Skip to content

Scenario Probability Table

Analytic artifact — instantiates Catastrophic-Risk Bargaining De-escalation

A lightweight table of how things could go — each scenario with a likelihood band, consequence, key assumption, and the action threshold that would trigger a response — for when a full model is overkill.

Sometimes a crisis cell needs a shared picture of how a standoff could unfold fast — in an afternoon, legible to everyone in the room — and a full probabilistic model would be too slow, too data-hungry, or false precision dressed up as rigor. A Scenario Probability Table is that quick artifact: a structured table whose rows are plausible scenarios and whose columns are likelihood band, consequence, the key assumption each rests on, and the action threshold that would trigger a response. Its defining trade is speed and legibility over rigor. It deliberately uses coarse bands ("likely / possible / unlikely") rather than computed probabilities, so a mixed group can build and read it together and immediately agree on which futures matter — the fast, communicable cousin of a full Probabilistic Safety Analysis, not a replacement for it.

Example

Two nations' coast guards are shadowing each other in contested waters, and a crisis cell needs everyone — diplomats, operators, lawyers — on the same page within hours. They build a scenario probability table. Rows: routine shadowing (likely / low consequence / assumes disciplined crews); accidental collision (possible / high / assumes crowded sea-lane and fatigue); miscommunication-driven escalation (unlikely / severe / assumes a garbled signal read as hostile). Each row carries an action threshold — e.g., "if closing distance drops below X, open the clarification line."

The table's value is that in one page, a group with very different expertise now shares the same map of what could happen and what each of them will do at which trigger. No one waited for a modeling team; the disagreements that remain are about the entries, which is exactly the productive argument to be having.

How it works

  • Enumerate the scenarios that matter — cover the routine, the accidental, and the escalatory paths, deliberately including the boring-but-likely and the rare-but-severe.
  • Band, don't compute — assign coarse likelihood and consequence bands rather than precise numbers, keeping the artifact fast to build and honest about its own resolution.
  • Expose the load-bearing assumption — record what each scenario's rating depends on, so a shift in conditions flags which rows to revisit.
  • Attach an action threshold per row — pair each scenario with the observable trigger and pre-agreed response, turning the table from description into a decision aid.

Tuning parameters

  • Scenario granularity — a handful of broad futures or many fine ones; more rows capture nuance but erode the at-a-glance clarity that is the table's whole point.
  • Band scheme — how many likelihood/consequence levels and how they're defined; coarser bands are faster and less falsely precise but risk lumping very different risks together.
  • Assumption visibility — how prominently each row's key assumption is recorded; surfacing them enables fast updating, hiding them lets stale ratings persist.
  • Threshold sharpness — how observable and specific each action trigger is; vague thresholds ("if things get worse") defeat the purpose, over-tight ones misfire on noise.

When it helps, and when it misleads

Its strength is shared, fast legibility: it gives a diverse crisis team a common map and a common set of triggers in the time a full model would still be scoping, which is often worth more in a live standoff than a more accurate analysis delivered too late or understood by too few.

Its danger is the flip side of its coarseness. Compressing risk into a few bands can genuinely mis-rank scenarios — a well-documented failure of risk matrices, where two very different risks land in the same cell and a lower risk can even be scored above a higher one.[1] Neat rows also invite false confidence and the illusion of completeness: the future not on the table reads as a future that can't happen. And like any such artifact it can be back-filled to justify a preferred response. The discipline: treat it as a fast shared sketch, not a verdict; keep a row for "something we haven't listed"; escalate to a real model when a decision turns on a rating the bands can't resolve; and revisit entries as assumptions move.

How it implements the components

  • stochastic_catastrophe_risk_state — the likelihood and consequence bands per row are a lightweight, legible statement of the current catastrophe-risk state.
  • escalation_pathway_and_risk_ladder — the scenarios, laid from routine to severe, sketch the rungs of the escalation ladder without a formal model.

It renders risk quickly and coarsely; it does NOT define the catastrophe boundary or model event chains rigorously (Probabilistic Safety Analysis), nor adversarially test the list for the futures it omitted (Red-Team Verification Review).

References

[1] Ranking risks by coarse likelihood-by-consequence cells is the familiar risk matrix, and its documented weakness — that low-resolution categories can mis-order risks and even rate a smaller risk above a larger one — is a recognized critique of the method. It is a reason to escalate to a quantitative model when a decision hinges on the ordering. ↩