Baseline Intelligence.
Reference profiles and capability assessments for nation-states, organizations, networks, and systems.
- TYPICAL CADENCE
- One-shot; updated on-request
- TYPICAL SOURCE COUNT
- 1,500–4,000
- TYPICAL OUTPUT LENGTH
- 2,000–6,000 words
- TYPICAL CONFIDENCE REGISTER
- Descriptive, with sourcing emphasis
// QUESTION CLASS
What is the thing, structurally?
Most intelligence consumers want a forecast. But you cannot produce a defensible forecast without an equally defensible baseline | the structural facts about the subject that don't change week to week. The Baseline engine produces that baseline. An organization's architecture, chain of command, and key dependencies. A person's career arc, public positions, networks. A system's components, interfaces, and known failure modes. These are the reference documents that anchor every other engine's work. Baseline reports are longer than the others because they are built once and then used. A good baseline outlasts the question that prompted it. The output is organized as a reference document, not a narrative | the reader will come back to specific sections, so every section must stand alone.
// ANATOMY
A Baseline report contains:
A Baseline report is structured as a reference, not a narrative. The reader does not read it cover to cover; they come back to the section they need. Every section is independently coherent and cross-referenced to others. The "Open Questions" section is mandatory — it names the specific gaps in the baseline so that downstream analysis knows what it's building on. This is the only engine whose output is expected to outlive the question that prompted it.
// TRADECRAFT
A baseline earns trust by being boring, correct, and complete.
Every structural claim has a documentary source.
Baseline reports are read by people who need to cite them. If a claim appears without a footnote, it does not appear. Oral history, analyst inference, and "widely understood" are not source categories; they're flags to investigate or to exclude.
ICD 206 · Full source attributionOpen questions are a section, not a footnote.
Every baseline has gaps. An honest baseline names them explicitly — the question we couldn't answer, the source we couldn't reach, the contradiction we couldn't resolve. This is the difference between a reference document and a PR document.
ICD 203 · Standard 2: Properly expresses uncertaintiesThe baseline is versioned.
Baseline reports are stable — but not static. When a structural fact changes (a merger, a new appointment, a doctrinal shift), the report is re-issued with a new version, and the change is diffed at the top. No silent updates.
ICD 203 · Standard 7: Explains change to analytic judgmentsOften paired with: Situational (the baseline Situational monitors against), Forecast (the baseline a forecast projects from).
// OTHER ENGINES