Scenario forecasting with calibrated confidence, alternative hypotheses, and pre-declared indicators.
// QUESTION CLASS
Anticipatory is the engine that produces estimates — structured forecasts in the shape of a National Intelligence Estimate. The output is not a prediction; it is a set of alternative futures, each assessed for likelihood, each accompanied by the specific indicators that would confirm or refute it. The point of an estimate is not to be right about the future. The point is to be useful about uncertainty: to tell a decision-maker which outcomes are plausible, how they would be distinguished, and what confidence the analysis carries. This is the engine where IC probability language matters most. "Likely," "unlikely," "roughly even chance," "almost certain" — these are not conversational terms on an Anticipatory report; they are calibrated probability ranges, defined at the top of every document. A reader who doesn't know the difference between "likely" and "very likely" will learn it from the first page.
// ANATOMY
An Anticipatory report is modeled on the National Intelligence Estimate — a product shape refined by the IC over seventy years. The Key Judgments lead; they are the one-page version the customer reads first. Each judgment carries an explicit confidence level (low / moderate / high) with calibrated probability language. Scenarios are plural by default — an estimate that produces only one future is not an estimate. Indicators are pre-declared, which means they were named as diagnostic before any reading was taken. This is the only way to avoid the post-hoc rationalization that makes most public forecasting worthless.
// IC PROBABILITY LANGUAGE, CALIBRATED
// WORKED EXAMPLE
// THE KIQ
"Project the plausible succession scenarios for [position of power] in [country] on a 24-month horizon. Identify the most-likely scenario, the most-dangerous scenario, and the warning indicators that would signal departure from the estimate."
We assess that [incumbent] will retain [position] through the 24-month horizon with MODERATE confidence (roughly 55–70% probability). This judgment rests on the institutional mechanisms favoring continuity¹ and the absence of a consolidated challenger position among the three most prominent potential successors.² A departure from this judgment would most likely follow an incapacitating health event,³ the consolidation of a challenger coalition around [named figure],⁴ or an external shock of a type and scale we do not currently see indicators for.⁵ We judge the most-dangerous scenario — a contested succession producing a significant policy reorientation toward [direction] — to have a LOW-to-MODERATE likelihood (roughly 20–35%)…
// TRADECRAFT
Every likelihood on an Anticipatory report maps to a defined probability range, published at the top of the document. "Likely" is not a vibe; it is a range (roughly 55–80%). This is the IC convention, and it is how a reader can actually do something with the estimate.
ICD 203 · Standards 2 & 8An Anticipatory report produces 3–5 scenarios and identifies the analyst's judged most-likely. It does not present only the most-likely. This is Analysis of Alternatives — ICD 203 Standard 4 — and it is what prevents the single-point failure mode that produced famous estimate disasters.
ICD 203 · Standard 4: Incorporates analysis of alternativesEvery scenario is accompanied by a set of indicators that would confirm or refute it. These indicators are named in the report itself. If observed, the estimate is re-issued with a diff. This is how you avoid the after-the-fact rationalization that hollows out forecasting credibility.
I&W Methodology · Pre-declared indicatorsOften paired with: Foundational (the baseline an estimate projects from), Argumentative (runs throughout, pressure-testing each judgment).
// AUTHORED BY
// OTHER ENGINES
