Every single person alive on this planet is reachable from you in six handshakes or fewer, and a team of mathematicians just proved that number is not a coincidence.
Six. Not five, not seven, not some sliding scale that shifts as civilisations grow or shrink. Six. Locked in. Fixed. A number that holds whether the network is 1,000 people or 8 billion, and that has held across every serious test ever run on human social networks at scale. Findings published in Physical Review X in May 2023 quantify exactly why this number is six and not anything else, tracing it to a mechanism so embedded in ordinary human behaviour that it has been quietly building the same invisible architecture into every society ever studied.
Start with what this actually means in your daily life, because the number is weirder than it sounds. You know roughly 150 to 600 people well enough to ask a favour. One of those people knows someone in a village in rural Indonesia. That Indonesian contact knows a government minister. That minister knows a tech billionaire in Taipei. That billionaire knows a subsistence farmer in the Sahel. You and the farmer are five handshakes apart, which puts you both inside the six-degree limit with a step to spare. That chain works in both directions. It works for you and the farmer, and it works for the farmer and the Taiwanese billionaire, and it works for the minister and the person who delivers your post. Every pair, anywhere on earth, six steps. The number does not budge. This is not a metaphor or a loose estimate. It is a structural property of the network, and it has been measured directly.
The first serious attempt to test this idea came from Stanley Milgram in 1967, working with roughly 1,000 people in the United States. Milgram gave participants a target person they did not know and asked them to forward a letter through personal acquaintances toward that target, one connection at a time. The chains that completed averaged around six steps. Decades later, a study tracked 24,163 separate chains across 13 countries using email, with 18 different targets, and recorded an average of six steps per completed chain. A 2007 dataset drawn from 30 billion conversations among 240 million Microsoft Messenger users produced an average path length of six. Facebook’s network in 2011, containing 721 million users and 69 billion friendship links, logged an average node distance of 4.74. Twitter measured 3.435 steps between random users. Every platform, every methodology, every era: the number converges on six or drifts slightly below, but never blows out. Something is holding it there.
What is holding it there is not the size of the network. This is the part that should stop you cold. If human networks behaved like simple mathematical graphs, where distance scales predictably with population, a network of 8 billion people should have dramatically longer chains than one of 1 million. It does not. The diameter of a human social network, meaning the longest shortest path between any two nodes in the entire system, flatlines around six regardless of how many people you add. Mathematically this is described as an ultrasmall-world property: a state where the network’s diameter is independent of its size across multiple orders of magnitude. The standard logarithmic scaling you see in other small-world networks does not apply here. Human social networks break the expected pattern in a specific and reproducible way, and until this paper, nobody had a proven explanation for why.
The answer comes from something you do every single day without thinking about it: deciding whether a new connection is worth the effort. Every person in a social network is constantly running a quiet calculation. Making and keeping a friendship costs something, time, attention, energy, social capital. Gaining a new connection also delivers something, information, opportunity, status, access to resources you did not have before. The research team modelled this as a game played across an entire network, where each person independently adds or drops connections based on whether the benefit outweighs the cost. The specific benefit they modelled was betweenness centrality, which is a measure of how often you sit on the shortest path between two other people. In plain terms, the more you act as a bridge between separate groups, the more information flows through you, and the more influence you have. People want this, consciously or not, and they build connections to get it.
When you run this game forward in time, something unexpected happens at the network level. Individual people making purely selfish connection decisions, each one looking only at their own immediate social neighbourhood, collectively produce a global network with a fixed diameter of six. No coordination required. No central plan. No awareness of the larger structure. The network self-organises into the six-degree state the same way water finds its own level, as a consequence of each molecule doing what physics requires it to do. The mathematicians proved this is not an accident of initial conditions or a quirk of specific network sizes. Any network where individuals balance connection costs against centrality benefits, starting from virtually any initial arrangement, reaches a stable equilibrium with a diameter of exactly six.
The mechanism driving this involves a concept called 2-independence, which is simpler than it sounds. Imagine three people who are all strangers to each other and who are also not friends of any of the same people, meaning no shared contacts, no common ground within two degrees of connection. In network terms they form what is called a 2-independent set, a cluster of isolated points with no bridging links anywhere near them. Now imagine one of those three people realises they could connect to the other two, bridging the gap. The moment they do, they land on the shortest path between a large number of previously distant pairs. The benefit is enormous and immediate. It far exceeds the cost of maintaining two extra connections, so they make the move. This process repeats across the entire network, with highly connected individuals acting as bridges, pulling isolated clusters together. The mathematics show that this bridging behaviour, driven entirely by self-interest, inevitably produces a central node reachable from any other node in three steps or fewer. Once that central node exists, the maximum distance between any two people in the entire network drops to six: three steps from person A to the central node, three steps from the central node to person B.
The scale-freeness of real social networks makes this outcome essentially guaranteed. Real networks follow a pattern where a small number of people have enormous numbers of connections and the vast majority have relatively few, described by a power law. The most connected nodes in a network of any realistic size already satisfy the mathematical condition required for the bridging game to take hold. This means the process that locks in six degrees is not some special feature of ideally structured networks. It is already present in the messy, unequal, real architecture of human society, and it kicks in automatically as networks grow. The simulations used to test this ran 10,000 different network configurations at each scale tested, and the equilibrium diameter held at six across all of them.
What makes this stranger still is that the six-degree outcome survives the addition of local clustering and hierarchical structure, the features that make real social networks feel like they are made of tight-knit groups nested inside larger communities. When the game runs on networks already containing dense friend clusters and hub-and-spoke hierarchies, those structures survive intact. The equilibrium state preserves the clustering, preserves the hierarchy, preserves the power-law distribution of connections, and still produces a diameter of six. The six-degree property is not imposed on top of the network’s other features. It coexists with them, emerging from the same underlying game without erasing anything else.
There is also a direct connection here to something sociologists have known since the 1970s: the strength of weak ties. Research established that the most common way people find new jobs is not through close friends but through distant acquaintances, people you barely know who move in entirely different social circles. Those distant contacts provide access to information that your close network cannot supply, precisely because they sit in parts of the social graph that your immediate connections never reach. The 2-independent bridging behaviour identified in this research is mathematically the same process Granovetter described sociologically. Every time someone extends a connection across a social gap, they are simultaneously searching for centrality and weaving a weak tie. The game and the sociology are the same thing, operating at different scales of description.
The network currently tracked by Facebook records an average separation of 4.74 degrees across 721 million users, a number that has been remarkably stable across multiple measurements taken years apart. Twitter’s network sits at 3.435 degrees. Neither platform has broken the six-degree ceiling despite adding hundreds of millions of users, which is exactly what the mathematics predicts. Six is not the limit the network is straining against. It is the equilibrium the network is always already at.
Source:
Samoylenko, I., Aleja, D., Primo, E., et al. “Why Are There Six Degrees of Separation in a Social Network?” Physical Review X, Vol. 13, 021032. Published 31 May 2023. American Physical Society. DOI: 10.1103/PhysRevX.13.021032
