Ross Mayfield's Weblog

Wednesday, February 05, 2003

Whole System Maps

The Blogmap project continues with new maps of the December and January data of Blog Network Friendships within Ryze. Last time we revealed the structure of the network core and how it evolved over the course of one month. This time we reveal the context of the network core in the Friendships of Blog Network members -- "whole system" maps.

December 2002:

- 64/90 Tribe members had links
- 358 internal links amongst the 64
- 1444 external links to 973 other Ryze members

January 2003:

- 151/180 Tribe members had links
- 583 internal links amongst the 151
- 2567 external links to 1693 other Ryze members

Both maps reveal the Blog Network to be open. Some network structures are relatively closed, a sign of an insular community. Ron Burt, a leading network expert, explains that a tightly closed network "amplifies predispositions, creating a structural arthritis in which people cannot learn what they do not already know"[PDF...][Orgnet].

For example, Valdis Krebs recently used the "book buddy" data that a major online book retailer generates through collaborative filtering. This strikingly revealed two closed networks that could be categorized as liberal and conservative in a bow tie pattern.

Social Network Analysis, amongst other things, looks for patterns of weakness such as the bow tie above or network holes. A bow tie pattern in an organization could spell its death. But there are also strengths in a closed network. If a group has dense internal ties (short path lengths and high clustering), information flows more directly within the core to foster agility. Purely open networks, by contrast, have more external ties. The open outreach of these networks is optimized for seeking opportunity, information and knowledge. The "golden ratio" of internal:external ties for network structure for an organization remains unknown.

11:12:53 AM

comment []

Modelling the Web

Just came across an interesting 'Power Law' paper, published by a team at NEC, which offers some thought-provoking data: "NEC researchers discovered that the degree of "rich get richer" or "winners take all" behavior varies in different categories and may be significantly less than previously thought." The key is competitiveness: in very competitive scenarios (NEC looked at ecommerce sites) 'preferential attachment' resulted in distributions that were very close to power law. But, in less competitive environments, the distributions moved steadily away from power law. In fact, deviation from power-law distribution becomes an index for competitiveness. I wonder what the Weblog index looks like? The team also pointed out that 'preferential attachment' did not prevent the rapid rise of a new star (they cite Google)... [Chris Gulker]

This paper makes a bold claim: "The model accurately accounts for the true connectivitydistributions of category-specific web pages, the web as a whole,and other socialnetworks."

Now its known that link structure drives traffic and traffic in e-commerce drives revenues. The chart to the right (right most point being Amazon) is a clear example of a Power-law in effect. By this example, the business of the net is in the hands of the few.

Aside from a fit and viral node's unlikely emergence, this is the world we live in. Or is it...

In "less competitive" markets, such as the one to the left for photographers, the distribution is less skewed. NEC Researchers offer geographic concentration as a possible reason for this decentralization.

But I believe the structure of the net is changing because the cost of sustained entry has fallen. Blogging, Web Services and other decentralized models allow new nodes to develop linkages at a lower cost. And these links aren't just driving traffic, they are conversations -- Micro-content forming micro-markets.

As the Net evolves, the definition of a link becomes more refined, the notion of Local blurs and the emergent pattern may trend towards more shared opportunity.

12:06:57 AM

comment []