What do software development teams have to do with Dunbar’s number and primates?

How people form social organizations has an impact on our industry: the question of how big a team or company should be often serves as a guide Dunbar number of 150. It is intended to indicate what group size people usually get along with. Unfortunately, this presentation is completely wrong. Dunbar’s scientific publications says something entirely different.

Advertisement

Eberhard Wolff is Head of Architecture at SWAGLab and has been working as an architect and consultant for over twenty years, often at the interface between business and technology. He is the author of numerous articles and books, including Microservices, and is a regular speaker at international conferences. His technical focus is on modern architecture and development approaches such as the cloud, domain-driven design, and microservices.

software-architectur.tv: Company policy for software architects

He studied non-human primates, often colloquially but incorrectly referred to as “monkeys.” The ratio of the volume of their neocortex to the rest of the brain is related to the size of the group that these primates form. From this data, Dunbar estimates the size of the human group at 147.8. As is usual in science, the value has a spread. With 95% probability the value is between 100.2 and 231.1. A Other papers gives a completely different range for the number, with a 95% confidence interval of 3.8 to 292.0. The paper presents several other data analyses, but all of them have confidence intervals ranging from low single digits to a few hundred. A look at this paper shows that this number has no practical use.

But Dunbar is not so concerned about numbers: In groups, primates use mutual grooming, among other things, to remove parasites but also to strengthen social cohesion. Dunbar establishes a relationship between the time spent on grooming and the size of the primate group. Larger groups require more grooming. According to Dunbar, this would lead to a time commitment that cannot be imagined for people. According to Dunbar, humans therefore developed language in order to build social cohesion more efficiently. Language did not originate, as other scientists claim, to coordinate hunting or tool making, but to maintain social relationships.

In other words, Dunbar’s central thesis is not group size, but that human language evolved to strengthen social ties.

Dunbar’s number of 150 is the estimated maximum size of groups of people who use language as a process rather than a form of community grooming. He talks about clans/villages. He also names other groups: bands of 30 to 50 people and tribes of 1,000 to 2,000 people.

Accordingly, Dunbar’s paper discusses many examples of human groups of very different sizes. Dunbar says that groups with sizes significantly different from 150 belong to a different category. So his thesis is definitely not on the size of groups, but on the mechanisms by which groups maintain themselves – and he writes it that way himself.

Particularly interesting: Dunbar sees groups in the military as confirmation of his thesis, since it is said that there are groups of 100 to 200 people. But of course there are also much smaller groups in the military, such as a squad (2 to 8 soldiers in the Bundeswehr) or larger groups such as a battalion (300 to 1,200 soldiers), which he does not consider further.

The paper has been widely criticised by other scientists, so almost every part of the paper is controversial. For example, there is evidence of fission/fusion behaviour among primates with regard to group size. These are groups that individuals join and then leave. For example, individuals may sleep together in the same location but spend the day apart. Such groups are therefore only temporary. However, species with such behaviour require little time to form together and are sometimes quite large. Apparently, primates can form large groups even without complex human language.

Why is the Dunbar number so interesting?

So you can’t learn anything from Dunbar’s number for the organization of human teams. Dunbar himself says that there can be human groups of practically any size. You don’t even need to resort to widespread criticism to make this statement.

There are also other interesting points in the reviews. For example, it is not at all clear why the number of people who know a person affects the size of a group in any way. So if we only know a certain number of people and talk to them regularly, a group can still be quite large. It is enough if we act together in a coordinated way. Thanks to language, people can coordinate on a large scale, up to nations or even beyond. It should be clear to everyone that people know and trust each other in different ways. This is also used in everyday project life. You ask a third person to give you information instead of giving it directly to one person because the relationship of trust between these two people is better.

Cause of misinterpretation

To me, the misinterpretation of Dunbar’s number is indicative of a fundamental problem: complex human and social behaviour is being simplified. In the end there is a number for the ideal group size. It is a simple rule to follow.

Actually, intuition should tell everyone something different. Because everyone knows from their lives that people can function in different groups – in the private and professional context: company, club, neighborhood, friends, family. The groups are of different sizes. Especially for larger groups, there are hierarchies, like in the army, but there are also hierarchies in companies with teams, departments, locations, etc.

These groups often do not exist for long. For example, during a training session or at the first counseling appointment, trainers and counselors have to work with a group they have never seen before – and it works. This is of course a different group, with a different type of relationship from your own family, but the goals of such a group are also different.

Now one could argue that, among other things, trust only grows over time. But trust can also be built quickly: if a patient is admitted to hospital, in extreme cases they will trust the treating doctor even without having ever seen that person before.

Dunbar’s research can certainly be used as inspiration for thinking about mechanisms to strengthen group cohesion. Dunbar’s thesis is that language emerged only to strengthen social cohesion, and he presents statistics on how much time is spent talking about social relationships and gossiping. Measures to strengthen cohesion, for example through casual conversation, could be useful. Where is such a platform for a team? It does not have to be forced team bonding, but regular lunches together could serve such a function well.

tl;dr

The Dunbar number says nothing about the likely size of teams or companies. They can be of any size and structured in different ways. Misinterpreting the number shows that our industry is prone to counter-intuitive simplifications. Teams need a mechanism to create social cohesion.

(RME)