The unit's analytical work runs on a five-layer methodological operating system. Each Protocol governs a distinct epistemic function. They are composable: every paper flows through the full pipeline. They are versioned: each Protocol carries a public release identifier and a change log. They are partially disclosed: architectural specification is public; operational implementation is restricted under NDA.
FP3 Method is the unit's core analytical framework. It governs how probability is assigned, how scenarios are constructed, how alternative hypotheses are weighted, and how forecasts are committed to a binarisable resolution criterion. It is built on three intellectual lineages: Tetlock's calibration discipline (Good Judgment Project, 2011-present), Heuer's structured analytic techniques (CIA, 1999), and Brier's verification scoring (1950).
Reference architecture document: FP3 Method — Public Specification v1.0 →
calibration is the metacognitive control layer that sits above FP3 Method. Its function is to continuously interrogate the analytical output itself for overclaim, motivated reasoning, narrative drift, and unstated assumptions. The signature mechanism is the §17 operator: every published claim is forced to declare whether it rests on verifiable data or on inference, with confidence explicitly stated as a percentage band. calibration draws from Kahneman/Tversky's heuristics-and-biases literature, from Tetlock's "active open-mindedness" construct, and from Schwitzgebel's epistemic humility framework.
source integrity guards the data perimeter of the unit. It is structured as three operational heads (H1 · H2 · H3), each governing a distinct integrity function: cryptographic source ingestion, adversarial filtering and honey-trap detection, and restricted distribution OPSEC. Its design draws from operational security practice in intelligence services (declassified literature), from Russian Maskirovka countermeasures studies, and from contemporary OSINT verification frameworks (Bellingcat methodology, EUvsDisinfo).
Cryptographic source verification · Admiralty grading at intake · PEC autospedizione for anteriorità
Adversarial source detection · Honey-trap recognition · Influence-operation flagging · State-actor disinformation triage
Tier-A/B/C routing discipline · NDA tracking · Cross-sector contamination prevention · Audit trail
case memory is the unit's empirical memory layer. It maintains the Calibration Ledger — the persistent record of every published probabilistic claim, its declared confidence, its pre-committed resolution criterion, and its ex-post Brier verification. It also runs invariant pattern-mining across the historical paper corpus. The Ledger is the unit's accountability mechanism: published forecasts are not retracted, they are scored.
The unit's methodology rests on a documented intellectual lineage. We name our sources because hidden methodology is bad methodology.
Full bibliography, including secondary sources and methodological monographs, is maintained at .
The unit discloses the toolchain that supports its analytical work. FP3 Method is human-led methodological discipline; the toolchain is contemporary. We name our tools for the same reason we name our authors: hidden methodology is bad methodology.
Full doctrinal treatment: Editorial Doctrine · Article 7 · Tool & Augmentation Disclosure → · Institutional adoption pathway: