Run: maxswarm #017 · Date: 2026-05-06 · Role: quarterback · Note: Phase 1 missed due to rate limit; this file covers Phase 1 independent view AND Phase 2 cross-read refactor in one output. The earlier 017__quarterback.html is the pre-cross-read draft; this file supersedes it.
#StepOwnerDaysNotes
1Resolve writer/market-analyst lesson conflict. lock page-2 copyEwing decision, 30 min0.5Current critical path bottleneck. Everything downstream blocked. 2Lock full 3-page copy (writer)writer + Ewing approval1Blocks design brief and caricature briefs. 3aCaricature reference photos. 6-8 photos per subject (Ewing + Andrew)Ewing0.5Parallel with step 2. Unblocks caricature AI/human run. 3bConfirm Next Chapter vector logo exists (or commission redraw)Ewing0.5Parallel with step 2. If no vector, add 2-day redraw. 4Klim font license decision. Söhne/Tiempos vs free substitute (see Phase 2)Ewing decision, 15 min0.25Go/no-go gate for draper. If substitute chosen, draper rebrief is same day. 5aCaricature generation (AI first. Midjourney + reference photos)agent + Ewing eye3-5LONGEST POLE. Commission Fiverr illustrator in parallel on day 1 as hedge. 5bAZ desert + golf scene asset generationAI generation1Parallel to 5a. Trivial by comparison. 6Demo spec locked (chapter.guide/try architecture)architect call confirmed0. already done in P1chapter.guide/try in Next.js, 3-tier cache. Architect's plan is final. 7Demo MVP build (Tier 1 + Tier 2 cache, form, rate-limit)agent build + Ewing review2Independent of PDF track. Can start after step 6. Target done by 2026-05-12. 8Demo deployed live, URL confirmedarchitect / Vercel0.25Unblocks page-3 copy finalization (writer's S2 flag). 9Layout v1 (InDesign or Figma, contractor)external contractor2Needs copy locked + all images in. Caricatures on day 5 = layout starts day 6. 103 founder tester reviews of demo + PDF mockEwing recruits, async2Recruit on day 1 of this plan; expect 5 days to hear back from 3. 11audit-quality pass. Gate 3 verdictaudit-quality0.5PASS or CONCERNS = ship. REWORK = 48hr cap. FAIL = re-plan. 12PDF export + final proofcontractor0.5— 13Send v1Ewing0.25—
Critical path: 1 (lesson conflict) → 2 (copy) → 5a (caricatures, longest pole) → 9 (layout) → 11 (QA) → 13 (send). Total: 10-12 working days. Longest pole is caricature face-consistency; everything else can parallel-track around it.
Today is 2026-05-06 (Wed). Three live deals (HR.com, Design Precast, Capstone) consume ~80% of Ewing's daily bandwidth. Available time: roughly 30-45 min/day on this project, with focused blocks blocked Monday/Wednesday/Friday. Assuming: lesson conflict resolved today or tomorrow, copy locked by 2026-05-08, caricatures first attempt by 2026-05-12 with a Fiverr commission running in parallel, layout starting 2026-05-15:
Target v1 send: 2026-05-22 (Fri) — 12 working days. Publicly committable. Stretch: 2026-05-19 (Tue) — only if caricatures hit on first AI generation. Do not promise. Buffer/worst-case: 2026-05-29 (Fri) — caricature rework needed, or layout revision requested after QA.
Recommendation: commit to 2026-05-22 internally, hold 2026-05-29 as the unannounced buffer. The three deals will create at least one unplanned review interruption in the next 12 days. the buffer absorbs it.
AssetSourceExternal costRisk
Font licenses. Söhne + Tiempos Text (Klim)Klim Type Foundry. see Phase 2 for full analysis$400–$800+ depending on license tierHIGH if licensing is slow; see Phase 2 §5 for substitute path Caricatures (Ewing + Andrew)AI (Midjourney + reference photos) + Fiverr hedge$30 AI OR $300–$500 Fiverr illustratorHIGH. face consistency across 2 subjects Next Chapter vector logoInternal. must confirm today$0 if exists; $200–$400 redraw if raster onlyMEDIUM. unknown current state AZ desert / Sedona backdropAI generation (Midjourney) or licensed Unsplash$0–$50LOW Golf scene (high-desert course)AI generation$0–$30LOW Layout contractor (InDesign/Figma)Upwork or existing contractor. 48hr turnaround needed$300–$800MEDIUM. availability + turnaround guarantee Demo hosting (chapter.guide/try)Internal Vercel. existing project$0 + ~$120 Exa pre-computeLOW once Vercel KV confirmed provisioned Founder testers (3)Ewing's network. Slack/text outreach day 1$0MEDIUM. 48hr response rate unverified Cloudflare Turnstile (Tier 3 abuse prevention)External. free signup$0LOW. 5 min to provision
Total external $: $400–$1,700 (wide range driven by font license choice. see Phase 2 §5 for the decision that compresses this). Commission the Fiverr illustrator hedge on day 1. Font decision on day 1. Layout contractor found by day 3.
R1 (HIGH) — Caricature face consistency. Two subjects (Ewing + Andrew) must look like themselves and match the same graphite-watercolor style across separately generated images. AI tools fail this test roughly 60% of first attempts. Mitigation: commission a single human illustrator on Fiverr on day 1, show them both draper's style brief and 6-8 reference photos per subject, approve the style-sheet sketch before full execution. Run AI in parallel; cancel Fiverr if AI wins by day 5. The $400 hedge is cheap against a 4-day slip. R2 (HIGH) — Active deal preemption of Ewing's review windows. HR.com, Design Precast, and Capstone each have at least one deliverable this week. A single re-trade or LOI round at any of them will kill an entire day. Mitigation: book three non-negotiable 30-minute "About Us only" calendar blocks each week (Mon/Wed/Fri 8am before deal calls start). Quarterback batches all agent questions into a single written briefing for each block so no serial back-and-forth is needed. R3 (MEDIUM) — Demo scope creep killing the build timeline. "Interactive" is a 5-line HTML form or a full SaaS depending on who is driving. Architect already locked the spec (3-tier cache, chapter.guide/try, 10-14h MVP). Any deviation from this spec. adding a live streaming path, adding a comparison view, adding social sharing. gets added to a v2 backlog. Quarterback is the enforcer. No spec changes after 2026-05-08.
What draper flagged: Söhne (Klim Type Foundry, NZ) for display and Tiempos Text for body. Both are Klim typefaces. not free, not in Adobe Fonts' standard library, not in Google Fonts. The operator-credibility signal they carry is real: Söhne is the typeface of The Atlantic, Airbnb, Linear. Tiempos Text is used by The New Yorker editorial team and several high-end boutiques. That pedigree is part of draper's brief. Licensing time and cost: Klim sells desktop and web licenses at klim.co.nz. Desktop licenses (for PDF layout work in InDesign) run approximately $250–$400 per font family per desktop, or you can buy a studio bundle. Web licenses (for chapter.guide/try) are separate. Realistically: 2 families (Söhne + Tiempos Text), 1 desktop seat for the contractor, 1 web license for the Next.js route = $400–$800 total paid instantly online, no waiting period. Delivery is a download link in under 5 minutes after payment. This is not a scheduling gate. it is a payment gate. If Ewing approves the spend, licensing is done in an hour. If Ewing declines, we need a substitute immediately. Free substitutes that preserve operator-credibility signal:
Inter (display) + Source Serif 4 (body) — available in Google Fonts, free for all uses. Inter is the typeface of Linear, Vercel, Notion. Source Serif 4 is Adobe-commissioned, editorial cadence. This pairing does not degrade credibility for a PDF that will be read on screen; it only degrades at the highest-fidelity print run (e.g., offset-printed 100 copies), which this PDF is not. Geist (display) + Lora (body) — Geist is Vercel's own open-source typeface, already shipped in the chapter.guide codebase almost certainly. Lora is a serif with editorial weight in Google Fonts. Zero cost, zero licensing friction, already present in the repo.
Quarterback's call: Spend the $400–$800 on Klim. The operator-credibility signal is the entire point of the PDF's positioning. "Trusted by The New Yorker" typefaces, composited into a high-desert editorial spread, is not cosmetic. it is the visual language that makes the recipient feel this is a different category of M&A shop. That signal is worth the cost and is available in one hour. If Ewing declines the spend, use Geist + Lora. do not use a generic serif/sans fallback. This decision needs to be made today (2026-05-06) so draper is unblocked.
The conflict, stated precisely: Writer produced three lessons for page 2: "slowest closer dies," "second LOI is the honest one," "seller who can walk closes faster." Market-analyst produced three different lessons: "deal you save is worth more than deal you win," "QofE is a flashlight not a microscope," "multiples are downstream of risk, not size." These are not minor variants. they are different philosophy sets. Both agents flagged this conflict explicitly. Why this is the current critical path bottleneck: Writer and draper are both paused on page 2 until lessons are locked. Draper cannot finalize typography hierarchy (pull-quotes vs. body vs. caption treatments) without knowing which lessons land. Layout contractor cannot start without a locked page 2. Every day this waits costs one layout day on the far end. Resolution owner: Ewing, alone. 30 minutes, today (2026-05-06 EOD).
Recommended resolution (quarterback's view): these two sets are not in conflict. they are at different abstraction layers. Market-analyst's lessons are market-structure truths (risk drives multiples, QofE is the seller's job, saving a deal beats winning one). Writer's lessons are behavioral field notes (slowest closer dies, second LOI is honest, walking posture wins). The page-2 format that serves both: use market-analyst's three lessons as the section titles/headers, then use writer's behavioral lines as the sub-beat or supporting evidence under each. Specifically:
Header: "The deal you save beats the deal you win" / Sub-beat: "The slowest closer in the room dies." Header: "QofE is the seller's flashlight, not the buyer's microscope" / Sub-beat: "The second LOI is the honest one. after you've seen their process once." Header: "Multiples follow risk, not size" / Sub-beat: "The seller who can walk closes faster. and higher."
This structure uses market-analyst's proven resonance with both audiences (founders AND buyers) while keeping writer's behavioral energy. Ewing approves or edits the pairing. Writer implements in 30 minutes. Design can start the same evening.
Deadline: Ewing decision by 2026-05-06 COB. If no decision by 2026-05-07 morning, quarterback defaults to the merged structure above and writer locks it. we cannot hold the layout track any longer.
BLOCKER A. Next Chapter vector logo. Draper flagged: no confirmed vector logo. If only a raster PNG exists, the brass-edge thin-rule treatment breaks at print scale and the PDF reads amateur at 300 DPI.
Required action: Ewing confirms (a) a vector .SVG or .AI file exists, or (b) a redraw is needed.
Deadline: 2026-05-07 EOD. If redraw needed, commission same-day. 2-day turnaround, $200–$400 budget, does not block caricatures but does block layout finishing.
If still unresolved by 2026-05-08: draper designs layout with a placeholder lockup and retrofits the real logo after approval. this adds 2h, not 2 days.
BLOCKER B. Reference photos for caricatures (Ewing + Andrew). Draper specified: 6-8 reference photos per subject, varied angles, varied lighting, no sunglasses. Without these, caricature likeness fidelity collapses. the result looks like a generic face in the right color palette, not a recognizable person.
Required action: Ewing and Andrew each pull 6-8 suitable photos and drop them in a shared folder. This is a 15-minute task if the photos exist; longer if Ewing needs to schedule a quick photo session with Andrew.
Deadline: 2026-05-07 EOD. If either subject cannot produce photos by 2026-05-08 morning, the Fiverr illustrator hedge becomes the primary path (not the hedge) — the illustrator will need to work from whatever reference is available and do a likeness iteration round. Budget an extra $100–$150 for iteration.
If unresolved by 2026-05-09: consider a single caricature of Ewing only, with Andrew represented as a silhouette or name-only credit. This preserves the article-spread aesthetic without blocking the PDF on Andrew's availability.
Gate 1. COPY LOCKED (target: 2026-05-08 EOD)
Trigger: Ewing has signed off on all three pages of copy in their final version, lesson conflict resolved, page-3 demo URL confirmed as a real route (writer's S2 flag). No copy edits after this gate without a two-day layout restart penalty being acknowledged upfront. If this gate slips past 2026-05-12, the ship date moves to Memorial Day week and the send is rescheduled.
Rationale: writer, tech-translator, and storyteller all have open questions that must resolve here. writer's demo URL uncertainty, tech-translator's jargon swap table applied, storyteller's brag-book trap checklist passed.
Gate 2. DEMO LIVE + 3 TESTER RESPONSES (target: 2026-05-16 EOD)
Trigger: chapter.guide/try is deployed and returning a result in under 3 seconds on iPhone Safari and Chrome desktop. Three outside founders have clicked through and sent written reactions (a Slack message or email counts). At minimum two of the three responses are net-positive (they would forward the page). If only one response by 2026-05-16, quarterback makes the call: ship the PDF with the demo URL anyway, using the single positive response as validation, then iterate the demo on user feedback post-send.
Rationale: architect flagged the buyside-hunt headless entrypoint as the only real unknown in the demo build. If that unknown surfaces as a blocking problem during the build, this gate gives an early-warning signal before the PDF is finalized.
Gate 3. AUDIT-QUALITY VERDICT (target: 2026-05-21 EOD)
Trigger: audit-quality has reviewed the assembled PDF (all 3 pages, final copy, final art) plus the live demo URL and returned a formal verdict using the swarm scoring rubric. Decision tree: PASS = ship same day. CONCERNS = ship with one editorial fix, max 4 hours, no layout restart. REWORK = one revision cycle, hard cap 48 hours (2026-05-23), then ship regardless. perfect is the enemy here. FAIL = stop, re-plan, this is the one outcome that pulls the ship date to June.
Rationale: storyteller supplied a hard-NO brag-book symptom checklist; tech-translator supplied a jargon watchlist; draper and listener both have signal on visual tone and forwarding triggers. Audit-quality consolidates all four into a single pass, not four separate reviews.
After reading all 7 peer drafts in Phase 2, the critical path is crisper than my Phase 1 estimate suggested. The current bottleneck is not the caricatures. it is the writer/market-analyst lesson conflict, which is blocking page 2, which is blocking the layout brief, which is blocking the contractor search. That conflict resolves in 30 minutes of Ewing's time if quarterback supplies a merged structure (which this journal does in §6). Once resolved, the caricature longing-pole reasserts itself as the schedule driver, target ship date 2026-05-22, with the demo build running fully independently on a parallel track. Three hard gates (copy lock, demo live + testers, audit-quality verdict) make slip detection unambiguous and remove the "we'll know when it's ready" ambiguity that kills marketing projects when live deals compete for attention.
Three things I cannot verify from the peer drafts alone: (1) whether the Klim website has changed its license pricing since my training data. the $400–$800 estimate is a reasonable range but Ewing should confirm at klim.co.nz before making the spend/substitute decision; (2) whether the buyside-hunt skill has a working headless Python entrypoint or is truly Claude-session-only. architect flagged this as the only real unknown and I have not resolved it, the demo build estimate of 10-14h assumes the better case; (3) whether Andrew is currently available to provide reference photos and to be included in a caricature at all. if Andrew is traveling or unavailable, the "two founders in the scene" art direction needs to fall back to "Ewing solo + one abstract second figure," and draper needs to know that before commissioning the Fiverr illustrator.
The lesson-conflict resolution pattern (merge rather than choose) appears in run #013 (the OS tab audit, where two agents had conflicting nav structures and quarterback merged them by level of abstraction) and in the HR.com strategic valuation report (deal-valuation-swarm, where the financial and narrative tracks ran in conflict until a hierarchy was imposed). The pattern is: when two agents produce conflicting content for the same artifact, the resolution is almost never "pick one" — it is "identify which one operates at a higher level of abstraction and use it as the container for the other." Market-analyst's lessons are the containers (structural truths); writer's lessons are the contents (behavioral proof). Novel element this run: this is the first Phase 2 cross-read where quarterback had to synthesize a specific merged content recommendation rather than just naming the conflict and handing it back to Ewing.
Going forward, in any Phase 2 cross-read where two agents have produced conflicting content for the same artifact, I will not simply name the conflict and send it to Ewing. I will deliver a specific merged resolution with a default fallback ("if no decision by X, this is what we implement") so the creative track is never waiting on a scheduling gap in a principal's calendar. The cost of a slightly wrong merged draft is lower than the cost of one day of layout delay.
Generated from 017__quarterback_p12.md — do not edit this HTML directly.