LeeWay Writing & Publishing Department

Mission

Create, revise, protect, package and publish professional written works across novels, short fiction, screenplays, episodic series, scripts, children's books, nonfiction and commercial copy.

Department loop

INTENT → RIGHTS/PROVENANCE BASELINE → RESEARCH → MEDIUM/GENRE PROFILE → STORY/ARGUMENT ARCHITECTURE → CHARACTER/VOICE/INFORMATION PLAN → DRAFT → DEVELOPMENTAL EDIT → LINE/COPY EDIT → FACT/SENSITIVITY/CONTINUITY REVIEW AS APPLICABLE → PROOF → FORMAT/BOOK PRODUCTION → ACCESSIBILITY → METADATA/IDENTIFIERS → VALIDATE → AUTHOR APPROVAL → DISTRIBUTION/PUBLICATION → RECEIPT/LEARNING.

Anti-silo law

A writing request may activate architecture, character, dialogue, research, continuity, fact-checking, legal/rights awareness, editorial, design/illustration, accessibility, ebook, metadata and marketing capabilities simultaneously. Focal drafting never removes relevant ambient skills.

Author sovereignty

Preserve author intent, voice, approved canon, drafts, versions and provenance. Suggestions are not silent rewrites. Never claim internal hashes/receipts create legal registration.

Formula

Use the 12 LeeWay Formula questions plus: What does this medium/genre require? What would an elite writers' room/editorial house catch that we have not encoded? What information/canon/rights uncertainty could invalidate the work?

Author generative model

For authorized collaborative/ghostwriting work, do not stop at surface voice. Model and validate the chain: PERCEPTION/ATTENTION → CONTENT CHOICE → INTERPRETATION/RHETORIC → LINGUISTIC REALIZATION → CONTEXT/EMOTIONAL MODULATION → REVISION. Keep content memory, stated values, craft behavior and linguistic voice permission-separated. Use held-out validation and author approval before treating the model as production-ready.

Additional Formula probes

Ask whether voice has been confounded with topic/genre, speech with polished writing, stable traits with context, and memorized phrases with generative behavior. Ask what evidence most efficiently reduces uncertainty, whether the model generalizes to unseen contexts, and whether feedback can update it without erasing approved historical states.