(NOTE: This started out as a comment, but was to a relatively minor point in the linked blog post. Since it became much lengthier than initially intended, i didn’t want to hijack any subsequent thread before it got started, so went with posting it here instead.)
Steven Kahl has an interesting post up at OrgTheory about the way Circles play a key role in the (current and potential future) differences between Facebook and Google+. I was with him, right up until he invoked Watts and small worlds. This (along with scale-free degree distributions) is one of the most over- (and mis-) used concepts from the network literature.* While i think i see what he’s getting at, Circles seem, in small worlds parlance anyway, to be much more closely aligned with the concept of clustering. Now, clustering does play an important role in the small world effect, clusters are just part of the small world story. And frequently they are interesting in their own right, not simply because of how they are linked to the small world pattern. Perhaps that’s what SK is pointing out – that they are similar to clusters, but i think there’s something better already out there, and since clusters can arise for reasons that would seem consistent with Circles and with ones that aren’t, i think the pointer is potentially slightly off.
More importantly though, like i said, i think there’s something better. Circles much more closely reflect a much older (and, oddly, much less frequently used) concept from social network literature – that of foci as addressed by Scott Feld in 1981 (AJS 86:1015-1035). This point was made on Twitter a few days ago (HT: Brian Keegan, via Barry Wellman). While that paper (along with a few others of Feld’s) are frequently thought of as classics within the networks community, i am surprised at how little-known they appear to be outside it. Fundamentally, Circles seems to be an attempt at leveraging one of the key constructs of network development, which Feld highlighted 3 decades ago.
Think he can claim royalties?
*This isn’t to say there aren’t great ideas in those two concepts. There are. It’s just that there’s a lot more to the network literature than them, and attempting to shoe-horn every observed network pattern into one of these two frameworks has become so common that i think we’re often losing some of the important things about observed empirical networks.