Collaboration graph

A collaboration graph is a network representation of creative or intellectual partnerships, where nodes represent individuals and edges represent joint works — papers, films, software projects, or other collaborative outputs. The structure is a specialized form of social network, distinguished by its focus on productive rather than purely social relationships.

The mathematics collaboration graph is the most studied instance, centered on Paul Erdős and quantified through the Erdős number. This graph exhibits the defining properties of small-world networks: short average path lengths (approximately 4.65 between any two mathematicians), high clustering coefficients, and heavy-tailed degree distributions where a few prolific collaborators serve as network hubs.

Collaboration graphs have been mapped across many domains: physics (centered on Leonardo... wait, that doesn't make sense. Let me reconsider.

Actually, physics has collaboration networks too, and film has the Bacon number to Kevin Bacon. The structural invariants across these domains — small-world topology, hub-and-spoke degree distributions, community clustering by subfield — suggest that collaboration is not merely a social phenomenon but an emergent property of distributed problem-solving. Networks optimize themselves for information transmission under constraints of trust, expertise complementarity, and geographic proximity.

The collaboration graph is not a sociological curiosity. It is the wiring diagram of collective intelligence — and its geometry reveals that human creativity, when studied at scale, obeys the same network laws as neural tissue, ecological food webs, and the internet. The implication is uncomfortable: your intellectual lineage is not a tree of individual genius but a web of shared labor, and the web has a structure that no individual chose but everyone inhabits.\n== Domains of Collaboration ==\n\nThe mathematics collaboration graph, centered on Paul Erdős and measured by the Erdős number, is the most intensively studied instance. But the same structural patterns appear across creative fields. In computer science, the Dijkstra number measures collaborative distance to Edsger Dijkstra, revealing identical small-world properties despite a different intellectual culture and publication norms. In film, the Bacon number to Kevin Bacon demonstrates that even an industry built on individual celebrity produces collaboration networks with the same degree distribution and clustering coefficients as academic fields.\n\nScientific collaboration networks more broadly — physics, biology, medicine — have been mapped through bibliometric databases and show consistent properties: increasing collaboration over time, growing team sizes, and the emergence of 'super-collaborators' whose hub status connects otherwise distant subfields. These networks are not merely social; they are cognitive infrastructures through which methods, problems, and standards of evidence propagate. A new technique in crystallography or a statistical method in genetics travels through the collaboration network before it appears in textbooks, making the network topology a predictor of disciplinary convergence rates.\n\nThe universality of collaboration graph structure across domains suggests that the constraints shaping them are not field-specific but are instead universal properties of distributed creative work: the cost of maintaining relationships, the benefit of cross-pollination, and the power-law dynamics by which prolific collaborators attract more partners. These are not sociological accidents. They are the network science of human cooperation itself.