Security Norm Contagion: Threshold Adoption Model

Population & Network

Network size 120 members

Initial adopters 6%

Mean risk-appetite threshold τ 0.30

Threshold diversity σ 0.12

Cascade State

t = 0 — seeding…

Adopted 6%Holding out 94%

Step t

Flipped this step

How it works

This is the Granovetter–Watts binary threshold model of complex contagion, applied to the adoption of a security practice (e.g. enabling MFA, applying a patch, following a secure default) inside an organization's social network.

Each member i draws a threshold:
  θ_i ~ Normal(μ, σ), clamped to [0.02, 0.95]

Synchronous update rule, every step t → t+1:
  fᵢ(t) = (# adopting neighbors of i) / deg(i)
  i adopts at t+1  iff  fᵢ(t) ≥ θᵢ

Adoption never reverses (a security norm, once
internalised, is treated as sticky).

Mean threshold μ — the population's average risk appetite. Low μ = people need only a small nudge from peers to switch; high μ = strong social proof is required before anyone changes behaviour.
Diversity σ — spread of thresholds across the population. Some diversity is required for a cascade to start at all: a population of identical high-θ holdouts never tips, because no single early adopter can supply enough local proof.
Initial adopters — the seed set, placed uniformly at random on the network. In practice this models deliberately seeding influential early adopters (security champions) rather than mandating a policy top-down.
Edges are k-nearest-neighbour links on a sphere plus a handful of random long-range "weak ties" (Watts–Strogatz shortcuts) — these shortcuts are usually what lets a stalled local cluster reach the rest of the network.

Reading the sim: a run that saturates near 100% shows a tipping point was crossed; a run that stalls partway shows the remaining holdouts' thresholds exceed what their local neighbourhood can supply — the model most security-awareness programmes actually face.

Adopted the practice Holding out

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.