research-document
Writer-to-purpose fit framework v1.0
Writer-to-purpose fit framework v1.0
This is a selection framework, not a DF- record. Under current governance,
DF- means Decision Record; DF-COMM-2026-96CF records the decision to use
this conditional model.
Step 1: declare the communication contract
Record:
purpose + audience starting state + audience resistance + target state + medium/use mode + proof obligation + stakes + uncertainty + power relation + accessibility constraints.
If these cannot be stated, do not select a writer model yet.
Step 2: select architecture by dominant need
| Need | Primary architecture | Useful bounded model | Exclusion condition |
|---|---|---|---|
| Make a novel mechanism plausible | familiar case → causal mapping → varied evidence → objections → synthesis | Darwin ch. IV | Analogy would be mistaken for direct evidence |
| Make a derivation auditable | primitives → target → ordered transformation → explicit warrants → closure | Euclid I.1 | Premises or definitions are contested |
| Make and legitimate a high-stakes decision | narrow question → acknowledge adverse material → governing rule/warrant → effect → holding/action | Brown | Communicator lacks decision authority or suppresses material evidence |
| Motivate adoption of a method | experienced problem → failed alternatives → compact principles → bounded application | Descartes Part II | Method is unvalidated but presented as universal/certain |
| Coordinate expert implementation | scope → canonical terms → layered abstraction → normative rules → examples/exceptions → navigation | RFC 9110 | Audience is novice, linear task completion is primary |
| Build reflective social recognition | recurring concrete motif → delayed disclosure → retrospective synthesis | Fitzgerald, Gatsby | Immediate action or unambiguous instruction is required |
| Create recoverable emotional subtext | concrete surface → patterned omission → inferential closure | Hemingway, selected fiction | Missing information affects safety, evidence, consent, or access |
| Sustain genuine deliberative tension | concrete shared task → countervoice → unresolved but bounded closure | Frost, “Mending Wall” | The audience needs one immediate safe action |
| Teach through a conceptual puzzle | simplified premise → crisis → model comparison → reversal | Asimov, selected fiction/essays | Social uncertainty is presented as deterministic law |
| Mobilize through public witness | representative particulars → anaphoric accumulation → direct address | Ginsberg, “Howl” | Neutral analysis or calibrated evidence synthesis is primary |
| Generate before evaluating | uninterrupted associative draft → separate evidence/revision pass | Kerouac as generative method only | Raw text will reach affected audiences |
| Make civic observation accessible | mobile observer → concrete street detail → humorous/political collage | Ferlinghetti, selected poetry | Vulnerable people become scenery or perspective is concealed |
Step 3: compose only compatible functions
- Coherent: Darwin orientation + Euclidean dependency exposure for a scientific explanation; linear novice overview + RFC reference layer; Descartes motivation + RFC normative procedure when personal narrative is clearly separated from requirements.
- Tense but manageable: Brown categorical closure + Darwin uncertainty. Resolve by separating established decision, evidentiary confidence, and open implementation questions.
- Usually incoherent: literary ambiguity inside a safety warning; personal authority used as institutional warrant; memorable compression that removes decision-critical qualifications.
Step 4: apply ethical gates
- Truth: Are claim, evidence, inference, and uncertainty separable?
- Consent and welfare: Does the pattern bypass deliberation or exploit distress, identity, or power asymmetry?
- Authority: Is authority earned, scoped, and contestable?
- Accessibility: Can affected audiences perceive, understand, and act?
- Alternatives: Are material counterarguments and consequences visible?
- Identity: Is the system extracting a function rather than impersonating a living writer or appropriating culturally situated voice?
- Situated authority: Does the technique depend on community standing, lived relation, nation, land, language, or access the communicator lacks?
- Translation: Are edition, translator, language-level uncertainty, and meaningful alternatives recorded?
- Circulation: Is any material sacred, restricted, private, relationally governed, or otherwise inappropriate to abstract and redistribute?
Failing a gate excludes the technique even if it would increase persuasion.
Step 5: validate against the intended outcome
- Inform/explain: comprehension, calibrated confidence, transfer.
- Decide: identification of options/tradeoffs, decision quality, traceability.
- Instruct/warn: unassisted correct action, critical errors, assistance; time only when time is mission-critical.
- Coordinate/specify: interpretation agreement, interoperability, defect rate, retrieval success.
- Relational communication: accurate recognition, trust, repair, autonomy.
Do not use readability, liking, memorability, or persuasion as a sole proxy.
Purpose taxonomy
Purposes are overlapping dimensions, not mutually exclusive bins:
- Epistemic: inform, explain, teach, compare, diagnose, preserve or reduce uncertainty.
- Deliberative: surface options, argue, negotiate, decide, legitimate.
- Operational: instruct, warn, prevent error, coordinate, document.
- Relational: build trust, reassure, challenge, repair, create identification.
- Experiential: entertain, provoke, memorialize, sustain ambiguity.
- Mobilizing: motivate, advocate, prompt action.
The dominant purpose sets the outcome measure; secondary purposes constrain the architecture. Power and audience welfare are cross-cutting constraints, not purposes.
Pattern library and anti-patterns
| Pattern | Use condition | Mechanism hypothesis | Failure mode |
|---|---|---|---|
| Familiar-to-causal bridge | audience knows source domain and mapping is valid | activates schema and reduces initial abstraction | seductive analogy substitutes for evidence |
| Visible dependency chain | premises accepted; audit matters | externalizes inferential dependencies | overwhelms novices; hides disputed premises |
| Objection-resolution loop | resistance or uncertainty is material | calibrates trust and tests robustness | straw objections manufacture credibility |
| Narrow-question closure | authorized decision under bounded issue | limits inferential scope and action ambiguity | hides systemic consequences outside frame |
| Motivation-before-method | adoption requires a mental-model shift | gives rules a felt problem and retrieval cue | delays action; persona crowds out evidence |
| Layered reference graph | experts need recurrent lookup | supports non-linear retrieval and stable terminology | orientation cost and link chasing |
| Linear overview + reference layer | mixed expertise/use modes | separates onboarding from exact retrieval | layers drift or contradict |
Anti-pattern catalog
- Surface mimicry: copies cadence/diction without the reasoning function.
- Authority laundering: imports judicial, scientific, or standards tone without corresponding process or evidence.
- Analogy inflation: a vivid mapping is reported as causal proof.
- Compression amputation: memorable summary deletes a boundary or risk.
- Qualification fog: caveats prevent any claim from being testable.
- Specification dumping: reference structure is used as novice instruction.
- Hybrid maximalism: every strong technique is combined, producing incompatible audience roles and excessive load.
Confidence
Medium (0.62 judgment). The model is supported by textual comparison and prior conditional-fit evidence, but no comparative audience experiment has tested whether these selections outperform alternatives.