Research State Snapshot

• Theory Version: TH-COMM-2026-0001 v0.2.0, candidate
• Knowledge Base Version: ROS 1.0.0; domain knowledge-base version not established
• Highest Confidence Areas: textual move order in the five selected units; explicit dependency and navigation structures
• Lowest Confidence Areas: causal reader effects; cross-cultural transfer; relative weight of author, genre, and proof regime
• Largest Remaining Unknown: whether conditional architecture selection improves comprehension, transfer, calibration, and error prevention over strong alternatives
• Active Research Streams: MS-COMM-2026-0002 outcome measurement; MS-COMM-2026-0003 comparative program
• Recently Invalidated Ideas: famousness as effectiveness evidence; a writer as one stable transferable style; every DF- as a decision framework
• Priority Changes: second-rater rubric reliability and a comparative audience experiment become the next discriminating work

Executive Summary

A five-domain pilot supports a conditional model: purpose, audience prerequisites, use mode, authority, and proof obligation explain more of a text's progression architecture than a whole-writer style label. Darwin builds a difficult causal mechanism through familiar analogy, heterogeneous cases, objections, and synthesis. Euclid exposes a short dependency chain under shared primitives. Brown narrows an institutional question, acknowledges historical limits, connects present function and effect, and states a holding. Descartes makes a problem experiential before compressing a method into four rules. RFC 9110 layers scope, terms, abstractions, normative rules, and navigation for expert retrieval [EV-COMM-2026-D6B2; EV-COMM-2026-6A2D; EV-COMM-2026-8434; EV-COMM-2026-5479; EV-COMM-2026-3377].

The result supports TH-COMM-2026-0001 but does not establish causal audience effects. Confidence is Medium (0.63 judgment): primary-text order is directly observable and repeated coding was 12/15, while the tiny, Western, English- readable corpus confounds author, genre, era, and domain. The largest caveat is that fit recommendations remain hypotheses until audience experiments compare architectures on purpose-bound outcomes.

Original Objective

Execute the first bounded mission under the comparative-writer program: determine how explanatory progression differs across one landmark work from science, mathematics, law, philosophy, and engineering, and identify plausible cross-domain pattern candidates.

Success Criterion

Another analyst can reproduce selection and coding; observation and interpretation are distinct; transfer limits and counterevidence are recorded; and at least one initial hypothesis is supported, narrowed, or rejected.

Scope

Included

Five declared works and units; a frozen corpus manifest and rubric; primary text analysis; contextual scholarship; repeated-coding stability check; functional profiles; direct contrasts; fit framework; ethics and anti-patterns; ten applied cases; evidence, hypotheses, decision, and journal records.

Excluded

Ranking communicators, exhaustive writer coverage, claims that one text represents a discipline, word-level stylometry, causal reader claims, full cross-cultural generalization, and imitation of distinctive living-writer style.

Scope Changes

The uploaded prompt is a program-scale mandate. ROS already decomposed it into MS-COMM-2026-0003 and first mission MS-COMM-2026-0004. This REP completes that pilot rather than falsely claiming completion of the multi-REP program.

Repository Context

The baseline theory is conditional fit (TH-COMM-2026-0001). Existing evidence already rejects style-independent effectiveness and a universal plain-language rule. RP-COMM-2026-0002 remains partial, so reader-outcome metrics have not yet been validated on a selected comparative task. The worktree contained pre-existing user modifications to missions, context, handoff, and program files; none were overwritten.

Current Understanding

Explanatory progression is best modeled as a sequence of functions that moves an audience from a declared or inferred starting representation to a target representation. Every selected text orients, manages prerequisites, connects support to a claim under a domain-specific warrant, and closes. What changes is the order, explicitness, and acceptable omission:

• closed proof systems foreground dependencies;
• historical causal explanations accumulate and reconcile evidence;
• judicial opinions balance legal warrant, institutional legitimacy, and closure;
• reflective philosophy can motivate abstraction through narrated inquiry;
• specifications trade linear ease for stable definitions and retrieval.

These are bounded patterns, not author essences. A useful recommendation has the form: for purpose P, audience A, use mode M, proof/risk regime R, use functional pattern T, expect mechanism hypothesis X, avoid failure F, and test outcome O.

Key Discoveries

1. Warrant and example are not the same move. Darwin's examples can explain and support without having Euclid's deductive force; the rubric was revised to preserve this distinction [EV-COMM-2026-D6B2; EV-COMM-2026-6A2D].
2. Navigation is part of architecture. RFC 9110's section graph supports recurrent expert retrieval, making linear readability an incomplete quality criterion [EV-COMM-2026-3377; HY-COMM-2026-CFF5].
3. Authority is contextual, not stylistic. Brown's institutional voice and unanimity cannot be responsibly copied by a communicator without comparable authority and process [EV-COMM-2026-8434].
4. Compression requires practice support. Descartes's four rules are memorable, but scholarship treats them as a practice-dependent abridgment rather than a complete universal procedure [EV-COMM-2026-5479; EV-COMM-2026-7D8D].
5. Whole-work interpretations can conflict. Multiple scholarly models of Darwin's “long argument” have textual support, so coding must remain unit-bound and preserve alternatives [EV-COMM-2026-6F00].
6. High-risk protocols can dominate writer choice. Medical summaries, UI warnings, and sensitive interpersonal messages require safety, consent, action, and relational constraints before writer-derived architecture.

Evidence Registry

Hypothesis Registry

Failed Assumptions

Open Questions

1. Can independent coders apply the move rubric with acceptable agreement?
2. Does architecture add predictive value beyond genre, lexical, and sentence variables?
3. Which pair of architectures produces a measurable, ethical contrast on one task without confounding content accuracy?
4. How do non-Western, oral, Indigenous, dialogic, and multimodal traditions alter the taxonomy?
5. When does explicit warrant reduce rather than increase comprehension?
6. Can overview/reference layers remain synchronized cheaply enough to help?
7. How should relational and experiential goals be evaluated without reducing them to task completion?

Recommended Next Research

1. Reliability: give the frozen rubric and 20 stratified units to a second analyst; preregister agreement thresholds and adjudication.
2. Discriminating experiment: select one unfamiliar scientific or technical mechanism and compare familiar-case/objection architecture against explicit dependency architecture with identical claims and visuals.
3. Measures: comprehension, near/far transfer, calibrated confidence, critical misconceptions, assistance, and accessibility subgroups. Stop if variants cannot hold content constant or coding reliability is unusable.

Research Backlog

1. Crossed author-by-genre corpus to separate author, genre, and proof regime.
2. Non-Western and multilingual progression study with qualified scholars.
3. Oral/public-address and multimodal comparison.
4. Law: compare majority, dissent, brief, statute, and contract.
5. Engineering: compare tutorial, RFC, API reference, runbook, and incident postmortem.
6. Quantitative move-transition analysis after reliability is established.
7. Longitudinal retention and decision-quality tests.

Suggested Specialized Research Agents

• Comparative-rhetoric specialist: select culturally situated non-Western units, translation policy, and local scholarship; output a bias-corrective corpus proposal.
• Discourse-coding methodologist: test code definitions, agreement, and alternative segmentation; output reliability evidence and rubric v1.1.
• Experimental psychologist/HCI researcher: preregister the architecture comparison and outcome model.
• Accessibility specialist: identify exclusion mechanisms and recruitment/ representation requirements.
• Legal historian: audit Brown context and distinguish opinion rhetoric, bargaining, doctrine, reception, and remedy.

Parallel Research Opportunities

Corpus diversification, legal-context audit, and technical overview/reference design can proceed independently after sharing rubric v1.0. Audience experiments depend on measurement work and must not begin until content parity, outcome scoring, and accessibility plans are frozen.

Risks

• Epistemic: author, domain, era, and genre are confounded; close reading is interpretive; source access was uneven.
• Operational: registry generation may overwrite manual index edits if not rebuilt and diff-reviewed.
• Ethical: persuasion optimization can enable manipulation; institutional tone can launder authority; cultural abstraction can appropriate voice.
• Accessibility: formal proof and dense reference graphs can exclude readers without prerequisite knowledge or compatible formats.
• Adoption: users may collapse bounded functions back into famous-writer personas or use the matrix as a ranking.

Cross-Discipline Opportunities

Combine rhetorical move analysis with discourse coherence, educational text- structure research, HCI information scent/retrieval, legal argumentation, formal dependency graphs, cognitive accessibility, translation studies, and causal experimental design.

Knowledge Relationships

EV-D6B2/6A2D/8434/5479/3377 → HY-D246 → TH-0001 provides the central chain. EV-6F00 limits over-simple Darwin coding; EV-7D8D limits method compression. EV-3377 + EV-E9D5 → HY-CFF5. DF-96CF adopts the conditional framework without promoting the hypotheses to settled theory.

Provisional concept and glossary records

No canonical concept/glossary registry or schema exists. To avoid inventing a shared type, these definitions remain provisional here:

• Explanatory progression: ordered functional moves transforming an audience's starting representation into a target representation.
• Proof obligation: domain- and task-specific burden determining which warrants, alternatives, and uncertainty must be visible.
• Writer-function model: bounded architecture abstracted from a work, excluding distinctive persona and surface mimicry.
• Authority laundering: using the signals of institutional or expert authority without the process, evidence, or mandate that warrants them.
• Compression amputation: memorable simplification that removes a decision-critical boundary, uncertainty, or exception.

Theory Impact Assessment

• Affected Theory Records: TH-COMM-2026-0001
• Affected Engineering Principles: declare goal/audience/context; separate objective performance from subjective fluency; treat accessibility and ethics as constraints.
• New Principle Candidates: proof-obligation fit; separate example from warrant; navigation as architecture; protocols can dominate writer choice.
• Deprecated Principles: none canonical; “select a writer by document type” is rejected as a candidate shortcut.
• Confidence Changes: TH-COMM-2026-0001 remains Medium; qualitative support broadens from procedural/measurement sources to five textual domains but no causal outcome evidence warrants a numeric increase.
• Predictions Created: HY-COMM-2026-D246 and HY-COMM-2026-CFF5.
• Predictions Invalidated: universal writer ranking and frictionless hybrid maximalism are rejected as program assumptions, not prior registered HYs.
• Required Theory Registry Updates: none until independent coding and an audience experiment; link this REP during the next theory revision.

Research Quality Metrics

Research Debt

Missing Evidence

Independent reader outcomes, full-text review of some scholarship, reception evidence by audience, and within-writer genre contrasts. Consequence: do not claim effectiveness or author essence.

Missing Experiments

Second-rater reliability and randomized architecture comparisons. These gate promotion of both hypotheses.

Missing Disciplines

Comparative rhetoric, translation studies, disability studies, oral tradition, Indigenous studies, education practice, journalism, medicine, organizational communication, and interface research.

Weak Areas

Cultural balance, multilingual validity, historical audience reconstruction, and accessibility are weak. Revisit before expanding the framework's coverage.

Replication Needed

Repeat coding with independent raters and alternative segmentation; replicate any audience result across expertise and language groups.

Tool Limitations

Web source extraction supplied primary texts and metadata but not all paywalled full articles. No participant recruitment or independent coder was available.

Assumptions Awaiting Evidence

Architecture adds value beyond content and genre; overview/reference layering helps mixed expertise; explicit dependency improves calibrated understanding.

Repository Updates

Created the corpus manifest, rubric, profiles, comparison matrix, applied cases, executive synthesis, fit framework, eight evidence records, two hypotheses, one Decision Record, one journal, and this REP. Updated registries and mission status/links. Pre-existing unrelated user changes were preserved.

Website Updates

No hand-edited website source was necessary. The repository publisher consumes research artifacts; build output is generated and should not be committed unless repository convention later requires it.

AI Consumption Notes

Reliable retrieval terms: explanatory progression, proof obligation, writer-function model, conditional fit, visible dependency, layered reference, authority laundering. Safe fact: the five texts have the coded order documented in linked primary evidence. Caveat: do not say those architectures caused superior comprehension. Misuse risks: writer impersonation, universal ranking, authority laundering, analogy inflation, and using persuasion as the objective without truth and welfare constraints.

Handoff Instructions

1. Read the corpus manifest and rubric v1.0 without changing them.
2. Select 20 stratified units from the same works; have an independent analyst code them and record raw disagreements.
3. If agreement is usable, draft rubric v1.1 without altering v1.0.
4. Coordinate with RP-COMM-2026-0002 before preregistering outcome measures.
5. Create the next parallel-safe IDs according to IDENTIFIERS.md.
6. Run ./ros registry check, ./ros validate, npm run research:validate, and npm run research:build; diff generated indexes before accepting.

Research Journal

JR-COMM-2026-572D records repository reconstruction, conflict resolution, corpus/rubric freeze, evidence review, falsification, repeated coding, and the stop decision.

Appendix

Deliverable map

Assumption-change rule

Increase confidence only after independent coding and outcome evidence. Reduce confidence if a crossed corpus shows stronger author than proof-regime effects, if architecture adds no predictive value beyond content/genre, or if declared fit patterns systematically disadvantage an audience subgroup.

Program-mandated future research seeds

Competing hypotheses: (1) genre conventions alone explain progression once document type is known; (2) audience expertise explains more variance than proof obligation; (3) a small set of universal coherence operations dominates both author and domain. Each competes with HY-COMM-2026-D246 and can be tested in a crossed corpus.

Overlooked communicators/traditions: Sei Shōnagon, Lu Xun, Rabindranath Tagore, Chinua Achebe, Naguib Mahfouz, classical Chinese persuasion, Nyāya argument, Buddhist pedagogical dialogues, West African oral traditions, Indigenous knowledge protocols, disability-led plain-language practice, and community-authored technical documentation. Inclusion requires appropriate language and cultural expertise rather than translated excerpt mining.

Quantitative options after reliability: move-transition matrices; dependency-graph depth and fan-out; explicit/implicit warrant rates; navigation path length; qualification and counterargument density; multilevel models that separate author, genre, domain, era, and audience; and preregistered mediation tests linking architecture to comprehension or error outcomes. Counts cannot substitute for semantic coding or audience measures.

Completion Checklist

• Metadata is complete and internally consistent.
• Research State Snapshot is complete.
• All mandatory sections exist.
• Important claims trace to evidence and hypotheses/theory.
• Contradictory interpretations and failed assumptions are preserved.
• Theory impact and deferred registry update are explicit.
• Quality metrics state method and limitations.
• Research debt is prioritized.
• Completion is explicitly bounded to the pilot, not the program.
• Repository and website updates are accurate.
• Handoff permits continuation without conversation history.
• Links, identifiers, and registries validate; site build completed.
• A capable agent can reconstruct the bounded investigation.