Interactive · Social Network Science
Five simulations exploring how structure shapes everything — who hears what, who matters most, what spreads, and what gets stuck. Click nodes to intervene. Watch what happens.
The same 35 people, connected differently, produce radically different systems. Random networks are democratic but slow to spread information. Scale-free networks — where a few hubs have many connections — are efficient but fragile: remove a hub and the network fragments. Small-world networks combine tight local clusters with a few long-range bridges, achieving both cohesion and reach. Real social networks almost always look like this last type.
Before designing an intervention, ask: what is the topology of the network you are working in? A programme that works by reaching hubs will fail in a lattice. A programme that builds local trust clusters will be ineffective in a scale-free network if the hubs are bypassed. Structure is not neutral. It is the medium through which everything else travels.
Information, norms, rumours, disease, innovation — all travel through social networks by the same basic mechanism: contact between connected nodes. Topology determines reach. Seed a scale-free network at a hub and watch the whole system light up within seconds. Seed it at a peripheral node and it may never spread at all. This asymmetry — invisible from aggregate data — is one of the most important structural facts about social change.
The recovery slider introduces a simple SIR model — nodes can move from infected back to resistant. A high recovery rate produces short bursts that die out. A low one produces endemic spread. Notice that on a small-world network, the initial spread is fast (bridges carry it across clusters) but it can also be contained if bridges are cut early. Timing matters as much as topology.
Degree: most connections — the obvious measure, but misleading in sparse networks. Betweenness: sits on the most shortest paths between others — the broker, the gatekeeper. Closeness: shortest average path to everyone else — the best position to spread something fast. Eigenvector: connected to well-connected people — influence by proximity to power. Switch between them and watch how the "most important" node changes completely.
The question "who should we fund?" often defaults to degree centrality — who has the most connections. But in many change processes, betweenness centrality is more relevant: who sits between communities that don't otherwise talk? Cutting that node — through funding withdrawal, burnout, or relocation — doesn't just remove one actor. It severs the bridge entirely.
In 1973, Mark Granovetter showed that people find jobs more often through acquaintances than through close friends. The reason: your close friends move in the same circles you do — they know what you know. Your acquaintances connect you to worlds you don't have access to. Weak ties carry novel information. Strong ties carry trust. Both are necessary. Neither is sufficient.
Drag the bridge strength slider to zero. Notice that information from cluster A never reaches B or C — not because it's not spreading, but because there's no path for it. This is the structural isolation that many marginalised communities face: not a lack of internal cohesion, but a deficit of bridges to networks where resources, decisions, and opportunities are concentrated. The problem isn't effort. It's topology.
People don't adopt behaviours in isolation — they watch others first. Each person has a threshold: the proportion of their peers who must act before they will. A crowd riot, a norm shift, a technology adoption, a political movement — all follow this logic. Run the simulation twice with identical settings. Because the network structure interacts with the threshold distribution, seemingly similar conditions can produce complete cascade or total stall — with no obvious explanation from aggregate data alone.
Notice that lowering the average threshold doesn't guarantee a full cascade — it depends on who has a low threshold and how they are positioned in the network. A cluster of low-threshold nodes isolated from the rest does nothing. One low-threshold node sitting between clusters can unlock the whole system. This is why aggregate statistics about attitudes or readiness tell you very little about whether change will actually happen. Position is everything.
About this resource
These simulations are produced by the School of Systems and Complexity (SSC) — a practitioner school working at the intersection of systems thinking, complexity science, and organisational change. We build education, advisory practice, and community of practice programmes for those navigating environments where structure, emergence, and power intersect.
Visit systemsandcomplexity.school ↗