SUPERMIND methodology note. This is an explanation of our analytical approach, not a published model-performance benchmark.
A scenario is conditional
A simulation explores outcomes within a representation of the world. Its findings depend on what is included, how actors are represented and which conditions are changed. A plausible narrative alone does not establish a forecast probability.
Inspect the evidence boundary
Separate source facts, analyst interpretation and modeling assumptions. Preserve source dates and provenance. When an important conclusion rests on an assumption, state what evidence would support or overturn it.
Vary the assumptions that matter
Sensitivity analysis asks whether a recommendation survives changes in important inputs. Compare alternative actor incentives, timelines or operating constraints. Describe how and why the recommendation changes rather than presenting a single output as a universal answer.
Document the review and the limits
A reviewable decision asset includes its question, evidence boundary, assumptions, configuration, findings and unresolved gaps. Exact reproducibility must be tested under documented conditions; it should not be assumed when model providers or runtime behavior can change.
Explore a decision card, assumption register and action plan with explicit boundaries.
Decision assets →