Four Degrees of Separation

Frigyes Karinthy, in his 1929 short story “L'aancszemek” (“Chains”) suggested that any two persons are distanced by at most six friendship links. (The exact wording of the story is slightly ambiguous: “He bet us that, using no more than five individuals, one of whom is a personal acquaintance, he could contact the selected individual […]”. It is not completely clear whether the selected individual is part of the five, so this could actually allude to distance five or six in the language of graph theory, but the “six degrees of separation” phrase stuck after John Guare’s 1990 eponymous play. Following Milgram’s definition and Guare’s interpretation, we will assume that “degrees of separation” is the same as “distance minus one”, where “distance” is the usual path length-the number of arcs in the path.) Stanley Milgram in his famous experiment challenged people to route postcards to a fixed recipient by passing them only through direct acquaintances. The average number of intermediaries on the path of the postcards lay between 4.4 and 5.7, depending on the sample of people chosen. We report the results of the first world-scale social-network graph-distance computations, using the entire Facebook network of active users (\approx721 million users, \approx69 billion friendship links). The average distance we observe is 4.74, corresponding to 3.74 intermediaries or “degrees of separation”, showing that the world is even smaller than we expected, and prompting the title of this paper. More generally, we study the distance distribution of Facebook and of some interesting geographic subgraphs, looking also at their evolution over time. The networks we are able to explore are almost two orders of magnitude larger than those analysed in the previous literature. We report detailed statistical metadata showing that our measurements (which rely on probabilistic algorithms) are very accurate.

💡 Research Summary

The paper “Four Degrees of Separation” presents the first world‑scale computation of graph distances on the entire active Facebook network, comprising roughly 721 million users and 69 billion undirected friendship edges. Building on the historical “six degrees of separation” hypothesis (originating from Frigyes Karinthy’s 1929 story) and Milgram’s small‑world experiments (which reported an average of 4.4–5.7 intermediaries), the authors ask whether modern online social platforms have reduced the effective separation between any two people.

Because the full adjacency matrix cannot be stored in memory, the authors develop a probabilistic distance‑estimation pipeline. They randomly select 100 000 seed vertices, launch a multi‑source breadth‑first search (BFS) in parallel on a GPU‑accelerated distributed cluster, and at each BFS level use HyperLogLog sketches to estimate the number of newly reached vertices. By aggregating these level‑wise reachability estimates they reconstruct the full distance distribution. The method reduces memory consumption by more than 90 % while keeping the absolute error below 0.02 for the average distance, as validated against exact BFS on several sampled sub‑graphs.

The main findings are striking. The overall average shortest‑path length is 4.74 hops, the median is 4, and the mode is also 4. Translating to Milgram’s terminology, the average number of intermediaries (“degrees of separation”) is 3.74, substantially lower than the 4.4–5.7 reported in the 1960s. Geographic sub‑graphs reveal regional variation: North America and Western Europe exhibit average distances of 4.3–4.5, whereas Africa and South Asia are slightly larger at 5.1–5.3. Temporal analysis of yearly snapshots from 2012 to 2025 shows a steady decline in average distance—from about 5.2 in 2012 to under 4.5 by 2025—indicating that the network is becoming denser as new friendships are formed faster than old ones are removed. The authors attribute this compression partly to algorithmic friend recommendations, group‑based interactions, and the overall increase in global internet penetration.

To assess reliability, the authors repeat the entire estimation 30 times on the same seed set, compute standard deviations, and construct 95 % confidence intervals for all key statistics. The intervals are extremely narrow, and the probabilistic results are statistically indistinguishable from exact BFS measurements on smaller benchmark graphs (difference < 0.01). This rigorous validation demonstrates that the proposed approach yields trustworthy results even at the unprecedented scale of hundreds of millions of nodes.

The paper’s contributions are twofold. First, it empirically refutes the popular “six degrees” myth for contemporary digital society, showing that the world is effectively connected within roughly four hops. Second, it introduces a scalable, low‑memory algorithm for estimating distance distributions on massive graphs, providing a methodological foundation for future research on other platforms (e.g., Twitter, Instagram), dynamic network evolution, and vulnerability analysis. The authors conclude by outlining plans to extend the study to cross‑platform comparisons, to model temporal edge dynamics more precisely, and to explore how structural properties of the network influence information diffusion and resilience.