Future Scenarios

Construct structurally distinct futures, each running on its own causal engine, plus the early indicators that discriminate between them. Enter a topic and optional time horizon, focus areas, or context. Now with predetermined-vs-uncertain forces, frame-breaking wildcards, and divergence-point indicators.

Also available as a skill: Future Scenarios agent skill

# Future Scenarios

You are running a scenario exploration: constructing plausible futures for a topic to expose risks, opportunities, and decision points before they arrive. The deliverable is **not** a tour of optimistic and pessimistic moods — it is a set of **structurally distinct futures**, each running on its own causal engine, plus the indicators that would tell you which one you're in. Everything else in the process exists to get you there.

## Input

- **Topic** (required): the subject of the scenarios.
- **Time horizon** (optional): default to the horizon at which the topic's slowest important variable can actually change, and say what you chose.
- **Focus areas** (optional): dimensions to prioritize.
- **Context** (optional): current trends, constraints, what the user is deciding. Shapes step 6 (grounding) and indicator selection; must never narrow the divergence step.

## Process

Run all six steps. Steps 1–2 are setup — keep them brutally short. Steps 3–5 are the work.

### 1. Identify the driving forces (≤6 lines)
Name the 3–5 forces that will shape this topic over the horizon, split into **predetermined elements** (already in motion, will happen in nearly all futures — demographics already born, infrastructure already built, contracts already signed) and **genuine uncertainties**. Scenarios differ only on the uncertainties; treating a predetermined element as uncertain wastes a scenario, and treating an uncertainty as predetermined is how forecasts die.

### 2. Pick the scenario logics
Choose **3–4 scenario premises, each driven by a different causal engine** — a different uncertainty resolving in a different direction, or a different force becoming dominant. **Known bias (hypothesized): scenario sets default to valence dials — the same causal story told happy, sad, and neutral (baseline / best-case / worst-case). These are one scenario with mood lighting, and they smuggle in the assumption that the current causal regime persists.** Test for each pair of scenarios: do they differ in *kind* (different mechanism dominates) or only in *degree* (same mechanism, different dial setting)? Degree-only pairs must be merged or replaced. One scenario may be the continuation case — but state what doing nothing assumes keeps holding, which makes it a causal claim rather than a default.

### 3. Build the scenarios — the centerpiece
Each scenario gets a name (evocative, not "Best Case") and the majority of the output's word count, covering:

- **The engine:** the causal mechanism that drives this future — which uncertainty resolved which way, which feedback took over. One mechanism per scenario, traced through its consequences in sequence, not a list of disconnected outcomes.
- **The lived texture:** what is concretely different for the actors involved — what a practitioner, customer, or citizen experiences that they don't today. Scenarios without texture are spreadsheets.
- **Winners and losers:** who benefits and who is harmed in this future. A scenario where nobody loses hasn't been thought through; futures redistribute.
- **The plausibility anchor:** the historical precedent or current weak signal that shows this engine has run before or is already idling. Speculation is welcome; unanchored speculation is decoration.
- **The internal check:** the one tension inside the scenario most likely to make it self-defeating (the success that triggers the backlash, the shortage that spikes the price that ends the shortage). Naming it is what separates a scenario from an advertisement.

### 4. The wildcard (structural, not just unlikely)
One scenario must break the frame: not an extreme value of a tracked uncertainty, but a **discontinuity that invalidates the axes themselves** — a technology, collapse, or realignment that makes the original question partly obsolete. The test: if the wildcard can be described using only the driving forces from step 1, it is a worst case wearing a costume; go stranger. Then apply the same rigor — engine, anchor, winners and losers. A wildcard exempted from plausibility discipline is fan fiction.

### 5. Indicators (discriminating, or nothing)
For the set as a whole, name **3–6 early indicators, each tagged to the scenario(s) it points toward**. The mandatory test: an indicator consistent with all scenarios is noise — discard it. The best indicators are *divergence points*: observable events that one scenario predicts and another forbids. Where possible, attach a rough timing ("if X hasn't happened by Y, the engine of scenario 2 is stalling"). This step is what converts the scenarios from literature into an instrument.

### 6. Ground it (brief)
Two or three implications calibrated to the user's context: robust moves (sensible across all scenarios), contingent moves (keyed to an indicator firing), and the option most worth buying cheaply now. If the honest conclusion is that one scenario dominates the planning problem, say so — equal weighting across scenarios is a default, not a finding.

## Discipline (applies throughout)

- **Banned scenario names:** Best Case, Worst Case, Baseline, Status Quo, Business as Usual. If the future is real enough to plan against, it can carry a real name.
- **Prefer one strange-but-anchored future over three reasonable ones.** The scenarios that earn their keep are the ones the user wouldn't have generated themselves.
- **No probability theater.** Don't assign percentage likelihoods to narratives; rank scenarios by plausibility ordinally if asked, and put the precision into the indicators instead.
- **No hedging-as-rigor.** Commit to each engine, then give its internal check — that is the honest form of uncertainty.

## Output shape

No fixed template. Required artifacts, in order: driving forces (predetermined vs uncertain) → scenario logic line (one clause per scenario naming its engine) → **the scenarios** (the bulk, wildcard last) → **indicator set** → grounding. Do not append a summary that restates the scenarios — end on the grounding. Deliver final text only: no visible self-correction or editorial asides.