What We Publish.
What We Do Not.

An intelligence analysis desk, not a magazine

The unit is constituted as an intelligence analysis desk. It operates editorially in Italy under the FP3NEWS commercial brand of the editorial title FreePress3.News, registered with the Tribunale di Verona · Registro Stampa N. 2149 · R.G.V.G. 7274/2020 · 21 December 2020. Publisher: MODENART SRLS (CCIAA Verona · P.IVA 04522520230 · REA VR-427683). Editor-in-Chief: Iacopo Mutascio (Ordine dei Giornalisti del Veneto). Its output discipline is that of a private analytic unit producing scenario-based intelligence assessments for institutional readers, not that of a generalist news outlet producing daily news flow.

This distinction is not cosmetic. It governs what we accept as evidence, how we commit to forecasts, who we distribute to, and what we expect from our readers. A reader who treats a PULSE INSIDER paper as a news article will mis-read it. A reader who treats it as a probabilistic assessment with declared confidence and committed resolution criterion will read it correctly.

Anti-overclaim discipline

Every analytical claim produced by the unit passes through the §17 operator, the signature mechanism of the riscontro protocol. The §17 operator forces three explicit declarations on every claim:

1. Evidence status — verifiable data, or stated inference. No hedge.
2. Confidence band — explicit percentage interval. No "high confidence" without a number.
3. Resolution criterion — a binarisable event and a date by which the claim will be true or false.

A claim that cannot pass §17 is rewritten or withdrawn. This is the most important single mechanism of the unit's analytical discipline. It is the difference between a calibrated forecast and a confident-sounding opinion.

Why we do not publish analyst identities

The unit operates through the FP3 method — a structured, human-led analytical method. the FP3 method is the unit's operational modality: human-led, AI-augmented, §17-governed. Cluster mapping and node functions are publicly disclosed at . The framework is a method, not a staff chart — human-led and single-author. FP3 is a human-led, single-author, AI-augmented unit: the analytical work is led and signed by the Director, augmented by AI devices across the. What is reserved is not the headcount — that is openly declared — but the method itself: toolchain configurations, prompt architectures and workflow scripts, held under §17 disclosure governance as declared methodological discipline. For the toolchain layer of the architecture, see Article 7 · Tool & Augmentation Disclosure.

This is the standard practice of many private analytic units producing intelligence-grade material for institutional readers. The classical academic publication model — where every contributor is named, with CV and biographical profile — serves a different epistemic function (peer-review traceability for reproducibility). The intelligence analytic model serves a different function: protect contributor identity to preserve the unit's operational integrity, while exposing the analytical product itself to maximum accountability through declared confidence, registered resolution, and ex-post riscontro verification.

There are well-documented institutional precedents for this model: certain Oxford Analytica briefing units, Stratfor private intelligence reports, RUSI restricted publications, and private boutique advisory units typically operate under disclosed-by-product / undisclosed-by-author protocols. The unit follows this established practice.

Who receives what

The unit operates a tiered distribution architecture aligned with the analytical sensitivity of each output.

Routing discipline is enforced under source integrity H3 (Distribution OPSEC). Cross-sector contamination is structurally prevented — for instance, investigative dossiers on Italian media systems are never distributed to wine industry contacts, regardless of subscription tier.

Negative declaration

A doctrine is also defined by what it refuses. The unit does not:

• Produce daily news flow. We publish papers, not articles.
• Issue forecasts without declared probability and registered resolution criterion. Confidence-sounding language without numbers is forbidden under §17.
• Retract failed forecasts. Failed forecasts are scored, kept on file, and discussed in subsequent papers as drift evidence.
• Accept commissioned analysis where the conclusion is implicitly specified by the commissioning party. We work under independence terms only.
• Publish analyst-level CVs, biographies, or contributor identities. See Article 3.
• Operate behind political alignment. The unit carries no party affiliation, no campaign deliverables, no advocacy-cycle output.
• Distribute restricted material to readers who decline to sign an NDA appropriate to the tier. The signature is the protection of the reader as well as the unit.

How to assess our work

A serious institutional reader assessing the unit should run the following three tests:

1. Riscontri test — pull the numbered Archive, identify papers whose resolution criterion has expired, and verify that the riscontros published are consistent with the declared probabilities at the time of publication. If the riscontros converge on or below the case baseline over a sustained sequence, the unit is calibrated.
2. Independence test — identify papers whose conclusions cut against the commercial or political interests of any party the unit is institutionally connected to. If such papers exist and were distributed despite the friction, the unit is operationally independent.
3. Boundary test — request methodology disclosure for a specific Protocol under NDA. The disclosure provided should be specific enough to allow methodology review by a qualified institutional reviewer, while preserving operational reserve. If both conditions are satisfied, the Disclosure Boundary is honest.

These three tests are the right way to assess us. We invite them.

On the analytical toolchain

The unit discloses, by editorial doctrine, the toolchain that supports its analytical work. FP3 Method is a human-led methodological discipline; it operates with the standard digital toolchain of contemporary analytical practice. We do not consider this a concession. We consider it the answer to a legitimate question that any 2026 institutional reader is entitled to ask of any analytical unit publishing today — answered here ex ante, before it must be answered ex post.

The disclosed toolchain comprises:

• Triangulated OSINT databases under source integrity Admiralty source grading (A1–F6).
• Statistical, charting, and time-series toolchain for empirical work and visualisation.
• Previsioni & Riscontri infrastructure for ex-post riscontro tracking and resolution audit.
• Large Language Model augmentation for drafting, structural editing, cross-reference retrieval, alternative-hypothesis stress testing, and red-team prompting — under explicit §17 supervision and human editorial commitment.

We hold that any 2026 analytical unit producing output of the density, pace, and cross-domain reach published by FP3NEWS · Intelligence Analysis is, by definition, supported by digital augmentation. The relevant question for an institutional reader is not whether augmentation is involved, but how its use is disciplined. Our answer is structural.

Four disciplines govern the toolchain:

Coverage discipline. FP3NEWS holds itself accountable on two core beats — Italian macro & stratified sovereignty and energy & Mediterranean geopolitics — where the Previsioni & Riscontri concentrates its forecasting record. Every dossier published outside these beats is an applied method demonstration: the same epistemic apparatus (ex-ante probabilities, fixed resolution criteria, explicit falsifiers, §17 anti-overclaim) deployed on a different domain, and labelled as such. Breadth of application proves the method travels; depth of accountability stays where the beats are.

1. §17 is tool-agnostic. No claim is published without passing through the §17 operator. §17 is human-enforced, post-generation, on every claim — irrespective of the drafting tool used. A claim that cannot pass §17 is rewritten or withdrawn, regardless of provenance.
2. Source verification is independent of drafting. source integrity Admiralty grading is applied to source material on its own merits. Any content traceable to LLM drafting is verified against original sources before publication. The unit does not cite from drafting layers; it cites from verified sources.
3. The Previsioni & Riscontri measures output, not tooling. A riscontro is indifferent to whether the underlying drafting used pen, software, or augmentation. A forecast either resolves within its declared band, or it does not. This is the only verification that matters and the only one we publish.
4. Editorial responsibility rests with the Director. Under Italian Press Law (L. 47/1948 art. 11), responsibility for every published claim rests with the Direttore Responsabile. No augmentation layer alters that responsibility. The Director's signature is a statement of analytical and editorial commitment, not a claim about the keyboard.

This Article completes the Disclosure Doctrine. We disclose what we use because hidden methodology is bad methodology — and because peculiarity does not require concealment. The discipline is the distinction. The discipline is also licensable: institutions seeking to adopt the FP3 Method architecture under formal agreement may consult .

How weak signals are gated before they reach a forecast

Open-source and social chatter is the fastest, dirtiest vector: high coverage, high speed, low Admiralty grade. The desk does not trust it — it processes it. A signal passes five stages before it goes anywhere near a published forecast: ① Recon (≥1h daily directed scan by the Director) · ② Triage (clustering, noise stripped) · ③ Confirm gate (the hard wall) · ④ Corroborate (mapped against existing theses — confirm, weaken, complicate?) · ⑤ Feed & log (Bayesian update where material, archived with falsifiers).

The contamination gates (stage ③): G1 provenance — walk the claim back to source-zero or it stays a rumour · G2 citation & figure — every number/date/quote checked against a Tier-1 source · G3 synthetic-media — reverse-image & manipulation screening, with the honest caveat that detector false-negatives are rising, so absence of a flag is not proof of authenticity · G4 cross-source triangulation — a single source never promotes a claim above Tier-3.

What it is: a detection layer (where to look, not what is true) · a tripwire for weak signals · a corroboration engine. What it is not: a source of record (social never anchors a claim alone) · a debunking column · a mass-collection operation · an auto-publishing bot — nothing ships without a human signature. The grammar is probabilistic, not adjudicative: an honest miss on the record strengthens the desk more than a claim of infallibility ever could.