AfriAI Field Desk
SimulatorX Lab
An exploratory interactive surface for thinking through assumptions, evidence, and system behaviour.
Use boundary
SimulatorX is an educational scenario tool. It is not a clinical, public-health, policy, forecasting, or operational decision system, and it does not represent institutional adoption or validated outcomes. Every case in Inside the Model is synthetic.
Inside the Model
Two ways of predicting the same thing, a Bayesian network and a small neural network, watched from the inside. The question is never which one is smarter. It is what each one calculates, what it assumes, and what evidence would justify trusting it.
Predict · reveal · explain
Resample the cases. Which blackout rate can move?
Every dot represents one synthetic case. The arrangement is illustrative, not a physical map.
4,096 synthetic mini-grid cases, 37.2% of which are expected to end in a blackout.
Every dot is one simulated mini-grid on one day. Each view follows these sampled cases. Resampling draws a fresh sample; Drift changes their simulated conditions. Pick a view, or let it play.
The generator's exact long-run blackout rate across all 32 sensor patterns.
Counted: 1,549 of 4,096. 95% interval 36.3%–39.3%. Resampling changes the sample, not the generator.
Even a perfect model calls this share of cases wrong at a 50% threshold, because the world is random.
Last reviewed 2026-09-29 · Quarterly review cadence · New simulator concepts enter the public site only after a defined data, method, and disclosure review.