# What is the optimal degree of separation in a society?

Almost everyone has heard about the notion of “six degrees of separation” – that there are six jumps between any two persons in, say, the US. The experiment or game actually originates in a short story by Hungarian author Frigyes Karinthy, The Chain, where the object of the game was to connect two random individuals. It was then popularized by experiments where social scientist Stanley Milgram tried to send a package to a person in Boston and started with a random set of people who then only could send it on via someone they knew. The result – popular culture has it – was around 6 degrees of separation between any two people in the US (the actual experiment had a much more difficult outcome, and was not as clearcut). The hypothetical rule was also sometimes formulated as a logarithmic correlation between separation and size of the population.

Several different studies have been made of networks, determining their degrees and discussing what is sometimes referred to as the “small world”-phenomenon. Duncan Watt’s reimagined Milgram’s experiment with email and arrived at an average number of people an email had to go through on the Internet was 6. One of the most interesting experiments was the “six degrees of Kevin Bacon”, looking at how to connect any actor with Kevin Bacon.

The result – an actors Bacon number – describes how many jumps it takes for an actor to connect with Kevin Bacon. The average Bacon number in the network is around 3 (and Kevin Bacon is not the most connected – there are others that are closer to having an average actor-number of 2).

All in all, different networks of different sizes seemed to conform to the idea that group size and degrees of separation are heavily connected. Families are super connected, villages are close to professional networks and as we move up through cities to nations and the world we see greater and greater degrees of separation. In a simple picture:

Now, this is in itself interesting, and you could start asking questions about this relationship. Why is it that we see this correlation? What is it based on? Here we enter open speculation, to be clear, but let’s play with explanations. Maybe one explanations is that we are limited to how many people we can have social relationships with? This would lead us down the path to the Dunbar number, proposed by Robin Dunbar as the maximum number of connections we can have given our biological, neurological capacity (Dunbar arrived at it by comparing brain size in primates and group size). The Dunbar number is a round 150 people, and if we imagine that the world fragments naturally (in some biological sense) into groups of 150, then maybe this could explain the degrees of separation growing in the way they do.

But if we invoke the Dunbar number (admittedly controversial) we are suggesting that there is some kind of causation between biology and the degrees of separation and brain size and our inherent ability to correlate a group. The degrees of separation partly start to resemble a collective cognitive limit or a social organization limit. Let’s draw this out in all of its consequences to see if there is a there, there: societies would, under this line of speculation, have an optimal degree of separation and that is what we have detected – the six degrees.

Ok, so why is it optimal? What is it that is optimized by the six degrees of separation? And how is that connected with the Dunbar number? Here is where I want to go with this – I think you could argue that Dunbar has given us more than a limit of the group size of primates, the group size is actually the size of the most efficient evolutionary unit for any specific primate. If larger groups and more friends and a higher Dunbar number had been evolutionarily advantageous we should have seen it in evidence — group sizes in fish and birds are different and not related in the same way to brain size, so why is brain size the limiting factor for primates? Because – wait for it – we think together. Intelligence is a network concept. We think together and so the size of our groups are optimized for collective problem solving and thinking.

Then, to keep that as we scale up in group size, the degrees of separation reflect the fact that we are still keeping to those Dunbar-sized groups to continue thinking efficiently together.

We are deep into guesswork here, but it is interesting to just pursue this thought experiment a little further, I think. We can now ask what happens if we change the degrees of separation across different groups. Dunbar groups have a single degree of separation, and as they grow and connect to other Dunbar groups we see the logarithmic growth of separation with group size. What happens in our thought experiment if we reduce the degrees of separation in really large groups? What is it that changes – do we become better thinkers or do we do worse on cognitive tasks?

Here I think the answer is – although maybe controversial – that we become much worse thinkers if group size increases and degree of separation is held constant. We simply cannot process or think with that many people, or brain then moves from cognitive modes to ways of trying to reassemble the smaller groups around us. A hyper connected large network component is unable to think, and its collective intelligence falls fast because it is constantly overloaded by ideas, views, information and data.

Our thought experiment might even allow us to say that hyperconnected large network components, several orders of magnitude beyond the Dunbar groups, are unable to create and sustain knowledge.

Let us now turn back to today. How is technology changing things? One of the cliché observations often through-out there is that technology connects people. What this should mean is that technology is slowly reducing the degrees of separation, and yes – this is exactly what we find. Facebook researchers have shown that the Facebook network component – which is very large – is increasingly connected, and the degrees of separation have gone down from 4.74 to 4.57. It is 4.67 at Twitter (see here). In our picture:

So, what then happens to a social group when we reduce the degrees of separation? The group seeks its former stability and in doing so it reverts from what Julia Galef calls “scout mind” – a knowledge seeking position – to soldier mind, where we seek to reform the group through having the same views, in an attempt to regain the stability of the Dunbar group.

If our thought experiment is correct, we would end up with the insight that connecting people is what drives them apart, what really polarizes them, as they try to regain their ability to think together. Polarization is really an attempt at increasing degrees of separation.

Now, there are several weaknesses in this argument, and it is interesting to look at those too.

First, we could argue that the overall degree of separation in a network component does not reflect the fact that there can be many sub groups and that these can uphold the Dunbar requirements – so why should the average degree of separation concern us? I think this is fair, and just saying “our society is more connected, so polarization is the natural reaction to increase separation” may be far to simplistic – but that does not mean that we could not see that effect on top of several still working, smaller groups. Society as a whole may suffer and smaller groups still remain healthy, and in fact, the tension then between society and these groups would only increase polarization – but this time not along political as much as social dimensions.

Second, we could deny that there is any value in the Dunbar findings at all and suggest that technology actually augments our capacity to interact with more people, and so we should expect that the solution to our problem is if we can go back to village cognition globally for the entire human network. The way forward is shrinking the degrees of separation not trying to increase them, and what we need are new modes of thought. We need to re-learn thinking in groups and start thinking in hyper connected networks. This is a skill, not an evolutionary trait, and we should be optimistic about our ability to do this. I find this hypothesis alluring, but also slightly utopian – I am not sure if it is right, but think it is worth taking seriously. My worry is that we are constantly underestimating the biological limitations of technology use, and that this forgetfulness of our biological nature leads us into solutions that increase the mismatch between the Wilsonian layers – “our paleontological feelings, medieval institutions and godlike technology.”

Third, we could argue that polarization is really just the consequence of inequality and that inequality is the result of political choices or the lack of such choices. Talking about degrees of separation and Dunbar groups really just hides the ball. Our emergency, you could argue, is political – not biological, and not scientific, and not dependent on any abstruse network qualities. This view has a lot to recommend it, and I am not saying that there is not a political problem as well — but maybe the ability of our political system to solve problems has to do with the cognitive ability we collectively command? If so, it seems it could be worthwhile asking if we are better or worse at thinking together now.

Fourth, you could argue this is a pessimistic view of mankind, technology and the future – and that things have improved massively with ever lower degrees of separation. I would disagree with this — I actually think that if we can start to unravel institutional and biological mechanisms of social problem solving we will be better off, and if that means that we increase the degrees of separation that will just lead to better decision, and I am quite optimistic that we can do this. There is another thing here too. If we look at our little chart, we find that greater degrees of separation actually also can be loosely related to greater political freedom, or at least liberty. The city has more political freedom than the village, and the family has none. It seems as if the tighter the social cohesion, the less individual freedom. This leads us to the surprising, but interesting idea that individual freedom requires more degrees of separation. And technology can be used for both connecting people and increasing the degrees of separation — and I would predict that some of the technologies that we will see in the coming decades may actually help us do exactly that.

To be clear, I am not blaming social media at all for connecting people – I think we want to be connected, and that two evolutionary imperatives (belonging to a group and retaining social distance for thinking) are at conflict here. The solution is not less social media, but more innovation along the lines of smaller groups and greater degrees of separation. Room size limits modeled on the Dunbar number, network component max sizes, sub-networks — it can all be done really well and lead to a coming re-separation rather than disconnection between people.

This is important – it is not about disconnecting people, it is about increasing the degrees of separation in the connections themselves.

Well, just a thought experiment. But an interesting one, I think.