Scenario Ensemble¶
Method — instantiates Ensemble Decision Aggregation
Tests a candidate decision against a small set of discrete, qualitatively distinct plausible futures to see whether it holds up across all of them.
A Scenario Ensemble does not pool numbers — it pressure-tests a decision. Its defining idea, the one that separates it from every sibling, is that the members are a handful of discrete, internally-consistent, deliberately divergent futures, and the test is whether one candidate action survives all of them. There is no averaging, often no probabilities at all; there is a small set of qualitatively different worlds and a search for the choice that performs acceptably in each. The output is not an estimate but a verdict on an action: robust, or fragile. Where a simulation ensemble asks "what is the distribution of outcomes," a scenario ensemble asks "which decision won't sink us no matter which of these worlds we get."
Example¶
A regional utility must commit to a decade of grid investment now, under deep uncertainty about how the energy system will evolve. Rather than bet the plan on one central forecast, its planners build four discrete futures: rapid electrification of transport and heating, a slow stalled transition, a natural-gas-dominant world, and a distributed-rooftop-solar boom. Each is a coherent story, not a knob-twist of the same model. They then run each candidate investment plan — a big-transmission plan, a flexible-modular plan, a status-quo plan — through all four worlds.
The big-transmission plan wins the electrification world but strands billions in the stalled and solar-boom worlds. The modular plan is never the cheapest but avoids catastrophe in every future. Because no one can say which world will arrive, the planners choose the modular plan on robustness, not expected cost. The ensemble of futures changed the decision by exposing which plan had no acceptable failure mode.
How it works¶
- Frame the decision. Name the specific action set under test and what "performs acceptably" means — the criterion each future will judge against.
- Build few, divergent, coherent futures. A small set spanning genuinely different decision-relevant conditions; cosmetic variants of one baseline defeat the purpose.
- Run every candidate through every future. Hold the futures fixed and score each action in each — no probability weighting required.
- Select for robustness. Prefer the action that avoids unacceptable outcomes across all futures (minimax-regret or satisficing), not the one that is optimal in the single most likely world.[n1]
Tuning parameters¶
- Number and spread of futures — few and truly divergent beats many and similar; too many collapses into false comfort, too few misses the disruptor.
- Scenario axes — which uncertain drivers define the futures. The choice of axes decides what the exercise can and cannot see.
- Robustness criterion — minimax-regret (avoid the worst regret) versus satisficing (clear a bar everywhere) versus tolerating one bad world; this dial sets how conservative the choice is.
- Whether to weight futures — usually not; assigning probabilities quietly turns the exercise back into expected-value optimization and loses the robustness stance.
When it helps, and when it misleads¶
Its strength is guarding against the deepest failure of single-forecast planning: optimizing hard for one future that then does not arrive. When uncertainty is structural rather than statistical — you cannot honestly put a probability on the world — testing across discrete futures is how a decision earns its robustness.
Its failure mode is scenarios that are secretly the same. If all four futures miss the real disruptor, or are cosmetic variations of one worldview, the exercise manufactures false confidence in a plan that is robust only to imagined variation. It can also feed paralysis, or be run backward to justify a preferred inaction. The discipline is to stress the axes hard, include at least one genuinely adverse and one genuinely surprising future, and keep the set small and distinct.
How it implements the components¶
decision_context— names the specific action under test and the "performs acceptably" bar each future applies.diversity_criterion— requires the futures to span genuinely different, decision-relevant conditions rather than cosmetic variants.decision_rule_under_spread— its defining component: the robustness rule (minimax-regret or satisficing) that picks the action holding across all futures rather than the expected-case optimum.
It does not generate a quantitative outcome distribution from many stochastic runs — that is Simulation Ensemble, whose separating component is disagreement_measure; and it never pools its futures into one number, so it implements no aggregation_rule (the pooling siblings are Model Averaging and Diversified Forecast Pool).
Related¶
- Instantiates: Ensemble Decision Aggregation — the instance where members are discrete futures and the aggregate is a robust choice.
- Consumes: Simulation Ensemble — each future's outcomes may be quantified by a simulation run when a scenario needs numbers rather than a narrative score.
- Sibling mechanisms: Ensemble Model · Model Averaging · Simulation Ensemble · Expert Panel · Committee Scoring · Multi-Source Intelligence Synthesis · Diversified Forecast Pool
Editorial Notes¶
Form Classification¶
Form family: Analysis, Modeling & Optimization
Rationale: Scenario Ensemble operates as an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution because it tests a candidate decision against a small set of discrete, qualitatively distinct plausible futures to see whether it holds up across all of them.
Independent corroboration: The frozen evidence defines Scenario Ensemble as 'Tests a candidate decision against a small set of discrete, qualitatively distinct plausible futures to see whether it holds up across all of them', so its operative form is Analysis, Modeling & Optimization.
Nearest alternative: Assessment, Review & Assurance — Scenario Ensemble includes features of a bounded evaluation of existing evidence or work that produces a finding or disposition, but its defining operation is an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Futurism & Strategic Foresight
Origin pattern: Convergent development
Present-day reach: Multi-domain
Rationale: Testing decisions across qualitatively distinct plausible futures is canonical strategic foresight.
Related originating lineages:
- Operations Research — Operations research, optimization, and queueing analysis supplies a parallel or contributing lineage for the mechanism's defining operation: tests a candidate decision against a small set of discrete, qualitatively distinct plausible futures to see whether it holds up across all of them.
- Organizational & Management Science — Organizational design, management, and operational governance supplies a parallel or contributing lineage for the mechanism's defining operation: tests a candidate decision against a small set of discrete, qualitatively distinct plausible futures to see whether it holds up across all of them.
Review resolution: Both blind reviewers agree that futurism_foresight is the primary historical origin. Explicit reconciliation of alternate_origin_disagreement, origin_mode_disagreement starts from reviewer_a's mechanism-specific evidence: Testing decisions across qualitatively distinct plausible futures is canonical strategic foresight. Reviewer A proposed alternates=operations_research, origin_mode=convergent, domain_reach=multi_domain, and encyclopedia_synthesis=false; reviewer B proposed alternates=operations_research, organizational_management, origin_mode=single_lineage, domain_reach=multi_domain, and encyclopedia_synthesis=false. The final record retains every independently supported alternate from either review (operations_research, organizational_management) without an arbitrary cap, selects origin_mode=convergent to represent the combined lineage evidence, and records domain_reach=multi_domain and encyclopedia_synthesis=false. Present-day transfer is recorded as reach and is not treated as proof of historical origin.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Robust decision making inverts the usual predict-then-act order: instead of finding the optimal plan for a best-guess future, it searches for plans that perform acceptably across a wide range of plausible futures, using regret or satisficing criteria rather than expected value. It is the standard framing for decisions under deep (non-probabilistic) uncertainty. ↩