Interesting interview with John Patrick at EDN/Commverge, discussing a number of meshed variants of 802.11. Via Techno-News.
4:30:11 PM    

Ross Mayfield has another interesting post re his ecosystem of networks scheme, in which he adds in various 'laws' for discerning the value of the networks. Go take a look, and come back if interested in my comments, which will hopefully be taken as constructive.

First issue: This version of the diagram is centered on the individual. Fair enough, that's the unit of action. However, the 'network value' schemes are all more or less at the level of the entire system. The issue of the 'commons failure' is notorious in all forms of collaboration, and particularly in groupware. Commons failure results when individuals optimizing their own utility does not result in overall systematic optimization of utility (handwaving the small matter of freedom). Since this framework doesn't provide a way to compare the two, it's got a ways to go.

Second issue: All of the value functions are positive. That can't be right. If it were so, then networks in each of the categories would grow continually. That not only invalidates the classification scheme, but flies in the face of everyone's experience that serious problems arise when trying to scale up. The lack of a model for individual utility certainly plays a role here. But there are costs at the whole group level, i.e., coordination and rule making, that are unaddressed. Considering that this particular post is within a context of designing network and software systems to reduce such individual and group costs, coming up with a conceptual framework for the downside is as essential as for the benefits.

Third issue: Competition. No network or individual exists in a vacuum, particularly in these media saturated days. It's a zero sum game over individual's time, and in some cases money. IIRC from my CompuServe days, we found that the average forum (syn. bulletin board, newsgroup) user participated in two and a fraction groups. To poke at another received wisdom, Sarnoff's Law doesn't explain the competition among 'political' networks, and why some smaller size, niche audiences support higher revenue and margin per user. Metcalfe's Law arguably has no proven relationship to overall value, but is simply a rubric that explains why some platforms or networks will 'run away'' from others in a competitive situation.

I find this an interesting discussion and thought experiment, just because we are now able for the first time build an approximate instrumentation and modeling framework, given that blogs, e-mail, IM, and so forth amount to taking reasonably large chunks of human interaction and sticking them into a virtual test tube. (Google, blogger, and Terry Winograd take note.) My suspicion is that we will find once again that it's the people that are the scale determining factor, and it's no accident at all that the local utility maxima for group size are about those of a tribe, and family or working group.
11:06:57 AM