The Forrest for the Trees
  - Posted by Ross Mayfield at 3:17 AM

A Red Herring article on social networking by Jonathan Thaw captures the healthy skepticism of recent capital formation. What’s new about it is lots of threatened incumbents claiming its a feature they can tack on to their existing offerings. Backed up by an analyst who spends her time using Portals suggesting it should be a portlet, because the only time you would use it is when you are prowling for dates.

I’m poking fun, but its called for. Its good incumbents are paying lip service, but when an established player has to claim featuredom the threat is real. But nevermind the vcs, network effects, patents or growing like gangbusters.

The problem for incumbents is market muscle isn’t organic. If a portal offers a social networking service they will have a flood of profiles with minimal connection. Like a grove of saplings with little to share except resource consumption. If an enterprise player imposes a graph from the top-down, incentive conflicts arise. Food chains are more complex than mandating the five food groups. Oh, and we can all envison the ideal user experience, but disruptions don’t turn out that way.

UPDATE: An anonymous coward who hasn’t been paying attention wants me to delve into the detail of the last paragraph. Its all about how the graphs grow.

If an incumbent has a base of customers and a large budget for development and promotion, it doesn’t necessarily give them a sustainable competitive advantage. Challengers have grown their network beginning with a single node inviting his friends. And they tell two friends, and they tell two friends, and so on, and so on…generation by generation. By requiring people to be invited into the network you assure they are connected to the main graph and receive more immediate value. Anyway, the advantage of this organic growth is a densely connected structure throughout the life of the graph. The disadvantage is growth as measured by the number of users is hampered.

If an incumbent offers a social networking service to their existing customer base, the network begins with multiple isolated nodes that branch out. Take a browse through Tickle and you will see lots of profiles with few of them listing connections. This is akin to a searchable resume database, little in the way of social context and few opportunities to leverage the graph for services (like filtering messages by degrees of separation). You could offer mechanisms for people to meet strangers in other graphs with forums and the like, but these connections would be enabled without an intermediary friend risking their social capital to foster the relationship, resulting in less aggregate trust, decreasing the potential utility of the network. As I mentioned in my last post on merging networks, creating incentives for people to bridge networks is hard to do. Its easier if you have two graphs (e.g. via an acquisition), a little more difficult if initial generations are based on buddy lists and extremely difficult the service is made indiscriminately available — resulting in almost as many graphs as there are nodes.

Enterprise networks face the same challenges but also have an inherent incentive conflict. If the graph is grown company by company, it means graphs contributed to help selling make you a target for selling or could be used by a competitor, which is why contribution is often made selective. This incents gaming at the company level, contribute invalid data for erstwhile competitors while using your real graph for search. Gaming can be taken into account, but the gaming game is a lot like fighting spam, both provider and users incentives grow with the value of the network alongside the complexity of tactics. Meanwhile company by company growth lessens network effects and viral growth rates will not be achieved.

The other incentive conflict is between individuals and enterprises. If an enterprise forces a salesperson to contribute their data, its something they perceive as their own asset they lease to each company the work for. If the graph is developed through hybrid contribution of enterprise and individual, providing a service for individuals as well, individuals receive less utility. Their data is being packaged as part of an enterprise sale and they cannot afford the same level of functionality as an enterprise (e.g. selective contribution of data). Without individual control over their graph or at least opt-out and with profiling of third parties without their consent raises major privacy concerns. For enterprises, the intelligence social network analysis will become a must-have and will outweigh their privacy concerns, but individuals may revolt if incentive conflicts aren’t addressed.

For most readers, I believe I said all this in a single paragraph and hope the elaboration is worthwhile. Since I didn’t use an illustrative metaphor for my last point on how services are competitive, I won’t get into detail, have a day job to get back to. I am not a journalist, don’t have an editor, but this is the beauty of the social editing process of weblogs…a conversation ensues to build understanding. Of course, its hard to have a conversation with a person that doesn’t exist.