Use calibrated AI forecasting models to predict realistic outcomes and implementation metrics for ballot propositions (for example, how many homes a housing bond will actually produce and at what real cost). Voters and civic groups can then convert campaign claims into expected per‑unit costs and fiscal impacts rather than relying on sponsor projections alone.
— If adopted, this practice would raise the factual standard of ballot debates and change how voters evaluate campaign promises and bond measures.
Scott Alexander
2026.10.28
100% relevant
Scott Alexander cites an 'AI superforecaster' that predicted ~22,000 homes built from an $11.25B bond, which he used to compute an implied $500K per home figure and to critique the proposition's design.
← Back to all ideas