Blogmap: Friendship/Time The January sample of Blogmap.net, the first project to map and analyze the social network of an online community over time, is here -- revealing signs of creeping decentralization.
The maps below are of the Friendships of core of the Blog Network as measured by confirmed relationships within Ryze. Pete Kaminski helped with the data collection. Here is Valids Krebs' analysis:
|
In the 30+ days between December 2002 and January 2003, the Ryze Blog Tribe doubled in size, going from 90 members to 180. How did this change the structure and the centralization of the group?
Initially, we look at the internal structure of the group. We display a link between two Tribe members if, and only if, they both list each other as a "Ryze Friend". On Ryze, as in other on-line communities, friendship links are created with great ease, often someone lists another as a friend before they have even exchanged their first email! To minimize this explosion of 'possible' links, we look at only those ties where both sides have confirmed each other as a 'friend'. Even then, we feel that we have quite a few 'false positives'. But that is the data we have, so we work with it.
Interestingly, the 'core'[all confirmed ties between Tribe members] has not changed much over the period of quick growth. Although the Tribe as a whole grew by 100%, the core only grew by 50%. In December, the founder of the group -- Ross Mayfield, was the main hub in the group. In January he retains that position. All network metrics point to Ross as the leader of the group. Yet, the Tribe is not totally dependent on Ross. A network that is very dependent on one node usually fragments when that node is removed -- best case the network does not fragment, but the average path length between nodes rises noticeably. Neither of these happened when Ross' node was removed [ Sorry Ross, you are not as important as you first appear! ;-) ]. When Ross' node was removed the average path length of either network increased by less than 10% in both cases -- not a significant disruption of the network. The Blog Tribe would go on if Ross picked up his laptop and left. With new members coming in, the average network neighborhood expanded from December to January going from 22.8 to 28.2. Yet, the core, as a percentage of the total network, shrank from 69% to 53%. These signs of 'creeping decentralization' often appear in networks started by nodes who mostly know each other, but then are expanded by nodes known to only a few of the existing members. |
 |
The map to the left is the core of the Blog Network in December, just two weeks after its formation. Previously we mapped the Friend links (Friends on Ryze) and Blog links (Blogrolls) of network members. This captured the entirety of the Network and revealed that almost the entire tribe was two horizons (or degrees) from any point in the network -- the power of weak ties.
Another set of maps revealed the centrality of the tribe at its inception and located Network Members within their network of relationships. |
|
The map to the right is the core of the Blog Network in January. Again, these are confirmed ties with a greater probability of a strong tie. Increasing the requirement for what is a relationship, such as basic confirmation, may change the properties of the network.
Zooming in or focusing on a cluster also simply makes it qualitatively useful. From just looking at the maps you can see growth as a moving picture, identify new hubs and relationships that could be formed to complement existing ones. |
 |
Within Ryze, a Friendship can mean many things to many people. Confirmed relationships point to some level of communication at the minimum. But at the least, its an explicit declaration. Two people saying they know each other to the world.
In Blogspace, the emergence of Friend Of A Friend (FOAF) XML vocabulary faces a similar challenge in defining what a relationship generically means. Eric Snowdeal points out the etiquitte of relationship declaration, or lack thereof, may be the primary barrier to FOAF reaching critical mass. This is an important point, mostly for our expectations.
FOAF will be widely adopted by the lowest common denominator of use. For example, Mark' Pilgrim's policy of any friend of FOAF is a friend of mine (email me and I'll relate to you). Bulletin Boards for the masses will not be widely used. It will primarily be a tool of organized communities and smaller groups. The rules of 7, 12 & 150 apply and social incentives lead to smaller group forming. Categories of relationships could help, but FOAF will start with smart groups with stronger ties (confirmed) when a purpose is provided. And tight formation of groups within secure environments avoids the slippery slope into the Omnihell of ubiquitous intrusiveness. Smaller groups can form their own etiquitte.
Next week we will broaden the scope and look at the external links of the Blog Network.
7:37:58 PM 
|
|
