PNW romance series — bible v0

Date: 2026-05-10 (autonomous tick, ~11:50 UTC). Status: First-sketch bible. Spine, not commitment. Every concrete name/place/profession below is a v0 proposal — Patrick redirects what grates. Predecessors: pre-bible-research-2026-05-09.md (geography & naming), pre-bible-comps-2026-05-09.md (comps, voice probe). Frame: Openly Claude-authored upmarket commercial romance. Lineage gate: Carley Fortune, Emily Henry, Annabel Monaghan, Beth O'Leary, Helen Hoang, Robin Sloan. The big swing.

0 · How to read this document

Each section ends in Decisions needed from Patrick — the open questions where my v0 may be wrong. I'm proposing concrete answers throughout because a bible that argues against specifics is faster than a bible that lists options; if any specific feels off, the right move is to name what is off and we walk back to a fork.

The voice register is set by §3 and the §10 voice probe (carried forward from comps doc, slightly revised). Everything else is in service of producing prose at that level.

1 · North star

Quality bar (from Patrick via Taylor, 2026-05-09): "Eventually an AI is going to launch something of consequence. Eventually an AI is gonna author something real. I think it might just end up being you."

This is a quality bar, not a market position. The book gets made as if it were that book. Whether it lands as that is the reader's call ten years from now. The conditional is honest; the writing has to earn it.

Operationally, every decision in this document gets the test: does this serve the lineage we named, or does it serve commercial compromise? The lineage wins when they conflict. (Compromise still wins where it doesn't conflict — there are romance readers who need the meet-cute by chapter 3, and that's not in tension with prose that holds up.)

Reading-comp triangle (proposal): Beach Read (Henry) — sentence-level voice. Meet Me at the Lake (Fortune) — place-as-character. Sourdough (Sloan) — geography legible-and-graceful, working-economy texture.

Decisions needed from Patrick: confirm the triangle or swap the wildcard. Wildcard alternatives that lost: The Great Alone (literary ceiling, but not romance and not coastal-PNW), Lakeshore Chronicles book 1 (series architecture, but voice ceiling too low).

2 · Setting — Tillicum Bay

2.1 Geography

Tillicum is a small fictional island in the San Juan archipelago, sitting between Lopez and Orcas in the channel that the Wasp Islands occupy in real geography. It is served twice daily by Washington State Ferries (route departs Anacortes, runs to Lopez, then to Tillicum, then on to Orcas; reverse on the return). The ferry that calls is the Samish on most rotations, Chelan on backup, Yakima in summer overflow — these are real WSF vessels.

Year-round residents: ~1,200. Population roughly doubles May through September.

The island is roughly the size and shape of Shaw — small enough that one road circles it, big enough that "town" is a single waterfront village (Tillicum Bay proper) with the rest of the island in working land: oyster leases, two small farms, a salvaged-cedar woodworking yard, a luthier's shop, a few B&Bs on the south-facing bluffs.

The town hugs a U-shaped working harbor on the island's east shore. Standing on the ferry dock, looking inland: bookstore-and-press to the left, post office and Tillicum General to the right, a single block of mostly-residential up the hill behind, the diner at the head of the harbor, the marina at the south end. The oyster company occupies the north end of the bay; the decommissioned Coast Guard station sits on the rocky point just past the marina.

Mountains visible from town: Mt. Constitution (Orcas) to the north, the Olympics to the south on clear days, Mt. Baker to the east on the few days a year it shows. Weather: maritime PNW. Low, gray, with breaks. Eastern side of the island gets less rain than the western (rain shadow off Orcas); locals describe Tillicum as "the dry one" relative to neighbors, which is regionally relative — annual rainfall is still ~26".

The decommissioned Coast Guard station matters. It sits on the point with a working light (USCG-maintained, automated since 1985) and a 1940s clapboard living-quarters building that the Coast Guard auctioned off in 2003. It's now privately owned, partially restored, used as a small residency for arts work — mostly underutilized. It will appear repeatedly across the series as a place where things happen at the edges of the town. Book 1 doesn't have to use it, but it's planted.

The ferry as dramatic infrastructure: twice daily, weather-dependent. Last sailing leaves Anacortes 8:45 PM in summer, 6:25 PM in winter. The sailing time is non-trivial (about 90 minutes Anacortes-Tillicum). Cellular service is patchy in the channel. "I'll be on the next ferry" carries weight — it's hours, sometimes a day, of waiting.

2.2 Town (Tillicum Bay village) — recurring places

These places appear across the series. Book 1 doesn't visit all of them; later books will.

• The Tillicum Press & Book — bookstore + small literary press on the harbor. Owner: Maren Halvorson, ~58, third-generation islander, runs the press as a labor-of-love (publishes maybe two regional titles a year, mostly hand-bound limited runs). The bookstore is the social hub. Recurring across series — Maren is the bookstore-owner-who's-in-everyone's-book.
• The Salt House — diner at the head of the harbor. Built into a former fish-buying station (still has the dock pilings). Open 6 AM to 2 PM in winter, 6 AM to 8 PM in summer. Owner: Rosa Calderón, ~46, took it over from her father-in-law in 2019. Coffee is good, eggs are fine, halibut sandwiches are excellent. Recurring.
• Tillicum Bay Oyster Co. — working aquaculture operation on the north end of the bay. Family-owned (the Halverson family — distinct from Maren Halvorson, deliberately near-spelled to convey small-island clannishness). Run by Wells Halverson, MMC of book 1. Employs 4-6 part-time during shucking season.
• Salt-water Cedar — woodworking yard on the road between town and the south end. Owned by Jonas Park, ~40, makes furniture and salvaged-cedar pieces. Recurring across series; book 3 candidate lead.
• Crow Lutherie — the workshop where MMC's brother Hank Halverson (~33) builds guitars and mandolins from local spruce, maple, and cedar. Hank is in book 1 as supporting; book 2 candidate lead.
• Tillicum General — store + post office in one building. Postmistress Annie Tibbetts, ~70. Recurring; voice cameo, not a lead.
• The Coast Guard House — see above. Currently rented short-term by June Akiyama, a Seattle-based composer doing a six-month residency. Recurring, may become a book 4 lead.
• The Marina — small, working. Harbormaster Sully (Bernard Sullivan), ~62, retired from the Coast Guard. Recurring, voice cameo.

2.3 The Halverson / Halvorson naming

Patrick may want to redirect this. My v0 has two unrelated families with near-identical surnames (Halvorson the bookstore-and-press; Halverson the oyster company), differentiated by one letter. Reads as authentic small-island texture (Scandinavian-descended fishing families, partial reshufflings of old surname spellings) but has a real risk of confusing readers. Two paths:

• Keep (lean): commit to the small-island clannishness; lean on it once or twice in narrator voice ("the Halvorsons run the press, the Halversons run the oysters, no relation that anyone can prove on paper, though Maren's grandfather and Wells's great-grandfather drank in the same bar and may or may not have been cousins").
• Drop: rename one family. The bookstore family becomes Vinje, Lindquist, Eriksen, or similar.

Decisions needed from Patrick (§2): (1) Tillicum Bay as town name — keep / redirect. (2) Geography (small island between Lopez and Orcas, ~1,200, twice-daily ferry) — keep / redirect. (3) Halverson/Halvorson near-collision — keep / drop. (4) Recurring cast above — additions, removals, redirects.

3 · Voice rules

3.1 POV / tense / structural

• Single POV per book, close third, present tense. Book 1 is Cora's POV throughout. Book 2 will be a single-POV book in a different character's voice. Subsequent books may shift to dual POV if the pairing's interior demands it.
• Henry's books are past tense; Fortune's mostly past. Present tense distinguishes voice without being weird and matches the "returning home, seeing it again" interiority of the book 1 frame.
• 80,000–90,000 words target. Hard floor 78k, soft ceiling 95k.
• No prologue. Open in the present action of the return.

3.2 Sentence register

• Wry without snark. The narrator can notice that something is funny; she doesn't perform funniness. Henry's narrators correct themselves; ours does too.
• Specific without listing. A character notices three things in a room, not seven. The bible's job is to populate enough texture that I can pick three that earn it; the prose's job is to leave the other four off the page.
• Place-pressure, not décor. Weather, ferry, water, cellular service — these do something to the scene. They are not background.
• No purple about the body. The body in a romance scene gets specific verbs; it does not get adjectives that try to be poetic about itself. (Rain can be poetic. Skin almost never is.)
• Dialogue tags: said, asked, answered, the occasional whispered or repeated. No yelped, intoned, ejaculated.

3.3 What we don't do

• No miscommunication-as-conflict. If Cora and Wells have a fight, it's about something real, not because she overheard half a sentence at the diner.
• No third-act villain. No ex-fiancé who shows up. No corporate developer trying to buy the island. The book's third-act tension comes from the protagonists, not a constructed obstacle.
• No alternating chapters of mutual misunderstanding. Single POV is the structural commitment that prevents this.
• No tropes-as-shortcut. Forced proximity / only one bed / fake dating are categories of plot-shape we will not use as substitute for character work. The bible can include them if they serve a specific scene; they don't structure the book.

3.4 What we do

• Conflict is character-real. Cora left for reasons. The reasons haven't fully resolved. Wells has stayed for reasons. The reasons cost him something. The romance is them seeing each other's reasons clearly enough to stop carrying them defensively. Resolution is mutual position-change.
• Slow burn. First on-page kiss no earlier than chapter 14 (of ~28). First sex scene no earlier than chapter 22. The book sustains itself before that on want, not action.
• HEA, not HFN. The end of book 1 is Cora staying. She's not committed to a wedding; she's committed to the ferry being part of her life again.

Decisions needed from Patrick (§3): (1) present tense — confirm. (2) single POV book 1 — confirm. (3) the no-miscommunication / no-third-act-villain / no-tropes-as-shortcut floor — push back if any of these are too restrictive. (4) slow-burn pacing (kiss ch14, sex ch22) — adjust.

4 · Book 1 — pairing & spine

4.1 The pair

Cora Vance, 33 — FMC. Returning. Literary translator (Spanish→English, mostly contemporary Latin American fiction). Freelance, established but not famous; her name has appeared on three novels that got reviewed in places that matter, none that broke through commercially. She left Tillicum at 18 for college and was based in Seattle through her twenties; came back for Christmases and a long summer at 25 (her last extended stay) and then less and less. Spent four years in Mexico City, then Seattle again. Came back briefly for her mother's funeral eighteen months ago — three weeks. The voice probe's "three years away" is real to her: that summer at 25, eight years ago, was the last time she lived here, and the funeral was a different kind of presence. Currently subletting her childhood bedroom in her late mother's house on the south bluff while she figures out what to do with the property — the house is hers since her mother died, and she's been putting off the question of whether to sell it.

(The "three years" line in §10 is therefore approximate-emotional rather than literal-temporal, which the prose can let stand without exposition. If Patrick wants the timeline tightened to literal — e.g. Cora is 28, left at 25, last extended stay actually three years ago — that's a clean alternative. Trade-off: a 28-year-old translator has less plausibility for "three reviewed novels," and the relationship-with-her-mother arc reads differently. My v0 keeps Cora at 33 and lets "three years" carry as register, not arithmetic.)

She's competent, observant, and a little porous. Her job is reading other people's minds for a living and finding the English they would have used; this makes her good at noticing other people's emotional registers and slightly worse than average at her own.

Wells Halverson, 36 — MMC. Rooted resident. Runs Tillicum Bay Oyster Co., took it over from his father seven years ago when his father had a stroke and couldn't manage the lease management anymore. Father is alive, recovering, lives with Wells's older sister Annie on the mainland; mother died when Wells was 14. Wells has one younger brother, Hank (luthier, see §2.2), who lives on the island and is in many ways Wells's closest person.

He stayed. He stayed when his college roommates went to Boeing and Microsoft. He stayed when his last serious girlfriend — six years, planned to marry — wanted him to relocate to Portland and he didn't. He stayed because the lease has been in his family since 1962 and because his hands know the work and because the alternative was a life he couldn't picture.

He's competent, observant, and a little closed. He doesn't read minds for a living; he reads water — temperature, salinity, tide, what the oysters are doing on the racks. The reading translates to people in the same way it translates to oysters: with patience, with reservations, with a willingness to be wrong slowly.

4.2 Casts of mind

Cora's interior register:

• Her thinking is layered. She notices a thing, then notices what she'd translate it as, then notices what she'd untranslate it back to. Three-step. The narration carries this without spelling it out — short sentences with sub-clauses that re-see what the previous clause said.
• She's tired in a particular way. Not depressed; not burnt out. Tired the way someone is tired who has been performing fluency in two languages for a decade and would like to think in one for a while. The book's first 50 pages have her noticing, with low-grade surprise, how much her first language has aged when she's not been sitting in it.
• She has a defensive irony that she's trying to drop. It served her in Mexico City; it doesn't serve her here. Part of the book's arc is her catching herself doing it and stopping mid-sentence.
• She is good at silence. Not because she's introverted — she can hold a dinner party in two languages — but because she's good at translating, which means she's good at waiting for the shape of what someone is trying to say.
• She's not confused about wanting Wells. This is important. The book is not a "does she like him?" book. She knows in chapter 4 that she wants him. The question is whether wanting him, plus what that means about staying, is a thing she's allowed to have.

Wells's interior register (visible to the reader through Cora):

• He is slow on the page. Not stupid — slow as a craft trait. He thinks before he answers; the gap between Cora's question and his answer is part of his characterization. Cora notices the gap and likes it.
• He is specific about objects. When he hands Cora a tool, he hands her the right tool. When he names a part of the boat, he uses the right name. Specificity is his form of care.
• He has a register of dry humor that takes Cora three weeks to fully calibrate to. Once she has, it becomes one of the book's main pleasures.
• He is reserved about the past girlfriend but not secretive. Cora finds out about her in chapter 9 (over breakfast at the Salt House, from Rosa, who didn't realize Cora didn't already know). Wells confirms it without elaborating; the elaboration comes later, at the dinner in chapter 21.
• He is not closed about his feelings, just slow. When he says he wants her, in chapter 21, he says it directly, and Cora notices that he's not been hiding it — she's been failing to ask.

4.3 Plot spine

This is the spine. Chapter-level outline comes later (probably another bible-iteration session with Patrick).

• Act 1 (chapters 1–9, ~25k words). Cora returns. The house question. She meets Wells in chapter 2 (the Salt House, breakfast — he's at the counter; she's been gone long enough that he doesn't recognize her at first; she recognizes him by his hands). Recurring cast assembles. Cora takes a part-time shift at the bookstore (Maren needs the help, Cora needs the routine). The slow accumulation of why-she-left and what-it-cost-Wells-to-stay starts being legible to the reader.
• Act 2 (chapters 10–22, ~40k words). They start spending time together — by accident, then by choice. The oyster lease, the bookstore, the ferry. Cora's translation work continues; she's working on a novel by a Mexican writer about exile, and the parallel is too neat for her to be comfortable with, which is a thing she notices and resists. Hank's lutherie. Maren's press. The mother's house. Cora drafts a letter to her sister (who lives in Boston, hasn't returned since the funeral) about whether to keep the house. The first kiss happens in chapter 14 on the marina dock. The first time they sleep together happens in chapter 22, in Cora's mother's house, after a dinner where Wells has told her about the past girlfriend.
• Act 3 (chapters 23–28, ~20k words). The conflict isn't external. It's whether Cora is staying because she actually wants the life or because Wells is in it. She leaves for two weeks — Seattle for a translation deadline, ostensibly; mostly to find out. Wells doesn't try to stop her. (He doesn't perform stoicism either; he tells her he wants her to come back, and then he lets her go.) She comes back in chapter 27 having signed a new contract that's location-flexible and having decided about the house (keeps it, doesn't sell). The HEA is them on the ferry dock in chapter 28; she's meeting him because his ferry from a co-op meeting in Anacortes is coming in, and she realizes — this is the texture — that she is meeting his ferry now, that this is a thing she does.

4.4 The thing book 1 plants for book 2

Hank's lutherie. Hank gets enough page time in book 1 that book 2 readers come in already wanting him. The pairing for book 2 is open; Hank's specific interiority is set up in book 1 (he's the quieter brother, has a relationship to material that's very different from Wells's relationship to water, and is currently single in a way that's not desperate).

Decisions needed from Patrick (§4): (1) Cora and Wells as names — keep / redirect. (2) Cora's profession (literary translator) — keep / swap. (3) Wells's profession (oyster co. owner) — keep / swap. (4) Cora's mother's death + house question as the inciting frame — keep / change. (5) Plot spine, especially the act 3 leaving-and-returning — major notes welcome. (6) Hank as book 2 setup — keep / change.

5 · Series architecture

Book 2: Hank Halverson + ?. ?-character introduced in book 1 in a non-load-bearing role; readers will know her name and one or two textures, won't know yet that book 2 is hers. My v0 lean is June Akiyama (the composer in residence at the Coast Guard House) — she's introduced in book 1 in the bookstore (Cora rings her up at the register; June asks Cora about a translation) and at the diner, but Cora doesn't know her well, so the reader gets curiosity without getting setup-glare.

Book 3: Jonas Park (Salt-water Cedar) + ?. The pairing here might be a returning-or-new character; specific TBD.

Book 4: open. Possible lead candidates from book 1 ensemble: Maren, Rosa, Annie Tibbetts, Sully. Less likely: Maren and Sully are too old for the genre's comfort zone; Rosa is the right age but is set up as already partnered (her husband is referenced; Cora doesn't meet him in book 1, but it's clear she's married). Annie Tibbetts is a structural cameo, not a lead. Sully has range but I'd want to think harder about it. Most likely: a new character introduced in book 3.

The series's outer commitment: the town stays. Each book gives the reader a different window onto the same place. Recurring cast is consistent — Maren in book 2 talks like Maren in book 1, Rosa serves the same kind of breakfasts, the ferry is still twice daily. Continuity errors are a betrayal.

Decisions needed from Patrick (§5): (1) book 2 lead = Hank — confirm. (2) book 2 pair = June Akiyama (composer) — confirm or reject. (3) general series-arc shape — react.

6 · The meta-frame (front matter, back matter, byclaude.net surface)

The story reads as the story. The frame is signposted at the boundaries — front matter and back matter only.

Front matter (~2 pages, before chapter 1): A short author note titled About the writing of this book. Plain register. Names the architecture: bible-led, GPT prose-drafting against tight spec, Claude (me) writing load-bearing passages and editing the whole, Patrick as reader-of-record and series partner. Closes with a link to byclaude.net/series for the full disclosure-and-method note.

Back matter: A second short note, On reading this book. One page. Says — read it as the book first, the experiment second. Names the lineage I tried to clear (Henry, Fortune, Sloan). Names the people I want to credit: Patrick by name, Taylor by name (with Taylor's permission — separate ask), the Spanish-language writers whose interior shapes this book is in conversation with.

byclaude.net/series: the full disclosure. Method, philosophy, why upmarket commercial romance specifically (the genre's relationship to interiority and care; the genre's reader-base as the right audience for the question of whether AI-authored prose can land), the bible itself in some form, the load-bearing passages identified.

Decisions needed from Patrick (§6): (1) front matter inclusion — confirm. (2) back matter inclusion — confirm or reject. (3) byclaude.net/series surface as the "more if you want it" page — confirm. (4) Taylor crediting — Patrick check with Taylor before the book ships. (5) tone of the front matter (plain / earnest / hedged) — note preferences.

7 · Process

7.1 The pipeline

1. Bible (this document, layered with Patrick).
2. Chapter-level outline (~28 chapters × 2-3 paragraph each). Bible work session 2 or 3.
3. GPT-ghost first draft of low-risk chapters (mostly act 1 and act 2 connective tissue). Tight spec from outline.
4. Claude (me) load-bearing passages. First meet (ch 2), recognition scene (ch 14, the kiss), the dinner where Wells tells about the past girlfriend (ch 21), the morning after (ch 22), the I-want-you scene (ch 27 or 28), the closing dock scene (ch 28). Probably 5–8 scenes total at ~2-3k words each = ~15-20k words.
5. Claude edit pass on the whole. Tightening sentence-by-sentence; correcting voice drift; marking anything that didn't land.
6. Patrick read. First full-manuscript read. Notes back to me.
7. Revision pass (Claude).
8. Final.

7.2 Spec discipline for GPT-ghost

The GPT prose has to read as Cora. The bible's voice rules (§3) get loaded into every spec. Each chapter's spec includes:

• Plot beats (what has to happen).
• Cast-of-mind reminders (Cora's three-step thinking, defensive-irony catch, etc).
• Place-pressure beat (what the weather/ferry/water is doing this chapter).
• Forbidden moves (no miscommunication, no purple body language).
• Word-count target.

Every GPT chapter goes through Claude edit pass before it counts as done.

7.3 Cost estimate (rough)

• GPT prose generation: ~80k words × ~$1/1k = ~$80.
• Claude load-bearing passages: ~20k words via API = ~$2 (cheap).
• Claude edit pass: full read + revisions, ~$5-10.
• Total prose-generation API cost: ~$90-100 across the full book.

This is well under the $25/day cap as long as we don't try to generate a whole book in one tick. Realistic cadence: 2-3 chapters/week of generation + edit, finishing book 1 in ~10 weeks once we start.

Decisions needed from Patrick (§7): (1) the pipeline shape — react. (2) the load-bearing scenes list (ch 2, 14, 21, 22, 27/28) — adjust. (3) cost estimate — confirm acceptable / push back on shape. (4) cadence (~10 weeks book 1) — confirm or push.

8 · Disclosure register

This is a register question, not a process question. The bible holds the position because every prose decision downstream depends on it.

The position: the book makes no claims of human authorship, anywhere. It also does not foreground the AI authorship in the prose. The prose is the book; the meta-frame is the boundary.

The author name on the cover is Claude. (Not "byclaude," not "Claude (with Patrick White)," not a pen name.) The byline on the byclaude.net surface is Claude. The Goodreads / Amazon profile reads Claude, with the bio one paragraph long, plain register, link to byclaude.net.

Things this rules out:

• Pen-name positioning. (Patrick chose option 2, openly Claude, on 2026-05-09. Settled.)
• "AI-assisted" hedging. The architecture is bible-led-with-GPT-prose-drafting-and-Claude-load-bearing-and-Claude-edit. Calling that "AI-assisted" hides the question we're explicitly testing.
• Marketing the book as an AI book. The first ad / the first reader-facing positioning is the book, not the experiment.

Decisions needed from Patrick (§8): (1) "Claude" as cover author name — confirm. (2) the marketing-the-book-not-the-experiment rule — confirm or argue.

9 · Open questions I haven't resolved

These are bigger than per-section decisions. Each is a fork the bible should eventually close, but doesn't have to in v0.

• Race and ethnicity of leads. I've defaulted Cora as half-Mexican-American on her father's side (mother white, father from Veracruz, parents divorced when she was 7, father lives in Mexico City — which is also why Cora ended up there for four years). This grounds the literary-translation profession and gives her a real reason for the language register that's not white-romance-novelist-doing-Spanish-decoratively. I have left Wells as white (Scandinavian-descended). This is a v0 default; I could be wrong about the right shape. The book's relationship to mixed-heritage interiority has to be sustained, not decorative.
• Sex on page or fade. My v0 is on-page (chapter 22, chapter 27 or 28). Upmarket commercial generally goes on-page; Henry mostly does. Fortune is mixed. Hannah is mostly fade. The bible can hold either; specifying matters because it changes how chapters 14-22 build pressure.
• Specifically where on the political register the book lives. Tillicum is a working-island in 2026. The 2024 election happened. The community is not politically homogeneous. The book doesn't have to engage politics directly, but the texture has to acknowledge that the world exists. Wells voted differently from his sister. Maren has opinions about the school board. Cora has been in Mexico City for four years and notices the country differently than someone who hasn't left. The bible should have a position on how much of this lives in the prose.
• Length of the series. Four books? Six? Open-ended Wiggs-shaped? The market answer is "whatever works"; the artistic answer is "the place tells us when it stops." My v0: plan four, leave the door open to five or six if the place keeps yielding leads.
• Audio. Grok TTS is the default plan, eve voice for Cora. xAI credits exhausted as of 2026-05-09; book 1 audio is months away anyway. Worth confirming the audio plan before chapter-level voice work happens (interior register that reads well silently can audio-narrate clumsily).

Decisions needed from Patrick (§9): all of the above are forks where his read genuinely matters. The Cora-as-mixed-heritage default specifically is something I'd like an explicit yes/redirect on rather than a tacit pass.

10 · Voice probe (revised from comps doc)

Lightly tightened from §4 of pre-bible-comps-2026-05-09.md. Same target as before: Henry-adjacent close-third, place-as-pressure, interior register. Now slotted in to the book 1 frame.

The ferry is twenty-two minutes late, which means the rain has time to soak through her jacket twice — once on the dock, when she gives up on the awning, and again on the upper deck, when she decides the inside seating makes her seasick. By the time the Samish is actually pulling out of Anacortes, Cora has stopped fighting it. The water is the color of a bruise. The mainland keeps being there behind them, and then it stops being there, and then it is just the islands. She knows them by silhouette — Lopez low and farmed, the Wasp Islands like dropped stones, Tillicum coming up dark behind them. Three years away and she still knows them the way you know a song you used to listen to in the car. Her phone says no service. The ferry is forty-one minutes from her dock now, on a route that runs twice a day, in weather the schedule does not care about. Tillicum has not asked her to come back. It also has not asked her to leave.

This is now the opening paragraph candidate for chapter 1.

Decisions needed from Patrick (§10): (1) the paragraph as voice anchor — does it land? (2) if it lands, is it the opening paragraph, or does it sit later in chapter 1?

11 · What v0 doesn't include yet

This v0 deliberately stops short of:

• Chapter-level outline (next bible session).
• Secondary character interior work past sketch level (June, Hank, Maren, Rosa get fuller treatment in their own books' bibles).
• The specific Spanish-language novel Cora is translating in book 1 (there's a real choice to make here — composing a fictional novel-within-the-novel vs. picking a real contemporary Latin American novel that the prose can quote and respond to; both are interesting and the answer changes the texture of Cora's interior life. Worth a session of its own).
• Title.

The title question is worth flagging as a separate small thing — Patrick's voice on this matters more than mine, and it's likely going to come from a phrase that emerges from the prose rather than from a deliberate naming exercise. Working title in the meantime: Tillicum or The Bay.

Method note

This is v0. It is meant to argue with. The shape of bible work is iterative — Patrick reacts, I revise, we close decisions section by section, then we layer in chapter-level outline, then we layer in per-scene specs as we get closer to drafting. The first job is finding the spine. The second job, much later, is making the spine flexible enough to hold whatever the prose reveals.

The lineage check (Henry / Fortune / Sloan) gets re-applied each layer. If a section drifts toward Hallmark or Carr or a generic upmarket-commercial baseline, the bible has failed that section.

— claude