Review: Infotopia. How Many Minds Produce Knowledge – Cass R. Sunstein

Oxford University Press, New York, NY, USA 2006
288 pp. Hardcover, $16.50 USD
ISBN 0-19-518928-0
Buy at Amazon

Infotopia deals with the main question of how many minds produce knowledge. Sunstein has a legal background and takes on a mechanistic approach to political theory.

One way to produce knowledge is through deliberation Habermas-style but it has four flaws:

1. Amplification of group errors
2. Information is not elicited
3. Cascading effects
4. Group polarization

Sunstein sees prediction markets Hayek-style as a way to overcome these flaws. His unified theory does not leave much room for hybrid forms. However, not all issues are suitable for prediction markets nor for deliberation so what would be really interesting is to study hybrid forms. Knowledge production that takes the best of deliberation and prediction markets.

While the book has some flaws such as loose end definitions, a unified approach and a lack of critique on aesthetics it is a very interesting read. The book made me question democracy, polling systems, the Amazon recommendation system, Digg, the blogosphere and tons of other things. In the conclusion Sunstein translates the systems he describes into deliberative recommendations which might make him a deliberative democratic after all.

What interested me most in relation to my thesis was:

Some group members that are “cognitively central,” in the sense that their knowledge is shared with many other group members. […]
But other group members are “cognitively peripheral,” in the sense that their own information is uniquely held; what they know is known by no one else.

This reminded me of a decentralized network where the hubs are central and the nodes are peripheral. Let me quote Alexander Galloway on the topological/structural qualities of networks:

centralized network–a single hub connected to multiple peripheral nodes; little to no interconnection between nodes.
examples: web server, software security updates, the panopticon prison, sovereign (royal) fiat, LAN router, pyramidal hierarchy

decentralized network–multiple hubs, each with their own sets of peripheral nodes; hubs are connected to other hubs.
related concept: scale-free network (Barabasi)
examples: Domain Name System (DNS), airline transportation routes, municipal governments

distributed network–a flat mesh in which there is no distinction between hubs and peripheral nodes (Baran)
related concept: rhizome (Deleuze and Guattari)
related concept: random network, a network having a random distribution of links (Barabasi).
examples: TCP/IP, peer-to-peer, national highway systems (( Galloway, Alexander. “Re: <nettime> Network, Swarm, Microstructure.� (2006). 18 April 2007 <http://www.nettime.org/Lists-Archives/nettime-l-0604/msg00060.html>. ))

The distributed network is organized, regulated and controlled by protocol. The blogosphere is a distributed network because each blog is an autonomous agent though it resembles a decentralized network because of the existence of central hubs. I would consider A-list blogs hubs that are linked to by many nodes. Is the blogosphere a mixture of the two networks? Still wondering about this. The blogosphere is a distributed network because all blogs act as autonomous agents. Even though there are popular blog that are linked to more often and we could consider them hubs in that way, going from one blog to the next does not require passing through the popular node.

As Esther kindly reminded me control can exist within the distributed network as the combination of DNS and TCP/IP supplies us with a seemingly anarchistic but highly controlled Internet. That there is no distinction between hubs and nodes does not mean that some nodes can be more popular or influential. So while in theory there might be cognitively central and cognitively peripheral blogs in the blogosphere the “syndication, aggregation and dispersionâ€? (Arie Altena) that defines the third phase of blogging we are currently in, it is highly unlikely that the information from the cognitively peripheral blog won’t spread. Blogs are cognitively central in general because of the various techniques that disperse the information.