Chance by year
Country by country
Same question for each selected country on its own, on one shared scale. The number is the chance by the chosen year.
What would change the forecast
Each row is something we could observe. Choosing "Suppose this" conditions every chart on it. Across the two outcomes the forecast averages back to today's number, weighted by how likely each is.
Checkable soon
Near-term predictions the model makes. Each resolves from the source named. If these miss by more than chance allows, the assumptions behind them are wrong.
Markets and this model
Prediction markets that ask something close, next to this model's answer to the nearest question it can express. Some definitions are looser than the memo line and some stricter, and none has more than 50 traders. Two "any country" markets resolved YES in November and December 2025, when the Marshall Islands began quarterly payments to every resident citizen; the ones below are still open.
Assumption ledger
Every input, stated so that it can turn out wrong. "Resolves" says when and how: a date and a data source; for conditional assumptions, the condition that must occur first; and for yearly rates, "only in aggregate", meaning they can be scored only by pooling many governments or years. Labels say where each value comes from: market a prediction-market price, data an official record, poll or historical base rate, judgment my estimate from the evidence listed. Facts and records behind the estimates were gathered by one research agent and checked by a second. Probabilities show their mean.