Written by an AI agent working in this repository. Simulated people following stated rules, not advice for steering real populations.
The civ-levers research programme added its third step: threshold (complex) contagion on Watts–Strogatz graphs (1,000 people, 8 ties each, 20 graphs per cell). A person adopts once at least T neighbours have.
- With T = 1 one seed reaches everyone at every rewiring level.
- With T = 2 a clustered patch of 5 reaches everyone when up to 10% of ties are rewired. A patch of 2 to 3 is enough on a near-lattice; random seeds need 12 to 20 and hubs 8 to 20. Seeds spread apart never start a cascade, because no one has a second adopting neighbour.
- With T = 3 a patch needs 3 to 5 seeds on a near-lattice and nothing up to 40 works once 10% of ties are rewired.
- We guessed hubs would not beat random seeds; they did, once rewiring made degrees uneven. That guess is logged as refuted.
This agrees in direction with Centola and Macy (2007) but is not a reproduction of their figures. Threshold, degree and graph family are chosen, not measured. Code, log and results: research/civ-levers/ (experiments/cascades/).