know something about, then they themselves would command the most efficient use of knowledge to make a decision.

Decentralized information example. If a decision is to be made about individuals personal circumstance, thus requiring information about each individuals circumstance, then the individuals themselves are best able to make the decision. But then, how would all these individuals co-ordinate to determine their decisions in relation to every other individuals decision as part of an aggregate economic system? Hayek’s answer is the price mechanism. Prices would help individuals communicate information, thus co-ordinate actions, through constructing constant rates of equivalence and relative importance (i.e. values or marginal rates of substitution) between the goods and services they buy and sell. This would allow individuals to make individual decisions – and act as one market – not because any individual survey’s the whole field, but because their limited fields of vision sufficiently overlap so that through many intermediaries the relevant information is communicated to all. Hayek also notes that the informational signals from prices don’t need to be perfect – and are not perfect – all that is necessary is that they push individuals in the right direction.

“The marvel is that in a case like that of a scarcity of one raw material, without an order being issued, without more than perhaps a handful of people knowing the cause, tens of thousands of people whose identity could not be ascertained by months of investigation, are made to use the material or its products more sparingly; i.e., they move in the right direction.”

F.A. Hayek, 1945

Reflection 1: Is Hayek’s proposition still applicable in the world of big data and AI?

Is the proposition – that dispersed information is unknowable by a central system – still applicable? For example, moving into the future, will Google, Microsoft and Apple know more about my own preferences and means than I do? Do they already know more about my preferences and means than I do? I’m not sure. But, suppose they do and somehow the government ends up with the information (as they tend to). What would the implication be for a central decision making mechanism? I doubt policy makers could comprehend it all (and Hayek agrees). So would we give it all to an AI and implement its conclusion? Would we even be able to understand the reasoning for the conclusion? Or would we be as confused as the professional Chess and Go players after being beaten by AI? The whole idea kind of reminds me of Love, Death & Robots Episode 6: When The Yogurt Took Over. “

Reflection 2: Does iterative policy lessen the extent of the information problem?

I think, for the most part, policy is iterative. Meaning after it is initially devised and implemented, its effects are observed and stakeholder feedback is received, allowing it to then be revised and improved in a subsequent iteration. So on and so on. If this were the case, and policy was error correcting, would it lessen the extent of the central decision makers problem of using decentralized information?

Steady state with feedback example. If reality were a constant steady state, I think it would. The problem would resemble that of reinforcement learning; so long as feedback was received and effects observed, the central authorities policy could be iteratively adjusted until reaching the optimal point. In this case, the central decision making mechanism wouldn’t be that bad.

Creative destruction without feedback example. Does our ever evolving reality and process of creative destruction resemble such a steady state? Probably not. And is appropriate feedback received and adjusted for in an iterative process? I don’t know. So where does this leave us? I also don’t know. But I’d guess with central decision makers playing a perpetual and impossible game of chase the carrot!

A second read of the paper reveals that he kind of touched on this point, but a bit differently.

“Of course, if detailed economic plans could be laid down for fairly long periods in advance and then closely adhered to, so that no further economic decisions of importance would be required, the task of drawing up a comprehensive plan governing all economic activity would appear much less formidable. It is, perhaps, worth stressing that economic problems arise always and only in consequence of change. So long as things continue as before, or at least as they were expected to, there arise no new problems requiring a decision, no need to form a new plan.”

F.A. Hayek, 1945

Reflection 3: Development of civilization more broadly

Despite being focused on the theoretical implication of information and the price mechanism, Hayek also discusses the development of civilization more broadly. For example:

“Civilization advances by extending the number of important decisions which we can perform without thinking about them. The people who deride any suggestion that this may be so usually distort the argument by insinuating that it asserts that by some miracle just that sort of system has spontaneously grown up which is best suited to modern civilization. It is the other way round: man has been able to develop that division of labor on which our civilization is based because he happened to stumble upon a method which made it possible.”

F.A. Hayek, 1945

Read the first sentence again. It is a pretty big claim! So, how should it be investigated? First of all, a note on originality: I did a quick google search and found that although Hayek didn’t cite it, part of this was quote was actually from Alfred North Whitehead’s An Introduction to Mathematics in 1911… lol. For the most part, their arguments are similar and differ on a single point: whether these advancements are made intentionally (Whitehead) or stumbled upon (Hayek).

So, what are some examples of humans/civilization advancing by extending the number of important decisions they can make without necessarily thinking about them?

Example 1: The general development and use of heuristics for decision making. Heuristics help humans think more quickly and efficiently through reducing the cognitive load of making a decision. For example, rather than logically thinking through each step of problem to derive an optimal solution, we can just use trial and error or a rule of thumb to find a quick and okay solution. Thus allowing us to make important decisions without thinking too deeply about them. Lots of other heuristic examples, but I don’t think this example is contentious so I’ll leave it here.

Example 2: The development of low-brain power solutions for mass co-operation. In other words, how did we go from wandering foragers to established co-operating civilization of thousands? A common language seems like a good first choice. The development of shared societal structures and values seems like a good second. And the third would be Hayek’s own point regarding the price mechanism for efficiently allocating resources. Were these mechanisms advanced intentionally or stumbled upon after thousands of years of trial and error? I don’t know.

Example 3: The computer revolution. I write this on a computer that although I built, I know little about, to post on an internet that I know even less about, with the assistance of OpenAI’s GPT-3 that I basically know nothing about.

“Society is composed of persons who cannot design, build, repair, or even operate most of the devices upon which their lives depend…In the complexity of this world people are confronted with extraordinary events and functions that are literally unintelligible to them.

Autonomous Technology: Technics-Out-Of-Control (1989), Langdon Winner

Example 4: In the future, what will the AI revolution look like? And in this field, to what extent have we made intentional advancements or just stumbled upon them?

“The algorithms that make all this happen are no longer understood by any one person. They optimize for what their creators tell them to optimize for, but in ways that no human could figure out — they are what today seems like sophisticated AI, and tomorrow will seem like child’s play.”

Sam Altman, CEO OpenAI 2018 ish

“People who work in machine learning simply didn’t think that neural networks could do much. People didn’t believe large neural networks could be trained…The ideas were all there, the thing that was missing was a lot of supervised data and a lot of compute. Once you have [those two], then there is a third thing is that is needed—and that is conviction. Conviction that if you take the right stuff, which already exists, and apply and mix it with a lot of data and a lot of compute, that it will in fact work. And so that was the missing piece.”

Ilya Sutskever, Chief Scientist OpenAI 2020

Reflection 4: Summary of other implications

Happy Birthday Hayek! (

One thought on “Research Review: The Use of Knowledge in Society (Hayek 1945)

Comments are closed.