Networks — social graphs, the internet's router map, neural wiring, power grids — are built by very different generative processes, and the process leaves a fingerprint on the shape of the graph. This simulation renders three classic models in 3D so you can see and compare their structure directly, then attack each one to see how it fails.
p. Degrees cluster tightly around an average; there are no real hubs.Scale-free networks are famously robust to random node loss but fragile to targeted attacks on their hubs — remove the top few hubs and the whole network can shatter into disconnected islands, while random graphs degrade far more gracefully but never had the same efficient long-range shortcuts to begin with.
A live 3D graph you can rebuild as a random, small-world or scale-free network, then attack node by node to see how each topology's shape decides whether it fragments or holds together.
Random, small-world and scale-free graphs are generated by different rules and end up with very different degree distributions — flat, mostly-local-with-shortcuts, and hub-dominated respectively — which directly determines how each one fails under attack.
Pick a model, adjust node count and connectivity, then repeatedly click "Remove node" under random or targeted failure and watch the largest connected component shrink at very different rates.
Scale-free networks like the internet's autonomous-system graph can survive massive random outages almost unharmed, yet a handful of targeted hub failures can fracture them completely — a resilience trade-off first formalised by Albert, Jeong & Barabási in 2000.