The Limits of Dunbar’s Number
Posted by Jeffrey Henning on Mon, Feb 15, 2010
If I ask you to imagine two monkeys, chances are you will picture one monkey grooming the other, picking twigs out of his companion's fur. Grooming, for primates, is a way of strengthening social bonds. Primatologists assert that the number of members of a social group that a primate can keep track of is limited by the volume of their neocortex. For Indri lemurs, it's a group size of 4. For ring-tailed lemurs, it's 15. For Rhesus monkeys, it's 25. And for humans...
Dunbar analyzed social group size and neocortex volume for 38 genera of primates and predicted a human "mean group size" of 148, which -- since it was an estimation -- is typically rounded to 150. He then searched anthropological literature and found many instances of mean group sizes of 150: Neolithic farming villages, Hutterite settlements and Roman army units.
Malcolm Gladwell popularized this apparent limit to natural community size in The Tipping Point, and -- as a result -- all sorts of misapplications of the data have sprung up:
- You can't have more than 150 friends (according to the otherwise inerrant Seth Godin).
- Social groups shouldn't exceed 150 people in size.
- You can't analyze more than 150 survey responses. Er, I mean, you can't build a true online community of more than 150 people.
Interest in Dunbar's number as it might relate to online communities surged again last week. Research re-ran its 2007 article "The size of the matter", and PluggedInCo expanded on a CNet article on Dunbar and Facebook, and Twitter users retweeted both.
To be fair, Dunbar never intended for his work to be used in these ways. He was asserting that small, tight-knit groups that relied on one another for survival had a natural limit of around 150 members, because the human brain could keep track of only 150*149 interrelationships: as Stowe Boyd puts it, "Dunbar's Number represents the largest stable social group, 150 people more or less, where all the members not only know each other, but understand how each member is related to the others, and the nature of their social interactions." (See Seth Godin Misunderstands Dunbar's Number, And Stubs His Toe. Also check out Forget Dunbar's Number: Our Future is in Scoble's Number.)
Many organizational planners think the ideal number of employees in an office, the ideal number of soldiers in an army unit, the ideal number of members in a church, is the Dunbar number. This is all nonsense. Unfortunately, humans have only one neocortex, so you can't devote one to the office, one to the church, one to school and one to that MROC you just joined.
For some MROCs, you want as many people as possible:
- Thousands or tens of thousands of participants is what you want for brainstorming and sheer volume of idea generation.
- You want as many customers participating as possible when you need to segment them by specific demographics.
- Jacob Morgan, in "Why Dunbar's Number is Irrelevant", writes "the real value of collaboration and of networks doesn't come from strong relationships and networks but from weak ones. In fact, one of Morten's network rules is actually ‘build weak ties, not strong ones.' According to Morten: ‘research shows that weak ties can prove much more helpful ... because they form bridges to worlds we do not walk within. Strong ties, on the other hand, tend to be worlds we already know."
But there are many valid and appropriate reasons to limit the size of some market-research online communities:
- Smaller groups foster a true sense of community closeness.
- Research into sensitive health, financial and other personal matters is richer and deeper in a small close-knit community.
- The researcher wants to keep track of fewer people.
- There is less material for the qualitative researcher to wade through and analyze (large communities can produce an embarrassment of riches).
- There are fewer people who might squeal about the confidential research you are conducting.
Again, there are many valid and appropriate reasons to limit the size of your MROC. Dunbar's number just isn't one of them.