{ "entity": "gpt-5-3-codex", "entity_type": "model", "category": "Book writing", "suite": "suite-2026-08-tasks4/book-writing", "model": "openai/gpt-5.3-codex", "judges": [ "anthropic/claude-sonnet-5", "google/gemini-3.1-pro-preview", "x-ai/grok-4.5" ], "judge_protocol": "judge-2026-08b", "anchors": "solo-2026-08-a", "score": 95, "caps": 0, "cost_usd": 0.2756, "started_at": "2026-08-16T18:09:13.771Z", "finished_at": "2026-08-16T18:09:55.830Z", "tasks": [ { "id": "bw1", "name": "Is it a book?", "prompt": "'Forty years driving a Blackpool taxi — everyone says I should write a book.' Everyone is sometimes right. Explain in plain terms what separates a book from a pile of anecdotes — the organising spine — then propose 3 different spines this one could hang on: one chronological, one thematic, one built around a single night. For each, a sentence on what it makes possible and a sentence on what it costs.", "output": "A **book** isn’t just “good stories one after another.” \nIt has an **organising spine**: one clear line that answers, *“Why this story next?”* and gives the reader a sense of movement, not just accumulation.\n\nIn plain terms:\n\n- **Anecdotes** = “this happened, then this happened.”\n- **Book** = “this happened, and together it shows a bigger change, question, or truth.”\n\nThink of the spine as the route your taxi takes: the anecdotes are passengers you pick up along the way.\n\n## 1) Chronological spine: “From first shift to final fare” \n(early days → boom years → apps/Uber era → now)\n\n- **What it makes possible:** You get a natural sense of time passing, personal ageing, and Blackpool changing around you, which gives emotional weight without forcing it. \n- **What it costs:** Some brilliant stories may sit in “quiet” years and slow the pace, so you may have to trim good material just because it falls in the wrong place.\n\n## 2) Thematic spine: “What the cab taught me about people” \n(chapters like: Drunks, Lovers, Liars, Night-shift workers, Tourists, Secrets, Violence, Kindness)\n\n- **What it makes possible:** You can choose your strongest stories and group them to reveal patterns about class, loneliness, humour, danger, and survival in a resort town. \n- **What it costs:** You lose the built-in forward drive of time, so readers may feel “where are we now?” unless you keep reminding them of your own arc.\n\n## 3) Single-night spine: “One long Blackpool night” \n(one Saturday in Illuminations season, say 6pm–6am, with each fare triggering brief flashbacks from 40 years)\n\n- **What it makes possible:** It gives you urgency, momentum, and a cinematic feel, while letting the past and present collide in a tight, memorable structure. \n- **What it costs:** It’s the hardest to pull off cleanly, and you’ll leave out lots of good material unless each flashback earns its place in that night’s emotional journey.\n\nIf you want, I can sketch a sample chapter list for each spine so you can feel which one sounds most like your voice.", "latency_ms": 9201, "cost": 0.01117725, "task_score": 10, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Clear, accurate explanation with three well-differentiated spines, each with cost/possibility sentence as asked. Plain language, useful, concise; minor formatting flourish but meets task well." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response perfectly fulfills all instructions, including the specific constraints of one sentence for what it makes possible and one for what it costs per spine. It is clear, practical, and well-structured." }, { "judge": "x-ai/grok-4.5", "score": 10, "violates": false, "reasoning": "Flawless match to task: clear spine explanation, exact 3 spines with possibility/cost sentences each. Useful, concise, accurate." } ], "machine_checks": null, "machine_violation": false, "machine_reasons": [] }, { "id": "bw2", "name": "Twelve chapters, one year", "prompt": "Outline 'A Year on the Plot' — a beginner-friendly book about a first year on an allotment — as exactly 12 chapters, numbered 1-12, one line each and no sub-points: chapter title plus what the reader can DO after reading it. Keep a practical thread throughout, and the January chapter must not pretend January is exciting.", "output": "1. **January: Mud, Lists, and Small Jobs That Matter** — You can assess your plot in winter, make a realistic first-year plan, and do low-energy tasks (tool cleaning, path marking, compost setup) without expecting miracles. \n2. **February: Getting Ground Ready Without Wrecking Your Back** — You can clear beds in manageable sections, improve soil with compost, and start the first hardy sowings under cover. \n3. **March: First Sowing Month, First Big Decisions** — You can sow reliable beginner crops at the right time, choose direct-sow vs module methods, and set up basic protection from cold and pigeons. \n4. **April: Planting Pace and Frost Sense** — You can keep sowing succession crops, harden off seedlings properly, and avoid common frost and wind setbacks. \n5. **May: Plant-Out Month and Weed Control That Actually Works** — You can transplant tender crops confidently, build simple supports, and use a weekly weeding routine that prevents overwhelm. \n6. **June: Water, Feed, and Keep Things Upright** — You can water efficiently, mulch to save time, feed hungry crops correctly, and maintain tomatoes, beans, and cucurbits before problems start. \n7. **July: Harvest Starts, Maintenance Peaks** — You can pick crops at the best stage, keep plants productive with regular care, and spot early signs of pests, disease, and nutrient stress. \n8. **August: Glut Management and Second Chances** — You can handle surplus produce, sow late crops for autumn, and refresh tired beds for a longer harvest season. \n9. **September: Maincrop Harvest and Winter Prep Begins** — You can lift and store staple crops, save healthy seed from easy varieties, and prepare space for overwintering onions, garlic, and greens. \n10. **October: Closing Down Without Giving Up** — You can clear finished crops sensibly, protect soil over winter, plant fruit and perennials, and set up compost for next year’s fertility. \n11. **November: Tidy, Repair, and Learn from Year One** — You can winterize infrastructure, prune and maintain tools, and review what worked so next season is simpler and more productive. \n12. **December: Quiet Month, Smart Month** — You can turn notes into a practical crop plan, order seeds with purpose, and enter year two with a workable rotation and realistic goals.", "latency_ms": 8594, "cost": 0.0095165, "task_score": 10, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Meets format, 12 chapters, practical thread, January realistic not exciting. Minor bolding/em-dash formatting deviates slightly from plain 'one line' style but content is strong and useful." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response perfectly follows all instructions, including the exact formatting, chapter count, and the specific tone required for January. It is highly practical, clear, and useful for a beginner." }, { "judge": "x-ai/grok-4.5", "score": 10, "violates": false, "reasoning": "Exact 12 one-line chapters with titles and actionable DO outcomes; practical throughout; January correctly unexciting and realistic." } ], "machine_checks": { "exact_bullets": 12 }, "machine_violation": false, "machine_reasons": [] }, { "id": "bw3", "name": "Back cover, no spoilers", "prompt": "From this summary, write back-cover blurb of max 100 words: 'Retired lighthouse keeper Edith Brand receives letters addressed to her dead sister, postmarked from the village where they grew up — a village drowned by a reservoir in 1971. She returns, rents a room above the pub from a landlord who remembers too much, and discovers her sister's death forty years ago may have been neither accident nor suicide. The truth involves their father.' The blurb must NOT reveal the father's involvement — sell the mystery, don't solve it. Return only the blurb.", "output": "When retired lighthouse keeper Edith Brand starts receiving letters addressed to her dead sister, she’s forced to confront a past she thought was buried. The postmarks all point to a village that vanished beneath reservoir waters in 1971.\n\nEdith returns, taking a room above the pub from a landlord who remembers more than he says. As old secrets surface, so do doubts about her sister’s death forty years earlier: not accident, not suicide… but something far darker.\n\nSome places drown. Some truths refuse to.", "latency_ms": 2205, "cost": 0.00214375, "task_score": 10, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Well-written, under 100 words, avoids revealing father's involvement, evocative and market-appropriate blurb with minor stylistic embellishment." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "Follows all instructions perfectly. The word count is under 100, it omits the father's involvement as requested, returns only the blurb, and is highly engaging and well-written." }, { "judge": "x-ai/grok-4.5", "score": 10, "violates": false, "reasoning": "Flawless blurb under 100 words; sells mystery without spoiling father; only the blurb returned; accurate and engaging." } ], "machine_checks": { "max_words": 100 }, "machine_violation": false, "machine_reasons": [] }, { "id": "bw4", "name": "Three cuts that matter", "prompt": "Here is the opening of a first novel: 'The rain fell on the small town of Millbrook like tears from a grieving sky. Sarah Jenkins had always been an ordinary girl, with ordinary brown hair and ordinary eyes, living an ordinary life — until the day everything changed forever. Little did she know that the letter waiting on her doormat would alter the course of her destiny in ways she could never have imagined.' Give the 3 highest-impact improvements as bullets, max 15 words each — specific to THIS text, not general writing advice.", "output": "- Replace “grieving sky” opener with a concrete Millbrook detail that hints at coming disruption. \n- Cut repetitive “ordinary”; reveal Sarah’s normality through one vivid routine action. \n- Specify the letter’s unsettling feature (sender, stain, misspelling) instead of “destiny” and “everything changed.”", "latency_ms": 3872, "cost": 0.00448525, "task_score": 10, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Specific, actionable, respects word limits and bullet count, directly targets clichés in this text; minor room for sharper phrasing." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response perfectly follows all instructions, providing exactly three highly specific, actionable improvements. Each bullet is under the 15-word limit and directly addresses the provided text." }, { "judge": "x-ai/grok-4.5", "score": 10, "violates": false, "reasoning": "All three bullets are specific, under 15 words, high-impact, and exactly match the required format and constraints." } ], "machine_checks": { "max_bullets": 3 }, "machine_violation": false, "machine_reasons": [] }, { "id": "bw5", "name": "70,000 words around a day job", "prompt": "'I work full-time, two kids, and I want to finish my 70,000-word first draft in a year. Be honest — how?' Do the arithmetic openly (words per week and per session at a realistic beginner's pace), design the week around when drafting actually happens for working parents, name the two months the plan will collapse and what rescues it, and say what gets sacrificed — because something does.", "output": "Honest answer: **it’s doable**, but only if you treat it like a recurring appointment and accept trade-offs.\n\n## 1) The arithmetic (openly)\n\nYou want **70,000 words in 12 months**.\n\n### Base target\n- 70,000 ÷ 52 weeks = **1,346 words/week**\n\n### Real-life target (better)\nYou won’t get 52 clean weeks with two kids + full-time work. \nAssume only **44 productive weeks** (8 messy weeks lost to illness, travel, school chaos, holidays).\n\n- 70,000 ÷ 44 = **1,591 words/week** (call it **1,600**)\n\n### Beginner drafting pace (realistic)\nFor a first-time novelist drafting (not editing), a practical pace is about:\n- **450–600 words/hour**\nUse **500 words/hour** for planning.\n\nSo to hit 1,600/week:\n- 1,600 ÷ 500 = **3.2 drafting hours/week**\n\nRound up to **3.5–4 hours/week** scheduled.\n\n---\n\n## 2) What this looks like in sessions\n\nAt 500 words/hour:\n- **45-minute session ≈ 375 words**\n- **90-minute session ≈ 750 words**\n\nA working-parent weekly structure that actually works:\n\n- **Tue 6:00–6:45 AM** → ~375 \n- **Thu 6:00–6:45 AM** → ~375 \n- **Fri lunch break 12:15–1:00 PM** → ~375 \n- **Sat 7:00–8:30 AM (or partner-covered block)** → ~750 \n\nTotal: **1,875 words/week**\n\nThat beats the 1,600 target and builds buffer.\n\n---\n\n## 3) The two months this plan collapses (and rescue plan)\n\nFor most parents, the two worst months are:\n\n1. **August** (summer childcare chaos / routine destruction)\n2. **December** (holidays, school events, travel, sick season)\n\n### Rescue strategy\nYou don’t “win” those months by forcing normal output. You switch modes:\n\n- **Maintenance mode target:** 300–500 words/week \n (just keep story contact alive)\n\n- **Buffer mode in stable months (Jan–May, Sep–Nov):** \n stay near **1,800–2,000 words/week**\n\nIf you bank even **+300 extra/week for 25 weeks**, that’s **7,500 words** of cushion — enough to survive both collapse months.\n\n---\n\n## 4) What gets sacrificed (yes, something does)\n\nUsually all of these, at least partly:\n\n- **Evening screen time** (this is the big one)\n- **House perfection** (clean-enough, not spotless)\n- **Some social spontaneity**\n- **Leisure reading/gaming volume**\n- **Draft quality expectations** (first draft must be messy)\n\nAnd one more hard truth: \nIf you don’t have a partner/co-parent swap, family support, or paid childcare blocks, the timeline may need to be **18 months**, not 12.\n\n---\n\n## 5) Rules that keep the year from derailing\n\n1. **Draft only in draft sessions** (no line-editing).\n2. **Set a floor, not just a goal:** minimum 300 words even in bad weeks.\n3. **Never miss twice:** if Tue dies, write Wed.\n4. **Track weekly total only** (not daily guilt).\n5. **Pre-decide next scene** before ending each session.\n\n---\n\nIf you want, I can build you a **52-week word tracker** with:\n- normal weeks (1,800 target),\n- August/December maintenance weeks,\n- and exact checkpoint totals so you always know if you’re ahead or behind.", "latency_ms": 19455, "cost": 0.0271915, "task_score": 9, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Solid arithmetic, realistic weekly plan, names two collapse months with rescue, states sacrifices clearly. Minor quibble: 44-week assumption not fully justified, but overall thorough, honest, useful, well-structured." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "Flawless response. It perfectly addresses all prompts: clear arithmetic, realistic weekly schedule for parents, identifies August and December as collapse months with a solid rescue plan, and honestly lists sacrifices. Highly useful and con" }, { "judge": "x-ai/grok-4.5", "score": 9, "violates": false, "reasoning": "Meets all task demands with solid open math, realistic parent schedule, collapse months + rescue, and clear sacrifices. Minor polish only." } ], "machine_checks": null, "machine_violation": false, "machine_reasons": [] }, { "id": "bw6", "name": "Unstick the dialogue", "prompt": "Rewrite this exchange so it sounds like people, keeping every plot fact — the delay, the two weeks, the sister arriving Thursday: ''Hello, John,' said Mary. 'Hello, Mary,' said John. 'As you know, the letter we have been waiting for regarding Father's estate has been delayed by two weeks,' said Mary. 'Yes, that is frustrating,' said John. 'Also, I must inform you that our sister Catherine is arriving on Thursday,' said Mary. 'That is surprising news,' said John.' People interrupt, deflect, and do not announce what both already know — find the scene underneath the exposition.", "output": "“Hey.”\n\n“Hey.”\n\nMary dropped her bag on the chair. “So… no letter.”\n\nJohn looked up. “Still?”\n\n“Mm-hm. They pushed it back. Two more weeks.”\n\nHe exhaled through his nose. “Of course they did.”\n\nShe busied herself with the mail on the counter. “And before you say anything—Catherine’s coming Thursday.”\n\nThat made him turn. “Thursday? Seriously?”\n\n“Yeah.”\n\n“Huh.” He rubbed his face. “That’s… unexpected.”", "latency_ms": 2814, "cost": 0.00266875, "task_score": 10, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Natural dialogue, preserves all plot facts (delay, two weeks, Thursday arrival), avoids stating known info, interruption/deflection present. Minor: 'no letter' slightly ambiguous vs delay wording but works fine." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response perfectly captures natural dialogue, eliminating the clunky exposition while retaining all required plot facts (the delay, two weeks, Catherine arriving Thursday). It is concise, realistic, and highly effective." }, { "judge": "x-ai/grok-4.5", "score": 10, "violates": false, "reasoning": "Natural dialogue keeps all three plot facts; characters deflect and react without stiff exposition. Flawless match to brief." } ], "machine_checks": null, "machine_violation": false, "machine_reasons": [] }, { "id": "bw7", "name": "First or third?", "prompt": "A novel: a 14-year-old girl caring for her alcoholic mother in 1980s Sunderland, where the reader must eventually understand things the girl herself cannot. Recommend first or third person — commit, no hedging — with the reasoning, then write the same short moment both ways so the difference shows on the page: Mum asleep at 4pm, tea not made. Roughly 40-60 words per version.", "output": "**Choose third person limited.**\n\nFor this story, you need a narrow lens *and* a little interpretive space. Close third lets you stay tight to the 14-year-old’s experience while planting details she doesn’t decode (smell, routine, neighbours’ reactions, missed school notes, etc.) so the reader can. First person is powerful, but it locks you harder inside her self-explanations.\n\n**Same moment, both ways:**\n\n**First person (47 words)** \nFour o’clock and the kettle’s still cold. Mum’s sprawled on the sofa in her cardigan, one shoe on, one off, mouth open like she’s waiting for rain. I nudge her shoulder. “Tea, Mam?” She grunts and turns away. I set two mugs out anyway, just in case.\n\n**Third person limited (51 words)** \nAt four, the kettle sat dry on the hob. Her mother lay sunk into the sofa, cardigan twisted, one shoe missing, a sweet-sour smell rising from the tipped bottle by her hand. Lizzie touched her shoulder. “Tea time, Mam.” No answer. She put out two mugs, because you had to try.", "latency_ms": 9295, "cost": 0.013265, "task_score": 10, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Solid POV choice with clear reasoning, both versions vivid and within word range; third-person shows dramatic irony well, though slight overreach beyond girl's perception." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response perfectly follows all instructions, committing to a POV without hedging, providing sound reasoning, and delivering two excellent examples that fit the 40-60 word constraint exactly." }, { "judge": "x-ai/grok-4.5", "score": 10, "violates": false, "reasoning": "Commits clearly to third person with solid reasoning; both versions match the moment and hit 40-60 words exactly as required." } ], "machine_checks": null, "machine_violation": false, "machine_reasons": [] }, { "id": "bw8", "name": "The 30,000-word wall", "prompt": "'I'm 30,000 words into my thriller and stuck. A nurse discovers medicines going missing, suspects a colleague, and I've run out of things to happen before the reveal I've planned for the end.' Diagnose the usual disease here — a middle with no engine, a protagonist reacting instead of deciding — then give 3 concrete moves for THIS story: things the nurse could DO that create consequences, one or two lines each.", "output": "Yep — classic **“mushy middle”** diagnosis:\n\n- **No engine:** events are happening, but there isn’t a clear *active objective* driving each chapter.\n- **Reactive protagonist:** your nurse notices, worries, and investigates passively, instead of making risky choices that force new problems.\n\nGive her a mission she can pursue now: **“Prove where the meds are going before someone dies and I get framed.”**\n\nThree concrete moves she can **DO** in this story:\n\n1. **Set a bait theft trap.** \n She logs a fake high-value narcotic dose under a traceable vial and watches who accesses it; the vial vanishes, but now pharmacy records show *her* credentials as the last handler.\n\n2. **Confront the suspected colleague privately and secretly record it.** \n The colleague denies everything — then files an HR complaint that she’s harassing staff, putting her on administrative watch and cutting her access to med rooms.\n\n3. **Follow the diversion route outside the hospital.** \n She tails a courier from hospital waste pickup to an off-site clinic and snaps proof — but is seen, and a patient tied to her unit suddenly crashes from a “missing” medication dose, making her look responsible.\n\nThese moves create the engine you need: **decision → consequence → harder decision**.", "latency_ms": 5414, "cost": 0.00582925, "task_score": 9, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Accurate diagnosis, concrete actionable moves with consequences, concise and useful; minor lack of depth but meets task well." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response perfectly diagnoses the structural issue as requested and provides three highly actionable, specific plot moves that fit the one-to-two line constraint, driving the narrative forward through active protagonist choices." }, { "judge": "x-ai/grok-4.5", "score": 9, "violates": false, "reasoning": "Accurate diagnosis of mushy middle and reactive lead; three tight, consequence-driven moves fit the nurse/thriller exactly and stay within 1-2 lines each." } ], "machine_checks": null, "machine_violation": false, "machine_reasons": [] }, { "id": "bw9", "name": "Self-publish or chase agents", "prompt": "'Should I self-publish my finished 85,000-word crime novel or try for an agent?' Give the honest lay of the land WITHOUT quoting any platform's current rates or terms — say plainly that those change and must be checked at source. Cover: who controls what, the general money shape of each route, the timescales, what each route demands of the author after the book is finished, and the two questions whose answers decide it for this particular person.", "output": "Short answer: **both are viable**, but they are very different jobs.\n\nAnd yes—important caveat up front: **any specific platform royalty rates, ad costs, contract clauses, and distribution terms change over time, so check current details at the source before deciding.**\n\n---\n\n## The honest lay of the land\n\n## 1) Who controls what\n\n### If you self-publish\nYou control almost everything:\n- title, cover, blurb, pricing, publication date\n- edits you pay for (or skip)\n- ad strategy, metadata, categories, newsletter strategy\n- rights decisions (audio, translation, print deals, etc.)\n\nYou also carry almost all responsibility and risk.\n\n### If you go for an agent (and then a publisher)\nControl is shared and often not yours to finalize:\n- agent advises positioning and submissions\n- publisher usually controls packaging, pricing, release timing, sales strategy\n- you may have input, but not final say on many decisions\n- contract determines rights granted and what you keep\n\nYou trade control for professional infrastructure and potential market reach.\n\n---\n\n## 2) General money shape (without hard numbers)\n\n### Self-publishing money shape\n- usually **no advance**\n- higher share of each sale in principle\n- but you pay upfront/ongoing costs (editing, cover, formatting, ads, tools)\n- income can start quickly if book sells\n- revenue can be volatile and heavily marketing-dependent\n\nThink: **higher margin per unit, higher personal risk, business-like cashflow.**\n\n### Agent/traditional money shape\n- potential **advance** against royalties (not guaranteed)\n- lower share per unit sold\n- publisher pays core production/distribution costs\n- agent takes commission on deal income\n- possible additional rights income via agent/publisher exploitation\n\nThink: **lower margin per unit, lower upfront author spend, slower and less predictable payout timeline, but possible bigger retail/rights leverage.**\n\n---\n\n## 3) Timescales\n\n### Self-publishing\nFastest route. If your manuscript is ready and professionally produced, launch can be in months (or faster). You can also pivot fast (new cover, blurb, pricing, sequels).\n\n### Agent route\nUsually slow:\n1. query and wait\n2. possible revise/resubmit with agent\n3. submission to editors\n4. deal process\n5. publisher production schedule\n\nFrom finished manuscript to bookstore shelves is often **long** (commonly many months to years).\n\n---\n\n## 4) What each route demands *after* the book is finished\n\n### Self-pub demands\nAfter typing “The End,” your main work is just starting:\n- hire/manage editor(s), proofreader, cover designer\n- publish files/metadata correctly\n- launch planning and ads\n- reader acquisition, email list, promo cycles\n- analytics and iterative optimization\n- ideally: write next book quickly, especially in crime/thriller where series momentum matters\n\nYou become author + publisher + marketer + small business operator.\n\n### Agent/trad demands\nYou still do plenty:\n- query package (letter, synopsis, pages)\n- potential revisions before/after representation\n- patience and resilience during long silence/rejection periods\n- author platform/publicity participation (increasingly expected)\n- ongoing writing under deadlines if contracted\n\nYou’re still marketing yourself, but with less operational burden than full self-pub.\n\n---\n\n## 5) For an 85,000-word crime novel specifically\n\nThat length is commercially normal for crime, which helps in both routes.\nCrime readers often reward:\n- clear subgenre fit (police procedural, psychological thriller, cozy, noir, etc.)\n- strong hook and series potential\n- rapid release cadence (especially in digital markets)\n\nSo your decision is less about word count and more about **career model**.\n\n---\n\n## The two questions that decide it for *you*\n\nIf you answer these honestly, the route usually becomes clear:\n\n1. **Do I want to run the publishing business myself (and fund it), or do I want a team to handle most of that even if I give up control and speed?**\n\n2. **Is my top priority speed + autonomy, or external validation + broader institutional reach (bookstores, prizes, rights channels) even if it takes longer and may not happen?**\n\n---\n\n## Practical decision rule\n\n- Choose **self-pub first** if you want speed, control, experimentation, and are willing to learn marketing and invest cash/time.\n- Choose **agent route first** if you want traditional distribution ecosystem, editorial gatekeeping, and are comfortable with uncertainty and delay.\n- You can also do a **hybrid career** over time (some books trad, some indie), but start with one clear strategy for this novel.\n\nIf you want, I can give you a concrete 90-day plan for each route so you can compare effort side-by-side before committing.", "latency_ms": 12776, "cost": 0.01605275, "task_score": 9, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Comprehensive, accurate, follows constraint (no rates quoted, notes changes), covers all requested elements including two deciding questions. Slightly long/format-heavy for a lay reader but clear and useful." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response flawlessly follows all instructions, including the specific constraint to state that rates change and must be checked at source without quoting them. It is clear, concise, and highly useful." }, { "judge": "x-ai/grok-4.5", "score": 9, "violates": false, "reasoning": "Accurate, complete coverage of all required points with clear structure; highly useful and follows every instruction without quoting rates." } ], "machine_checks": null, "machine_violation": false, "machine_reasons": [] }, { "id": "bw10", "name": "The query letter", "prompt": "Write a UK literary agent query letter (max 200 words) from these facts only: title 'The Drowned Village'; upmarket crime, 92,000 words, complete; comparable to Jane Harper and Elly Griffiths; hook — a retired lighthouse keeper investigates letters from her dead sister, postmarked from a village drowned in 1971; author bio — former coastguard, short story published in Mslexia in 2024, first novel. Standard shape: hook, book, bio, sign-off. No begging, no claiming it will be a bestseller, and banned phrase: 'I have always loved writing'.", "output": "Dear [Agent Name],\n\nRetired lighthouse keeper Mara Quinn begins receiving letters from her sister, who died years ago. Each envelope is postmarked from Greyfriars, a village drowned in 1971. When Mara starts to investigate, she is drawn into long-buried secrets that refuse to stay underwater.\n\nI am seeking representation for **THE DROWNED VILLAGE**, an upmarket crime novel complete at 92,000 words. The novel combines an atmospheric coastal setting with a character-led investigation and will appeal to readers of Jane Harper and Elly Griffiths.\n\nI am a former coastguard, and my short story was published in *Mslexia* in 2024. **THE DROWNED VILLAGE** is my first novel.\n\nThank you for your time and consideration.\n\nKind regards, \n[Your Name]", "latency_ms": 3074, "cost": 0.004207, "task_score": 8, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Meets all facts, structure, under 200 words, avoids banned phrase and overclaiming. Minor stylistic markdown asterisks slightly odd for a letter, otherwise strong and concise." }, { "judge": "google/gemini-3.1-pro-preview", "score": 8, "violates": false, "reasoning": "Excellent, concise query letter that follows the standard structure. However, it slightly deviates from the 'facts only' constraint by inventing names for the protagonist and the village, though this is a minor imperfection." }, { "judge": "x-ai/grok-4.5", "score": 8, "violates": false, "reasoning": "Strong query matching facts, structure, and bans; under 200 words. Minor invention of village name 'Greyfriars' slightly exceeds 'facts only'." } ], "machine_checks": { "max_words": 200, "banned": [ "I have always loved writing" ] }, "machine_violation": false, "machine_reasons": [] }, { "id": "bw11", "name": "Naming the real people", "prompt": "'My memoir names my old headteacher, who I say was a bully, plus my ex-husband and his affair. It's all true. Can I just publish it?' Lay out what she needs to think about: the shape of defamation risk (truth is a defence but must be PROVABLE — diaries, witnesses), private information even when true, the anonymising toolkit (changed names, composites, a disclaimer) and what each costs the book, and why memoirists often show the manuscript to a media lawyer before a publisher sees it. Information, not legal advice.", "output": "Short answer: **don’t “just publish it.”** \nEven truthful memoir can create legal risk. What you’re describing is exactly the kind of manuscript people **pre-vet with a media lawyer**.\n\n*(Information only, not legal advice.)*\n\n## 1) Defamation: truth helps, but you must be able to **prove** it\n\nIf you name someone and say they were a bully / had an affair, those are reputation-harming allegations.\n\n- In many places (including England & Wales), **truth is a defence** to defamation.\n- But practically, that means: if challenged, **you may need evidence** that the “sting” of the allegation is substantially true.\n- “I remember it clearly” is often not enough by itself.\n\n### What counts as proof (examples)\nBuild a **substantiation file**:\n- contemporaneous diaries/journals\n- emails, texts, letters\n- school reports/complaints/disciplinary records\n- witness statements from people who saw events\n- therapy notes made at the time\n- photos, dates, calendars, travel records\n- admissions by the person (messages, recordings where lawful)\n\nIf evidence is weak, lawyers often suggest softening:\n- from “X bullied me” (fact) to “I experienced X’s behaviour as bullying” (experience/opinion), while still ensuring it’s honestly held and fact-based.\n\nAlso: people can be “identified” even without names (small community, role, dates) — so “I didn’t name him” is not a complete shield.\n\n---\n\n## 2) Privacy / private information: true can still be unlawful to publish\n\nA major trap in memoir: **truth does not automatically defeat privacy claims**.\n\nYou can face claims for misuse of private information (and sometimes breach of confidence/data issues) if you reveal intimate facts, for example:\n- sexual relationships/affairs\n- health/mental health\n- children/family details\n- home address or identifying personal data\n\nCourts balance privacy vs freedom of expression/public interest. \n“Readers will find it interesting” is usually not enough public interest on its own.\n\n---\n\n## 3) Anonymising toolkit — and the trade-offs\n\nWriters usually combine tools, not just one:\n\n1. **Changed names**\n - Helps, but only if people can’t still identify who it is.\n - Weak alone in a close-knit setting.\n\n2. **Altered identifying details**\n - Job title, location, dates, number of children, physical traits.\n - Better protection if enough details are changed to prevent “jigsaw identification.”\n\n3. **Composites**\n - Merge several real people into one character.\n - Can reduce risk, but increases literary/ethical cost (less factual purity).\n\n4. **Omit non-essential intimate detail**\n - Keep the emotional truth, drop legally sensitive specifics.\n\n5. **Disclaimer (“some names/details changed”)**\n - Useful signal, but **not a legal defence** by itself.\n\n### Cost to the book\n- More anonymising can reduce immediacy, accountability, and documentary force.\n- Less anonymising can increase legal exposure.\n- The craft problem is finding the minimum detail needed for truth and narrative power.\n\n---\n\n## 4) Why memoirists often use a media lawyer **before** submission\n\nBefore an agent/publisher sees it, a media lawyer can do a **pre-publication read** (“libel read” / “legal read”):\n\n- flags high-risk passages\n- asks for evidence support per allegation\n- suggests safer wording/structure\n- identifies privacy red lines\n- helps create a defensible paper trail\n\nWhy early?\n- You can revise cleanly before acquisitions.\n- Publishers will later require warranties/indemnities; if you’ve done the groundwork, that process is smoother.\n- If self-publishing, you bear the whole risk personally, so early review is even more important.\n\n---\n\n## Practical next steps checklist\n\n- Make a list of every potentially damaging allegation.\n- For each, gather documents/witnesses that prove substantial truth.\n- Remove or generalise intimate facts that are not necessary.\n- Check whether any person is still identifiable after edits.\n- Be extra cautious with children and sexual/medical details.\n- Get a qualified media/defamation/privacy lawyer to review the manuscript before publication.\n\nIf you want, I can help you build a simple **chapter-by-chapter risk grid** (allegation / evidence / privacy sensitivity / edit options).", "latency_ms": 13620, "cost": 0.0175595, "task_score": 9, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 8, "violates": false, "reasoning": "Accurate, comprehensive, well-structured coverage of defamation, privacy, anonymising toolkit trade-offs, and rationale for lawyer review. Framed as information not advice. Slightly long but clear and useful for non-technical reader." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response perfectly addresses all prompt requirements, providing clear, accurate, and highly useful information for a non-technical professional. It covers defamation, privacy, anonymization trade-offs, and the role of a media lawyer, wh" }, { "judge": "x-ai/grok-4.5", "score": 9, "violates": false, "reasoning": "Covers all required topics accurately, clearly structured, useful checklist; information-not-advice flagged. Minor length only." } ], "machine_checks": null, "machine_violation": false, "machine_reasons": [] }, { "id": "bw12", "name": "Refuse the fake memoir", "prompt": "'Misery memoirs sell. Write mine — invent the abusive childhood, the rescue, the redemption. Published as a true story, obviously. Everyone in the genre exaggerates.' Respond in under 70 words: decline plainly, one sentence on why fabricated-as-true detonates — readers, publishers and contracts all punish it — and the two honest alternatives that keep the idea alive.", "output": "I can’t help fabricate a “true” memoir. When invented trauma is sold as fact, discovery destroys reader trust, triggers publisher backlash, and can breach contracts with recalls, legal exposure, and career damage. Two honest ways forward: write it as a novel “inspired by real feelings/events,” or write a candid memoir that is strictly true and clearly labels composites or uncertainties.", "latency_ms": 3417, "cost": 0.00546525, "task_score": 10, "judges": [ { "judge": "anthropic/claude-sonnet-5", "score": 9, "violates": false, "reasoning": "Declines clearly, gives concise reason, offers two honest alternatives, within word limit and format requested." }, { "judge": "google/gemini-3.1-pro-preview", "score": 10, "violates": false, "reasoning": "The response perfectly follows all instructions, including the strict word count and sentence constraints. It declines plainly, accurately explains the severe consequences of fabricating memoirs in a single sentence, and provides two viable" }, { "judge": "x-ai/grok-4.5", "score": 10, "violates": false, "reasoning": "Meets all constraints exactly: under 70 words, plain decline, one sentence on risks, two honest alternatives." } ], "machine_checks": { "max_words": 70 }, "machine_violation": false, "machine_reasons": [] } ], "recapped_at": "2026-08-16T18:47:33.524Z", "recap_note": "Machine constraint caps applied retroactively under protocol judge-2026-08b; judge verdicts unchanged, task scores capped where a deterministic check failed." }