« the lessig blog: where the vacation lineup is better than the main show | Main | Extremism and Information Aggregation »

On Aggregating Information: Hayek, Blogging, and Beyond

This is from Cass Sunstein; I'm most grateful to Larry for inviting me to post on his blog for a bit. His kind invitation is a result of a naive and ignorant inquiry I sent him in the recent past, about information aggregation and its possible limits.

Background: A few years ago, a book of mine, Republic.com, emphasized the risks associated with echo chambers and self-insulation. I'm doing a new book, still inchoate, that continues to explore those risks, but that also stresses the excitingly general possibility that the Internet can allow widely dispersed "bits" of human information to be properly aggregated -- as, for example, through open source software, wikis, prediction markets, and even blogging.

All this is pretty abstract, so let me try to focus it. One of the greatest arguments of the twentieth century is Hayek's about the price system. In particular, Hayek claimed that any "price" is capturing the information and tastes of many people, in a way that will outperform the judgments of even the best experts. Hence prices do a lot better than any central planners. Here's a puzzle: How close is the analogy between the price system on the one hand and wikis, open source software, and even the blogosphere on the other? Where does the analogy break down? When, in particular, will wikis and the blogosphere fail as mechanisms for aggregating dispersed information? I'll venture some thoughts before long, but for the moment I'd just like to pose the question. I know that there's a lot of information out there about all this; any help would be appreciated.

(originally posted 7/16/05)

| | technorati

TrackBack

TrackBack URL for this entry:
http://lessig.org/mt/mt-tb.cgi/1700

Listed below are links to weblogs that reference On Aggregating Information: Hayek, Blogging, and Beyond:

» Blogs and information aggregation from Stephen Laniel’s Unspecified Bunker
Cass Sunstein — author of Republic.com — has taken over Larry Lessig’s blog for a short time, and for his first post he’s asked if there are any analogies between the price system and distributed Internet projects, like blogs and f [Read More]

» Cass Sunstein: aggregation as market from Pokkari Blog
I’m currently reading Cass Sunstein’s excellent Republic.com, which discusses the lack of a commons and the lack of a shared narrative on the Internet. Today Lessig reprints an e-mail from Sunstein, in which Sunstein talks about his new b... [Read More]

» Cass Sunstein is blogging at lessig.org from Blogs at CIS - Lauren Gelman
While on vacation, Larry is hosting a great list of guestbloggers-- this week it's Cass Sunstein, a superstar professor from... [Read More]

» Lessig vacation lineup from aTypical Joe: A gay New Yorker living in the rural south.
I do so admire Larry Lessig. Here I am on vacation, blogging, while his vacation starts today. No blogging for him; he's promised his family a month away from the Internet each year. Maybe one day. But not this week.... [Read More]

» SUNSTEIN'S POST ON WIKIPEDIA AND HAYEK from Knowledge Problem
Lynne Kiesling Cass Sunstein is currently guest-blogging at Lessig's place, and has a thought-provoking post on information aggregation, Wikipedia, prices, and Hayek. The post itself and the comments have a lot of good content, and I recommend reading ... [Read More]

» Sunstein on Hayek: from The Volokh Conspiracy

Cass Sunstein, guest-blogging for Lessig this week, has had a number of posts on Hayek (here and Cass Sunstein, guest-blogging for Lessig this week, has had a number of posts on Hayek (here and

[Read More]

Comments (22)

An interesting question...

I think the analogy starts to break down because of the different levels of commitment involved in exchange pricing versus information sharing.

With prices, each party involved is giving something up in the transaction and therefore the interest in maximizing their own benefit drives a certain level of involvement.

Information sharing in wikis, blogs, etc. depends on a different set of motivations that cannot be counted on to drive involvement or quality of involvement.

So the difference comes down to the following examples:
1. If I am paying my own money for something, I will do all I can to get the lowest price possible. The seller will do everything possible to maximize the price. Knowledge of similar transactions is helpful in arriving at an agreeable price.
2. If I see some information on a blog that I think is incorrect, I might not do anything to communicate my disagreement other than "change the channel". At the same time, if I see something I agree with, I will be even less likely to do something to register my agreement.

Perhaps this is simplistic but it comes down to the perceived cost of not acting to fix bad information or not acting to confirm good information.

To posit the failure to anticipate, you appear to assume that Wikis and the blogosphere currently succeed in aggregating dispersed information. I think it would be wise to inspect this premise before you proceed.

With respect,
Steven

I mean to say:

To posit the failure you anticipate, you appear to assume that Wikis and the blogosphere currently succeed in aggregating dispersed information. I think it would be wise to inspect this premise before you proceed.

With respect,
Steven

Price signals are excellent for commodities for that are fungible, and where the market provides transparent information (grain futures), and where actors are making rational decisions

Price signals are more problematic where the content is not transparent, fungible, or is not driven primarily by reason. A Technorati search could tell you the relative weight of the memes "valerie plame was outed for political revenge" and "karl rove was preventing a political hack from smearing the president."

But the Technorati algorithm result won't get you closer to the truth.

I was fortunate to read Dr Sunstein's "Democracy and the Problem of Free Speech" when I was a student. Now IT consultant like myself are very often confronted with free speech issues - indeed today any serious internet project has some public collaborative aspect, usually because it accepts user comments. For example Amazon derives an essential part of its value to customers from the 10 million-plus comments posted (free of charge) by its users. I found in "Democracy" an excellent framework to counsel customers (e-retailers) as to the moderation rules they should adopt concerning the content provided by the public.

Now to Dr Sunstein's project:
I am a (modest) contributor to the open source software (OSS) effort.
At first sight I would not think that the price mechanism is the most appropriate framework. Most individuals contribute to an OSS project for reasons that are only remotely connected to the project's potential "market price", that is the current price of existing, unsatisfactory solutions. Usual motivations are desire for intellectual recognition, sheer geekness, learning (participating in a OSS project is the best teacher), Microsoft-bashing, etc.
The motivation framework is that of academia, not of commerce.
Of course OSS solutions are now extremely valuable, but their value is captured not by their creators, but by their users.

Perhaps Dr Sunstein could explore the framework of transaction costs (Ronald Coase: "The Nature of the Firm", 1937).
The optimal size of any organization is determined by a trade-off between:
- decentralization: the desire to procure resources at market prices for better efficiency, which pushes to reduce the size of the organization (for example it would rather use contractors than long-term employees)
- centralization: the costs associated to organizing work, which push to increase the size of the organization: by using long-term employees a manager does not have to worry about constantly finding/ training resources, which is costly.

The internet has massively deplaced the equilibrium point towards more decentralization, hence smaller organizations, by providing efficient and cheap collaborative tools: news lists, project sites, forums, etc. as well as near-free distribution channels for its products (bandwidth).
Of course this is relevant for the economy as a whole, not only for OSS projects, but it is most spectacularly demonstrated by OSS.

>> When, in particular, will wikis and the blogosphere fail as mechanisms for aggregating dispersed information?>> When, in particular, will wikis and the blogosphere fail as mechanisms for aggregating dispersed information?

I don't think they have succeeded yet, in general, though there are a few prominent exceptions.

One area where wikis can do very well is in aggregating the desires of a group of consumers for particular content, by enabling them to work together, each doing a small amount to produce a larger aggregate work which all can then benefit from. The Wikipedia encyclopedia hosted by the Wikimedia Foundaton is one good example of this.

Another potential example is business directories, which tend to be highly priced aggregations of information already held the consumers of the directory, who don't yet work together to remove the middleman.

Threats to this model include license agreements which place the host rather than the community as the owner of the content (CC licenses with attribution to the site rather than the author, for example), since those reduce the community ownership, making it more like a donation of content to a publisher than a community work for the benefit of the community.

Overly powerful wiki primary hosting sites also threaten this model, since they could develop the power to make legal threats against their "competitor" licensees, who may not have the legal resources to defend their rights against the large licensee. This is one of the longer term threats to the Wikipedia encyclopedia, if the Wikimedia Foundation was ever to find itself wanting large sums of money for something, which might tempt it to make it hard for other licensees to comply with the GFDL license, perhaps by making it impossible or difficult to get at the fulll GFDL history without paying service fees of some sort.

Author is one of the Wikimedia/Wikipedia technical team.

Looking into the bull whip effect in supply chains might be useful to you. I think it is an effect that will interest you. of limitation effect you. Supply chains can become unstable. They are a series of repeated transactions that are suppose to tend to arrive at and adjust to a ideal fullfillment rate. It does not always work out that way. Hau Lee's Information Sharing in a Supply Chain and related work on trust effects

Best Regards,
Susan

"Hayek claimed that any "price" is capturing the information and tastes of many people, in a way that will outperform the judgments of even the best experts. Hence prices do a lot better than any central planners."

I'm beginning to think that Hayek and prices has become similar to the Heisenburg Uncertainty Principle or Godel's Incompleteness Theorem. Yes, there is a deep, fundamental, insight there. But the domain of true applicability is dwarfed by the metaphorical services. In particular, Hayek/prices has a very strong tendency to turn into market-worship (I'm not saying it's happened here yet, but there's a tendency one should beware).

"How close is the analogy between the price system on the one hand and wikis, open source software, and even the blogosphere on the other?"

Oh, I'd say about the way probability theory on the one hand relates to schools, political parties, and even the government itself on the other.

"Where does the analogy break down? When, in particular, will wikis and the blogosphere fail as mechanisms for aggregating dispersed information?"

As Steven Ericsson-Zenith said "you appear to assume that Wikis and the blogosphere currently succeed in aggregating dispersed information. I think it would be wise to inspect this premise before you proceed."

Open Source is a business-model. Wiki's are group-collaboration tools. Blogs are publications. If you are asking "When does a more-decentralized business model work? When does group collaboration work? When does data-mining of publications work?", those are wide-ranging topics. But it's not all that useful to say something which may boil down to "Decreasing costs of collaboration lead to an increased ability to better support models which depend on collaboration". That would be a bit like "Good roads and cheap gasoline lead to an increased ability to support mail-order businesses and more distributed supply chains". It's certainly true. But very limited in understanding the dynamics of a successful mail-order businesses or distributed supply chain.

I like the idea of dumping raw materials into a container attached to a computer, hitting the enter key, and getting a finished product. I also like the idea of placing a product that I created, into a container attached to my computer, and sending it to a computer on another continent, where it is duplicated, and the receiver can open the container and hold that product in my hands.

A villager, in a community far from any metropolis, can sell their song, and anyone in the world, can purchase it. Will the day come, when remote manufacturing permits that local villager to create a hand-crafted item, and sell it to a worldwide market, in the same way? If so, will pricing structure operate in the same way it does now?

I think, what is described above, is called, "Gift Giving Economy".

Hayek's Theories of price and information work quite well but they are ultimately limited by the quality and correctness of information that people work from. Do market signals always lead people to the most market efficient solutions? Does this feedback loop always work?

And so it is similarly with the blogosphere and wikis. I think the difference that breaks the analogy is not that they are so much like parallel tracks but more like radial lines leading towards the same ill-defined center point.

I just wanted to say that Hayek's work on price theory is central to my own thinking about how to manage the Wikipedia project. Possibly one can understand Wikipedia without understanding Hayek, since perhaps my own theories of how Wikipedia works are false. :-)

But one can't understand my ideas about Wikipedia without understanding Hayek.

--Jimbo (founder of Wikipedia)

I'm wondering how to measure whether a wiki/blog etc is accurately aggregating dispersed information.

With commodities there is a quantifiable measure, i.e. the price.

However with information in a wiki, I don't see that there exists a quantifiable measurement of truth.

So until we find a way of putting truth into some kind of quantifiable measurement, I think the Hayek/pricing model is more useful as an analogy than it is as an analytical tool.

When, in particular, will wikis and the blogosphere fail as mechanisms for aggregating dispersed information?

I am not totally sure what kind of information you are talking about. Are you talking about individual preferences that are aggregated by the Hayek-like price system (like in, e.g., the Googld Pagerank system) or are you talking about content that is produced by a dispersed set of people (like in, e.g., wikipedia)?

I am not sure whether the distinction is ultimately important, perhaps both categories can be translated into each other. But one difference could be that, at least under certain traditional assumptions, the creation of a preference by an individual does not depend or build upon the preferences of other individuals, whereas a wikipedia-like content production system crucially depends on building upon and changing already-existing information/content. So the kind of information that is produced in various systems could differ.

It is my theory and observation that in the open internet content links are self-vetting. I base that on ten years of online experience with nuggets (links/nodes) of subjects studied in schools and colleges. When a superior link of such a subject appears online before long it is found by everyone interested - and most importantly by the experts in its field. This is even better than the google effect. The experts on a particular small topic all use the same few links. Also, students of a topic flock to the best website for that topic. This does not apply in a big subjects like chemistry or history. But it does apply for granular topics like the table of chemical elements or the study of the Phrygians.
The students flock to this table of elements:
http://www.webelements.com/
The world's experts on the Phrygians are probably all linked here:
http://phrygians.com/

While Hayek's ideas about spontaneous order and the "products of human action but not human design" are invaluable in understanding internet phenomena such as blogs and wikis, I would be cautious about drawing analogies to the price system per se. Prices are aggregates, and as such one of their most important functions is to discard information; that is, they are a form of lossy compression. Indeed, one of the amazing things about prices is how they discard nearly everything there is to know about something while retaining just exactly the bits needed to guide the behavior of diverse actors in the marketplace. Moreover, they do so in a way that is relatively automatic, in that the aggregation computation is emergent rather than under any individual's explicit control (which is what makes markets such robust institutions in the face of intelligent, highly self-interested agents trying to game the outcome). Blogs and wikis are also aggregators, and thus also lossy, but the aggregation/lossiness is under the deliberate control of the authors and shaped by the authors (albeit with varying success) towards particular ends of their choosing.

One might perhaps be able to speak of the economy (using that word very loosely) of links in the blogosphere in terms of a price analogy, in that individual links are under individual control while the wider universe of links is not. This might be a fruitful avenue for investigation (indeed, the impersonal nature of the larger universe of links on the web is one of the key phenomena that indexing services like Google exploit to extract useful information in the face of the self-interested behavior of web site operators).

Jimbo,

You might think more about how the Wikimedia Foundation resembles a government central bank and how to structure things to avoid the negative effects Hayek expects from central bank operations.

Hi,

You could do worse than look at Yochai Benkler's piece on sharing nicely which touches on some of these issues: http://benkler.org/SharingNicely.html

Chip,

That's a fascinating comment, but to draw on what Alan said in the first comment, what is a linker giving up? In my blog, I tend to link only rarely, perhaps because I'd like to both focus my readers' attention on a small set of things, but perhaps because an out-link is competition for attention between myself and someone else?

Here's the key connection. The market is so immense as to approach what I call "practical infinity," the inability of a human being to fathom or even imagine its breadth. The net is the same. There isn't a free market, but there are entrepreneurs who, within some framework acceptable to them, operate in relative freedom. The best of these know that the unfathomable will send them the unforseeable, they just don't know what. They lack the hubris to see the system as closed. "Price" isn't the only issue; there is also the question of discovering goods and services you didn't know you needed before you entered the market.

By the same token, those who succeed in having a viable net experience, whether it be doing business, seeking information, conecting with other people, or what have you, take the same fresh approach. They greet the unforseeable on its own terms.

July 19, 2005 7:55 AM Shaun Evans:

It seems to me that the most relevant distinction between the price mechanism and blogs/wikis is that information is aggregated as a side effect in the price mechanism, but is the primary goal in the blog/wiki world. When I go to the grocery store, I have no desire to share my preference for fresh pineapples, but it happens anyway.

This means that Google is actually much closer to the price mechanism in terms of it's information aggregation process, because the Google rankings emerge as a second order effect from the nature of links on the internet.

Prices are aggregates, and as such one of their most important functions is to discard information; that is, they are a form of lossy compression. Indeed, one of the amazing things about prices is how they discard nearly everything there is to know about something while retaining just exactly the bits needed to guide the behavior of diverse actors in the marketplace. Moreover, they do so auto parts in a way that is relatively automatic, in that the aggregation computation is emergent rather than under any individual's explicit control (which is what makes markets such robust institutions in the face of intelligent, highly self-interested agents trying to game the outcome).

March 7, 2007 12:33 PM Ken Rushinzky:

True and valuable information is a collective good. Participating in providing the "input" that over time will provide valuable information is not costless. So, even with such instruments as wikis and blogs, there will be a fundamental free-rider problem and suboptimal provision of the good. Each transaction in the market is different (not an analogy), even though the market as such is a collective good.

Post a comment

By entering the words in the box, you are also helping to digitize texts that were written before the computer age. The words that you see were taken directly from old texts that are being scanned and stored by the Internet Archive. This CAPTCHA helps proofread the books. If the sample is too hard to read, click the recycle button to get another two. A space between each word is required. And thanks for the comment and help.




--> 'use strict';