Where the number comes from
Robin Dunbar's 1992 paper used the correlation between neocortex ratio and mean group size in primates, then extrapolated the human point. The result was roughly 150, and it appeared to match company sizes, military unit structures, Christmas-card lists and the ethnographic record of village sizes.
Dunbar later described layers: about 5 intimate ties, 15 close, 50, 150, 500 and 1,500, each a band of decreasing emotional closeness. Mobile-phone and social-network data seemed to show the same bands in call frequencies.
The supporting argument is about time and grooming. Primates maintain relationships by grooming, which is expensive; language may have evolved as cheap grooming at distance, allowing larger groups. The ceiling is a budget, not a hard wall.
The reanalysis and the internet
A 2021 reanalysis by Lindenfors and colleagues, published in Biology Letters, redid the comparative statistics with modern methods and reported a 95 per cent confidence interval from about 2 to about 520. The point estimate remained near 150, but the interval is so wide that it cannot bear the weight usually placed on it.
Part of the problem is that the primate sample is small, the group-size measures are inconsistent, and the extrapolation assumes humans sit on the same regression line as other primates — an assumption that language and institutions directly challenge.
Online, the test is unavoidable. Average Facebook friend counts run in the hundreds, but the number of contacts people actually interact with regularly stays near the predicted layers. Dunbar's own 2016 analysis of 30 million users found the structure persisted — a smaller active network nested inside a larger passive one.
Survives: human relationships are layered by closeness and limited by maintenance cost; there is a small intimate circle and a wider active one, and that structure shows up in phones, firms and villages.
Does not survive: the precision of 150 as a cognitive limit. The confidence interval does not support a specific number, and the primate comparative base is thin.
Wrong framing: 'Dunbar's number says you cannot have more than 150 friends' is a misreading. It is about the number of relationships whose quality can be maintained, not the number of names you know.
The archive labels this live debate because the layered structure is well supported and the famous number is much weaker than its fame implies.
Why it is in the archive
Because it is one of the most cited numbers in popular science and one of the least certain, and because the online version of the question is genuinely open.
Related: echo chambers, the contact hypothesis.
Sources for this page
- Wikipedia, “Dunbar’s number” — the claim, its uses and criticismsen.wikipedia.org
- Dunbar (1992), Journal of Human Evolution — neocortex size and group size in primatesdoi.org
- Lindenfors et al. (2021), Biology Letters — “Dunbar’s number is fuzzy but most humans can maintain 150 relationships”doi.org
- Dunbar et al. (2016), Royal Society Open Science — online social networks and the 150 layerdoi.org