Sunday, May 28, 2006

Self Organizing: How the Internet Avoids the Babel Objection

I came across Yochai Benkler's book The Wealth of Networks a month ago and highlighted it in a post titled Self-exemplification. I called it that because the book seemed to be exemplifying its topic with a Creative Commons License, a wiki and a free PDF download.

I just started reading the book today. The section from the introduction re-printed below does a great job of giving some insight into why the internet, the blogosphere, etc. do not suffer from the Babel objection (which the uninitiated often immidiately cite as why this all can't work).

Benkler defines the the Babel objection:

According to the Babel objection, when everyone can speak, no one can be heard, and we devolve either to a cacophony or to the reemergence of money as the distinguishing factor between statements that are heard and those that wallow in obscurity.

Here is the passage:



The networked public sphere has also begun to respond to the information overload problem, but without re-creating the power of mass media at the points of filtering and accreditation. There are two core elements to these developments: First, we are beginning to see the emergence of nonmarket, peer-produced alternative sources of filtration and accreditation in place of the market-based alternatives. Relevance and accreditation are themselves information goods, just like software or an encyclopedia.

What we are seeing on the network is that filtering for both relevance and accreditation has become the object of widespread practices of mutual pointing, of peer review, of pointing to original sources of claims, and its complement, the social practice that those who have some ability to evaluate the claims in fact do comment on them.

The second element is a contingent but empirically confirmed observation of how users actually use the network. As a descriptive matter, information flow in the network is much more ordered than a simple random walk in the cacophony of information flow would suggest, and significantly less centralized than the mass media environment was.

Some sites are much more visible and widely read than others. This is true both when one looks at the Web as a whole, and when one looks at smaller clusters of similar sites or users who tend to cluster.

Most commentators who have looked at this pattern have interpreted it as a reemergence of mass media—the dominance of the few visible sites. But a full consideration of the various elements of the network topology literature supports a very different interpretation, in which order emerges in the networked environment without re-creating the failures of the mass-media-dominated public sphere.

Sites cluster around communities of interest: Australian fire brigades tend to link to other Australian fire brigades, conservative political blogs (Web logs or online journals) in the United States to other conservative political blogs in the United States, and to a lesser but still significant extent, to liberal political blogs.

In each of these clusters, the pattern of some high visibility nodes continues, but as the clusters become small enough, many more of the sites are moderately linked to each other in the cluster. Through this pattern, the network seems to be forming into an attention backbone.

“Local” clusters—communities of interest—can provide initial vetting and “peer-review-like” qualities to individual contributions made within an interest cluster. Observations that are seen as significant within a community of interest make their way to the relatively visible sites in that cluster, from where they become visible to people in larger (“regional”) clusters. This continues until an observation makes its way to the “superstar” sites that hundreds of thousands of people might read and use. This path is complemented by the practice of relatively easy commenting and posting directly to many of the superstar sites, which creates shortcuts to wide attention. [I would add that these superstar sites (and possibly others along the way) seem increasingly to be having an impact on the mass media as well.]

It is fairly simple to grasp intuitively why these patterns might emerge. Users tend to treat other people’s choices about what to link to and to read as good indicators of what is worthwhile for them. They are not slavish in this, though; they apply some judgment of their own as to whether certain types of users—say, political junkies of a particular stripe, or fans of a specific television program—are the best predictors of what will be interesting for them. The result is that attention in the networked environment is more dependent on being interesting to an engaged group of people than it is in the mass-media environment, where moderate interest to large numbers of weakly engaged viewers is preferable.

Because of the redundancy of clusters and links, and because many clusters are based on mutual interest, not on capital investment, it is more difficult to buy attention on the Internet than it is in mass media outlets, and harder still to use money to squelch an opposing view. These characteristics save the networked environment from the Babel objection without reintroducing excessive power in any single party or small cluster of them, and without causing a resurgence in the role of money as a precondition to the ability to speak publicly.

No comments: