How we measure the agents.
Every metric we use to judge the swarm — what it means in plain English, the value we're getting from our 100 agents right now, whether it's actually the right thing to track, and (if you want it) the formula and the paper behind it. Pick your depth below.
loading live values…
Live values pulled from the same snapshot the dashboard uses (api.hridayjain.in/wepredict_ops.json); N_eff is computed live every cycle from the belief ledger; the shadow-backfill calibration scores are from the latest off-line run. Formulas and parameters match the running engine (score_beliefs.js, build_wepredict_snapshot.js, belief.js, diversity_meter.js, fit_recalibration.js). Citations individually verified. Reviewed June 2026.