Wednesday, October 28, 2009

The Math of Twitter's Fail

  Last year, News Corp. offered to buy Twitter for $500 million dollars, but was rebuffed. I think News Corp. dodged a bullet.

  I cannot figure out a way for Twitter to overcome its inherent flaws. It suffers from mathematically provable problems that make it a candidate to be the next

  A school of anthropological research concerns itself with efficient sizes of human groups. The most coherent groups, in which all members know the inter-relationships of all other members, are theoretically limited by the Dunbar Number, which equals about 150. History shows again and again that social units like agrarian villages and military units optimize near this level but not beyond it, at which point bureaucratic control impairs coherence.

  Twitter has two math problems. The first is that the incentive to gain followers pushes most players to try an gain audience share. As audience size increases, the intimacy level decreases. We have a problem of dilution. Because Twitter users do not interconnect as a unit, a user can go well beyond the Dunbar number, but ultimately the same limiting principles kick in.

  The second is that as each Twitter member receives more and more tweets, more go unread and unacknowledged. Twitter lost its usefulness to get a question answered about a year ago, and frivolity has replaced functionality. This is a problem of magnification. If everyone expands their tweet reach by a factor of ten, we are all overwhelmed by the number received. As institutions are advised by PR gurus to join Twitter and broadcast tweets, the number of mouths will exceed the number of ears. It's shouting into the darkness.

  Some Twitter members have been invited to use a new feature, called Lists, to parse people into subgroups for purposes of prioritizing. This is a red flag. It suggests that Twitter's brain trust perceives the math undermining the model. It also smells of desperation: they don't know how to blunt the math problem, so they have subdivided it into numerous math problems, or more precisely, a sociopolitical math tangle.

  Next comes the return of the anthropologists, who will observe that lists are more likely to be sorted by status than any other category. The result: hard feelings, cliques, and finding yourself on lists with titles like "kooks" or "people who talk about food too much."

  Efficiencies in social networking are illusory unless you can monopolize them. The din of everyone chatting away to hollow empires of followers is likely to yield to networks that offer precision and value over network size.