Skip to content

White Paper Mode

White paper mode is for conceptual, executive, and strategic manuscripts where narrative clarity and argument structure matter more than primary data novelty. The draft is Markdown-first; the export includes executive-oriented deliverables.

When to use

Choose white paper mode when your materials are primarily:

  • Strategic analysis, internal research notes, or position documents
  • Literature synthesis without original experimental data
  • Thought leadership or industry perspective pieces
  • Executive briefings or policy papers

route_mode.py recommends this mode automatically when materials lack strong empirical signals (experimental logs, datasets, statistical results). If routing is ambiguous, you can override at the routing checkpoint.

Stage W2: Perspective Discovery

Goal: Map the viewpoint landscape before locking argument structure.

Folio scans your raw materials, idea.md, and available external literature to identify distinct perspectives. It produces:

  • planning/perspectives.json — Array of perspectives with perspective_id, viewpoint, key arguments, source materials, and relevance
  • planning/question_tree.json — Structured questions the paper must answer, grouped by planned section

Exit condition: At least three distinct perspectives identified, or user waiver recorded in logs/run_log.md.

Stage W3: Argument Structure

Goal: Thesis, counterthesis, and claim ledger aligned to perspectives.

Folio formulates the thesis and strongest counterthesis, then maps claims to perspectives. It produces:

  • planning/argument_graph.md — Argument flow as text/ASCII: sections → claims → dependencies
  • planning/claim_ledger.json — Claims with extended types: fact, interpretation, opinion, forecast

Gate C — Claim ledger validation:

cd plugins/folio && python scripts/build_claim_ledger.py workspace/

Folio blocks on failure and fixes JSON and claims until this passes.

White paper vs. research paper claim types

White paper mode uses extended claim types (fact, interpretation, opinion, forecast). Research paper mode uses quantitative, qualitative, methodological. Hybrid mode supports both sets.

Stage W4: Outline and Section Drafting

Goal: Markdown-first draft for executive readability.

Folio generates planning/outline.json and drafts drafts/paper.md. Key constraints:

  • Draft is Markdown, not LaTeX
  • Citations use [Author, Year] style; a parallel drafts/reference_list.md tracks references if refs.bib is not yet built
  • Every non-trivial claim must trace to claim_ledger.json
  • Internal code names and unexplained jargon are forbidden (also enforced by IP scan)

Optional: long works can split sections under drafts/sections/.

Gate: All question_tree.json questions are covered by outline sections; every non-trivial claim traces to the ledger.

Stage W5: Review

IP scan (required):

cd plugins/folio && python scripts/scan_redlines.py workspace/

Writes reviews/ip_safety_report.md. IP Gate: Violations block completion until cleared, redacted, or explicitly acknowledged.

Hostile review: Folio reads drafts/paper.md as an external critic and writes reviews/hostile_review.md covering logical gaps, unsupported assertions, competitive risk, clarity, and factual risk.

Standard review: Folio writes reviews/review_round_1.md and reviews/scorecard.json with numeric subscores for coherence, support, citations/clarity, and venue fit.

Gate G — Non-regression: If the overall score drops after a repair round, Folio reverts to the better draft. Maximum 3 review rounds; best-scoring artifact is retained with residual issues documented.

Stage W6: Executive Outputs

Folio produces export-ready deliverables:

  • exports/executive_summary.md — ~1 page for leadership: thesis, why it matters, risks, asks
  • exports/bd_talking_points.md — Bullets for business development conversations: customer pain, differentiation, proof, objections

Human Checkpoint 3: Folio asks you to confirm BD/summary tone, sensitive claims, and any deferrals before packaging.

Artifacts produced

Artifact Stage Description
planning/perspectives.json W2 Perspective map
planning/question_tree.json W2 Section-linked question structure
planning/argument_graph.md W3 Argument flow diagram
planning/claim_ledger.json W3 Claims with types and perspective links
planning/outline.json W4 Section structure
drafts/paper.md W4 Working Markdown draft
reviews/ip_safety_report.md W5 IP scan results
reviews/hostile_review.md W5 External critic review
reviews/review_round_N.md W5 Standard review notes
reviews/scorecard.json W5 Quality scores
exports/executive_summary.md W6 Leadership summary
exports/bd_talking_points.md W6 BD conversation bullets

See Artifacts for the complete cross-mode artifact list.

What white paper mode does not do

  • Does not produce a LaTeX draft (use Research Paper or Hybrid for LaTeX output)
  • Does not run a literature search and citation verification pipeline (only citation-style references in prose)
  • Does not generate figures from data sources (figures from raw materials can be referenced; new figures must be provided)