Cass Sunstein, guest-blogging for Lessig this week, has had a number of posts on Hayek (here and here). Most of the posts have actually been on why certain institutions, such as blogs, open-source software, Wikipedia, etc., are not actually Hayekian information-processing mechanisms. That is correct. There is a fundamental difference between a Hayekian information-processing mechanism, such as the price system, and the type of information-processing mechanism that Sunstein is exploring.
Oversimplified, the distinction is between those systems where the purpose is simply to centralize dispersed knowledge in one place versus systems that centralize knowledge for the purpose of decentralizing it again.
[Continue reading "Sunstein on Hayek" under hidden text]
Hayek is interested in coordination among many decentralized actors, not the centralization of information as an end in itself. He is interested in the way in which seemingly centralized systems, such as prices, emerge as an unintended consequence of decentralized individual decision-making. Nobody who is using the price cares if the price system produces the "correct" or "right" price (unlike the Wikipedia)--each individual just wants to know whether if they want to buy or sell something that they will be able to find someone to trade with. The price system helps them to identify one another and to conduct transactions. The marvel, Hayek notes, is that through individuals acting selfishly at all of these decentralized decision nodes, you get this spontaneous generation of a "price" that is a short-hand for all of their decentralized knowledge and decisions. The price, however, was nobody's intentional design--unlike the Wikipedia or open-source software. In fact, this squarely contradicts the central purpose of the Hayekian insight--Hayek's argument is that the value in prices is in the tacit knowledge that it conveys (i.e., the decisions and expectations of many decentralized actors), whereas Sunstein is interested in the explicit knowledge captured in these systems.
The value of prices, contra Sunstein's information-aggregation systems, is that prices permit the "centralization" of information in a short-hand form (market price) for the "purpose" of then decentralizing back to the individual market decision-makers for whom it provides incentives and information about the relative scarcity and preferences of many disaggregated market decision-makers. I put "purpose" in quotes, because of course, spontaneous orders have no purpose of their own, but in some sense market actors are responding to the information and incentives that prices provide in taking their own actions.
Thus, Hayekian information-aggregation mechanisms are not centralized information-collecting mechanisms of the type Sunstein discusses. Rather, they are short-hands for each dispersed individual being able to predict how other people are likely to act, which then allows each of us to plan our own actions and to coordinate better. Even better than the price system in illustrating Hayekian information aggregation may be language. Given terms within a language are wholly conventional--all that matters is that everyone else understands what you trying to communicate. So the correct word, like the correct price, is simply that word/price that enables people to interact with the maximum degree of coordination and minimal degree of friction. It is largely nonsensical to think about what would be the "correct" word for "car," "boat," or "train".
Similarly, customs and traditions, Hayek argues, are like this. They are created in a decentralized manner, and to the extent that people act in accordance with them (or more precisely, respond predictably to the incentive structures that they produce) the existence of these "rules" can help people to predict one anothers' behavior. You don't have to know why a given tradition was "created" in order to follow it (sometimes traditions such as dietary restrictions or other "taboos" may actually have a sound scientific basis, but that is largely beside the point). Here, however, Hayek does take the next step and argues that there is a metalevel competition among different sets of rules and traditions, cultural group selection, which does in fact select for the "best" rules, but again, in the empirical sense that groups that follow those rules will prosper relative to others. (I have written about this aspect of Hayek's thought here.)
Law too, is such a system. Hayek argues, in a forerunner to modern law & norms, that the "best" legal system is one like the traditional common law, which collected information from case-by-case adjudications of specific disputes that arise among people and infer more abstract principles from those, rather than law-making by centralized legislatures. Again, the point is not just that the decentralized information is being centralized; rather, the central point is that by collecting information from decentralized actors, and aggregating them into coherent rules, judges can then send the information back out to individual actors who can then make use of that knowledge to predict how others are likely to behave. The point of the information aggregation is simply that the rules produced are more likely to be "good" rules in that the reflect how individuals are actually acting already and so that rules grounded on that foundation are more likely to further coordination among decentralized actors, as opposed to rules that lack this foundation. Again, the point is that these rules will enable each individual to better predict each other's likely behavior, and to make his or her plans accordingly.
In other words, legal rules that build on spontaneous orders will, in general, promote coordination better than rules that run against the grain of spontaneous order. (Whether there is some normative reason to dislike the rule, or whether a given rule may be only a local rather than global optima and so should be altered, are different questions from the observation that rules that follow rather than contradict individuals' settled expectations will promote coordination better than those that run against them).
In an article I published several years ago, I argued that this is also the purpose of law, properly understood--to distribute information outward to individual decision makers on the periphery about incentives and expectations so as to allow individuals to better coordinate their affairs with one another and to allow the spontaneous order of the market to flourish. In turn, recognizing this tells us about what our legal rules should look like. Unfortunately the article is not available on-line, but the cite is Epstein & Polanyi on Simple Rules, Complex Systems, and Decentralization, 9 Const. Pol. Econ. 143 (1998). The rule of law, properly understood, also furthers this end of promoting coordination by dispersed individuals; it does not have an "end" of its own.
Conclusion: So Sunstein is absolutely correct to note that the information-processing mechanisms in which he is interested are not actually Hayekian systems. Hayek is not interested in the centralization of knowledge for knowledge's sake. Rather, he is interested in the way in which certain institutions (such as prices, language, and traditions) centralize huge amounts of information, boil it down into tacit knowledge, and then redistribute it to decentralized decision-makers in the form of prices, rules, traditions, etc. The "purpose" is not to collect the information at the center in order to make it more "accurate" or "better"; the purpose is to send it back out to decentralized decision-makers in order to allow them to better coordinate their affairs with one another.
If I have time later today, I will explain that I think the model of information-processing that Sunstein has in mind may actually be something more like the Condorcet Jury Theorem than Hayekian information aggregation.