— INSIGHT / SCENARIO CONSTRUCTION
← Back to Probability Lab

Better forecasts begin with better scenario sets.

One of the most common failure modes in probabilistic reasoning is false diversity. A forecast looks broad because it contains many scenarios, but several of them are cosmetic variations of the same basic story. Aurelon puts strong pressure on causal diversity before aggregation even begins.

— Causal family diagram

What the generator is designed to do

Raw generationMany plausible scenariosbut some are duplicatesDiversity filterreject renamed copiesforce distinct mechanismsban meta/adjudication scenariosDistinct scenario familiesstructural persistenceexternal shockinternal political shiftpartial / ambiguous caseimplementation failure
More scenarios do not create more insight unless those scenarios are genuinely different.
— Practical effect

Why this matters for low-count runs

At low scenario counts, elegant narratives can dominate too easily. Broader causal coverage makes even smaller runs less likely to be seduced by one polished story.

— Why we use this approach

Core operating rules

First, scenario families must be causally distinct. Second, cosmetic duplicates are discouraged before they ever reach aggregation. Third, adjudication logic is not allowed to enter the main scenario pool as a fake real-world pathway.