Voter Model Simulator

What you just saw

Each agent holds a binary opinion and occasionally copies a neighbor. Without zealots, the system always collapses to one side or the other — pure drift. A tiny fraction of zealots (~5%) completely changes the picture: stable, frozen clusters of both opinions coexist forever. Media influence adds a global nudge that can flip which side wins.

The science behind it

The voter model is one of the cleanest opinion-dynamics abstractions. Clifford and Sudbury introduced the version here in 1973. Extensions with zealots, media, and heterogeneous network topology are used today to study information cascades on social media platforms and electoral outcomes.

Try these experiments

1. 1. Pure drift

   Settings: zealot_fraction=0

   What to look for: One side wins — but which side is random. Consensus is inevitable.

2. 2. Frozen polarization

   Settings: zealot_fraction=0.05

   What to look for: Opinion clusters lock in and persist. No consensus.

3. 3. Media pressure

   Settings: media_strength=0.1

   What to look for: The media-aligned opinion slowly takes over, even against local social pressure.

from scirouter import SciRouter
client = SciRouter(api_key="sk-sci-...")
result = client.simulation.run(
    model="voter",
    params={"grid_size": 1000, "zealot_fraction": 0.05,
            "media_strength": 0.01},
    steps=1000, seed=42,
)