Dunbar's Numbar
Wednesday, May 13th, 2009 04:16 pmThere is a phenomenon in anthropology known as Dunbar's number, after its creator, Robin Dunbar, who speculated that there was a correlation between neocortex size and the maximum number of regular social contacts that a primate could maintain. Since social contact in primates is maintained primarily by grooming, I would have thought the crucial variable would be manual dexterity rather than neocortex size, but then I'm not a primatologist, so what do I know? Anyway, the idea caught on like lice on an ungroomed primate, and Dunbar proposed a number for humans based on the data from other primates. This number is 150, which sounds like a reasonable estimate. I mean, could you handle more than 150 friends on Live Journal, Facebook or whatever? (This, by the way, is the reason why so many IT movers and shakers are interested in Dunbar). There again, could you handle more than 100? There's the problem: Dunbar's number is actually 148 with a 95% confidence interval of 100 to 230. So we can predict with a fair degree of confidence that if a group grows to have a hundred members, it will either start to experience problems cohering and start to fragment, or continue growing up to as much as double its current size.
In other words, Dunbar's number tells us nothing that common sense doesn't. Meanwhile, other anthropologists have come up with some different numbers: Russell Bernard and Peter Killworth reckon the maximum could be a hefty 230 or 290 (depending on whether you take the median or the mean). But, as Wikipedia notes, "the Bernard-Killworth number has not been popularized as widely as Dunbar's," despite its being replicated in a variety of studies. To explain this, I propose Dunbar's Law: "Where there are two hypotheses to explain the same data, the one with the cooler name will be adopted." "Dunbar's number" beats "the Bernard-Killworth number" by sheer assonance.
In other words, Dunbar's number tells us nothing that common sense doesn't. Meanwhile, other anthropologists have come up with some different numbers: Russell Bernard and Peter Killworth reckon the maximum could be a hefty 230 or 290 (depending on whether you take the median or the mean). But, as Wikipedia notes, "the Bernard-Killworth number has not been popularized as widely as Dunbar's," despite its being replicated in a variety of studies. To explain this, I propose Dunbar's Law: "Where there are two hypotheses to explain the same data, the one with the cooler name will be adopted." "Dunbar's number" beats "the Bernard-Killworth number" by sheer assonance.