Penwright Research Program — Sub-Paper Plan

Provenance: verbatim copy of PENWRIGHT RESEARCH PROGRAM — SUB-PAPER PLAN .txt v1.0, drafted 2026-05-03. Defines the twelve-paper Penwright Research Program, three tiers, shared dataset strategy, product-integration mapping, and three-phase publication strategy. The cleaned and integrated form lives in program.md; this file preserves the source as-is for reference.

---

Drafting status (as of 2026-05-05)

The status notes below sit alongside the verbatim source above. The source plan is preserved unmodified; this section tracks which papers have been converted into prose drafts and where they live.

Paper	Tier	Status	File
Paper 1 — AI-Augmented Authorship and Human Skill Development	1	✅ Drafted (technical + public versions)	penwright-paper-01-technical.md (technical, outline) · penwright-paper-01-public.md (general-audience)
Paper 2 — Corpus Control and Epistemic Agency	1	✅ Drafted 2026-05-05	penwright-paper-02-corpus-control.md
Paper 3 — The Authorship Packet Model	1	✅ Drafted 2026-05-05 (PA-006)	penwright-paper-03-authorship-packet.md
Paper 4 — A Measurement Framework for AI-Augmented Writing Skill Development	2	✅ Drafted 2026-05-05 (PA-005)	penwright-paper-04-measurement.md
Paper 5 — Dependency and Independence	2	✅ Preregistration drafted 2026-05-05 (PA-007) — prose paper gated on pilot data	penwright-paper-05-dependency.md
Paper 6 — Learning Loops	2	✅ Drafted 2026-05-05	penwright-paper-06-learning-loops.md
Paper 7 — Genre-Specific Effects	2	✅ Preregistration drafted 2026-05-05 (PA-008) — prose paper gated on pilot data	penwright-paper-07-genre-effects.md
Paper 8 — Longitudinal Effects	3	⬜ Gated on external-operator pilot (PA-009)	—
Paper 9 — Transfer of Skill Outside AI Systems	3	⬜ Gated on Paper 8 pilot data	—
Paper 10 — Cognitive Load and Attention	3	⬜ Not yet drafted	—
Paper 11 — Corpus Composition and Output Diversity	3	⬜ Not yet drafted	—
Paper 12 — Human–AI Working Alliance	3	⬜ Not yet drafted	—

Phase 1 (0–3 months) Tier-1 set + Paper 4 is now drafted: Paper 1 (technical + public), Paper 2 (Corpus Control), Paper 3 (Authorship Packet Model), Paper 4 (Measurement Framework). The Tier-1 foundational set is complete. Phase 2 papers (5, 6, 7) remain queued; Phase 3 papers (8–12) remain gated on the external-operator pilot accumulating production data.

---

📄 PENWRIGHT RESEARCH PROGRAM — SUB-PAPER PLAN

Version 1.0 — Publication + Product Integration Strategy

1. Program Architecture (How this fits together)

You are not writing “a paper.”

You are building:

A coordinated research system with multiple publishable outputs sharing a common dataset

Three Tiers of Papers Tier 1 — Foundational (Theory + Framework) defines the problem introduces your model Tier 2 — Measurement + Mechanism operationalizes the system introduces metrics and constructs Tier 3 — Empirical Studies tests hypotheses using Penwright data 2. Tier 1 — Foundational Papers 📘 Paper 1 — AI-Augmented Authorship and Human Skill Development Purpose

Establish:

the problem (AI deskilling risk) the opportunity (developmental systems) Penwright as a model Core Contribution

Reframes AI writing from generation → human capability development

Source Domains phenomenology of skill attention theory cognitive apprenticeship feminist epistemology Data Required

None (theoretical + synthesis)

Status

✅ Already drafted (technical + public versions)

📘 Paper 2 — Corpus Control and Epistemic Agency in AI Systems Purpose

Formalize:

the epistemic problem of LLMs the role of corpus control Core Contribution

Introduces epistemic control layer as a design principle

Research Anchors standpoint theory epistemic injustice AI bias literature Penwright Tie-In corpus selection system attribution visibility source integration Data Opportunity compare writing outputs under: default corpus curated corpus 📘 Paper 3 — Human–AI Writing as a Structured Cognitive Process Purpose

Replace:

prompt-based models

With:

structured authorship workflows Core Contribution

Defines the Authorship Packet Model

Research Anchors transactive memory translation theory cognitive load theory Penwright Tie-In packet builder retrieval layer structured drafting 3. Tier 2 — Measurement & Mechanism Papers 📗 Paper 4 — A Measurement Framework for AI-Augmented Writing Skill Development Purpose

Formalize:

your measurement system (what we just built) Core Contribution

First multi-dimensional framework for measuring writing skill in AI environments

Key Constructs Writing Quality Index Independence Index Integration Index Metacognitive Index Data Source Penwright telemetry Output framework paper potential standard for others 📗 Paper 5 — Dependency and Independence in AI-Assisted Writing Purpose

Quantify:

when AI helps vs harms Core Contribution

Identifies dependency thresholds

Hypotheses early AI use helps excessive reliance reduces independent capability Data % corpus reliance independent writing performance Penwright Features Used constraint mode independent writing gate 📗 Paper 6 — Learning Loops in AI-Augmented Writing Systems Purpose

Evaluate:

whether structured feedback improves skill Core Contribution

Defines learning loop architecture

Data performance before/after interventions pattern recognition improvements Penwright Features Used teach moments practice mode reflection layer 📗 Paper 7 — Genre-Specific Effects of AI Writing Systems Purpose

Compare:

nonfiction vs memoir vs fiction Core Contribution

Shows that AI effects are genre-dependent

Hypotheses Genre Risk Nonfiction shallow argument Memoir emotional flattening Fiction generic narrative Data genre-tagged sessions skill progression by type 4. Tier 3 — Empirical Studies 📊 Paper 8 — Longitudinal Effects of AI-Assisted Writing on Skill Development Purpose

Core validation study

Design

Groups:

Penwright users Standard AI users No AI Duration

3–6 months

Output

Does Penwright create better writers?

📊 Paper 9 — Transfer of Writing Skill Outside AI Systems Purpose

Test:

independence Method blind writing tasks no AI allowed Core Contribution

Tests real-world capability

📊 Paper 10 — Cognitive Load and Attention in AI Writing Systems Purpose

Study:

how Penwright affects attention Anchors Stiegler attention ecology Data session duration interruption patterns flow states 📊 Paper 11 — Corpus Composition and Output Diversity Purpose

Test:

impact of corpus selection Core Contribution

Shows how epistemic control shapes output

📊 Paper 12 — Human–AI Working Alliance in Writing Systems Purpose

Study:

trust reliance collaboration quality Anchors working alliance theory psychoanalytic models 5. Cross-Paper Data Strategy Shared Dataset

All papers draw from:

Penwright session data telemetry events writing outputs reflection logs This is critical:

You are building a single data-generating system that feeds multiple papers

6. Product Integration (very important)

Each paper should map to:

Paper Feature Authorship model Packet Builder Corpus control Corpus UI Measurement Control Kernel Learning loops Practice Mode Dependency Constraint Mode Genre Mode switching 7. Publication Strategy Phase 1 (0–3 months) Paper 1 (done) Paper 4 (measurement) Paper 3 (authorship model) Phase 2 (3–6 months) Paper 5 (dependency) Paper 6 (learning loops) Paper 7 (genre effects) Phase 3 (6–12 months) Paper 8 (longitudinal) Paper 9 (transfer) Paper 10–12 (advanced) 8. Strategic Insight (important)

Most people would:

write one paper run one study

You are doing:

A system-level research program with compounding outputs

9. What You Now Have

You now have:

measurement framework ✅ technical theory paper ✅ public narrative ✅ research program map ✅