About this observatory
This visualization is inspired by research on how the structural position of biased agents within scientific communities affects their capacity to manufacture ignorance. The model follows Bala–Goyal (1998) and Holman–Bruner (2015): agents run experiments, share results with neighbors, and update beliefs via Bayesian inference. Biased agents never update and always report misleading evidence.
The central finding: a single biased agent at a high-centrality hub causes dramatically more damage than one at the periphery (Cohen’s d ≈ 1.42). Degree centrality — how many connections a node has — is the strongest predictor of an agnotologist’s influence.
The aquarium metaphor
Like watching fish in a tank, you observe a society of epistemic agents from above. Green and red particles visualize testimony flowing along network edges — making the invisible process of social learning visible.
Interactions
Hover any node for details. Drag nodes or draw a selection box to rearrange. The Laboratory lets you build custom networks and experiment with every parameter.
Key parameters
Topology — who talks to whom. Stars concentrate influence; cycles spread it evenly.
Bias placement — hub vs. periphery is the central comparison.
Bias strength — deception intensity (0.5 = honest, 1.0 = maximal).
Efficacy difference (δ) — how detectable truth is. Lower = harder.
Evidence per round — trials per tick. More = faster convergence.