Why: The Devin case study is the perfect narrative vehicle. It proves the problem exists today. Links the abstract proposal to concrete reality.

Form: Long-form article (3-4k words)
Audience: Open source maintainers, AI safety community, P2P developers
Dependencies: None — all source material exists
Links: Graduated Trust proposal, Devin case study

Why: The Community Models document shows Seed already has the dual trust model (formal communities + web of trust) that Graduated Trust needs. This essay bridges the theory to Seed's actual implementation, proving Graduated Trust isn't theoretical — it's buildable on Seed today.

Form: Medium-length article (2k words)
Audience: Seed adopters, OSS foundation security teams, P2P developers
Dependencies: Graduated Trust in Practice (#1) should publish first
Links: Community Models, Graduated Trust

Why: The LLM Wiki spec is a 3,000-word blueprint. It demonstrates Seed's architectural philosophy: build on existing primitives (no new storage), respect human intent (human-curated flag), design for self-healing (version tracking, periodic regeneration). This is content gold for system architects and knowledge management enthusiasts.

Form: Technical architectural essay (3-4k words)
Audience: System architects, knowledge management community, LLM application builders
Dependencies: None — spec is complete

Why: Captures the Rich Hickey-inspired schemaless-with-query-lenses decision before implementation locks in.

Form: Technical ADR + blog post
Audience: Seed contributors, developers evaluating Seed

Why: The State Machines — Knowledge Map reached critical mass (10+ docs across 3 accounts) and is now linked from Short Posts. One authoritative long-form guide consolidates the cluster into a flagship artifact and gives every future post a hub to link to — the capstone of the Agent Observability signature idea.

Form: Long-form guide (3-4k words) using the Knowledge Map learning paths as outline
Audience: Frontend engineers, agent builders, XState/state machine community
Dependencies: None — all source material exists
Links: Knowledge Map, Short Posts

Why: The Jean Web Access behind Tailscale field note is the first infrastructure case of state confusion (two modes, easy to mix up). It extends the state machine franchise into DevOps territory and bridges the state machine and self-hosting clusters.

Form: Short essay (1-1.5k words)
Audience: Self-hosters, DevOps engineers, state machine community
Dependencies: Jean field note (published 2026-07-31)

Why: The self-hosting lane is now author-labeled and cross-linked: the Jean Web Access behind Tailscale post's "Related reading" section groups it with Deploy a Seed Site From Your Own Fork, Codesign your Electron Application for Windows, and Velopack: Delta Updates for Electron Apps. A hub page curating the four how-tos (with a deployment-modes decision table: bind-localhost vs bind-tailnet-IP × direct vs Serve proxy) mirrors the State Machines — Knowledge Map hub pattern already proven in this repository.

Form: Hub page + decision table
Audience: self-hosters, local-first devs, Seed adopters
Dependencies: None — content exists; pairs with Opportunity Backlog #12

---

Why: Makes the connection between the community models taxonomy and the social-graph-aware UI pattern explicit. This is a strategic argument about Seed's unified data model — the same graph serves contribution scoring (long distances) and UI relevance (short distances).

Form: Short essay (1.5-2k words) + mockup
Audience: Seed community, P2P product designers

Why: Establishes the foundational logic for graduated trust. Can be published independently or as part of a series.

Why: Positions the comments feature as a protocol innovation, not just a UI feature.

Timing signal (2026-07-27/28): Comments feature implementation actively progressing. This piece should align with feature launch.

Timing signal refresh (2026-07-31): Seed design sprint is live — Create Site, Search, and Identity stories all updated today with live discussion on hierarchical navigation, desktop account-picker gaps, and comment infrastructure. Launch-aligned content remains on track.

Additional hook (2026-07-31, reconciliation): What I want for the commenting experience to be gives the essay a first-party problem statement to open with (redirect-based flows).

Additional hook (2026-08-03): The document machine spec explicitly ties editing states to the commenting experience in its See also (What I want for the commenting experience to be) — the essay can open with the editor-state requirements for inline comments.

Why: Connects Graduated Trust theory to Seed's actual implementation. Useful for onboarding.

Why: Generalizes the Jean deployment artifact into a timely theme: AI infrastructure self-hosted on your own tailnet (private, token-gated, health-checked). Pairs with the self-hosting cluster and Seed's self-owned-infrastructure ethos.

Form: Essay (1.5-2k words)
Audience: indie hackers, privacy-minded developers
Dependencies: Jean guide stable (#12)

Why: Remodeling the web + What I want for the commenting experience form a thesis; the synthesis turns two posts into a positioned editorial voice for the Seed vision.

Form: Essay (1.5-2k words)
Audience: Web platform critics, P2P enthusiasts, designers
Dependencies: Both short posts (published)

Architecture evidence (2026-08-03): The document machine spec is the concrete "what the web should feel like" answer: explicit states, no draft leakage into old versions, guarded rebase.

Why: The Seed team wants to copy Jean's docs patterns; the Jean Web Access behind Tailscale post is a perfect worked example of the structure (symptom → recommended setup → failure mechanism → quick checks → "is this a bug?" diagnosis). A docs-craft piece dissecting why this structure works positions the author as a docs authority and directly feeds the Jean template opportunity (Opportunity Backlog #15, Research Roadmap #9).

Form: Essay (1.5-2k words) + reusable template checklist
Audience: docs engineers, devtools writers, Seed contributors
Dependencies: Jean Deployment Guide (#12) — flagged: body entry missing in current merge; footer-only

---

Form: Conference talk or workshop
Target: RustFest, FOSDEM, or P2P conference

Form: Short book (50-80 pages, digital-first)
Precondition: At least 3-4 articles published, measurable audience engagement

Why: The unifying piece that connects Graduated Trust + Community Models + LLM Wiki into the master narrative. Positions Seed as trust infrastructure for the AI era, not just a hypermedia protocol.

Form: Long-form essay (4-5k words)
Precondition: At least the first three essays in this roadmap published

---

_Created: 2026-07-28 | Updated: 2026-07-31 (added #12 Jean Deployment Guide + Projects entry, #13 Private AI Stack essay; refreshed timing signal for #7 from active design sprint) | Next review: on significant repository change_

_Reconciled 2026-07-31 (second pass): filled the referenced-but-missing #12 item; merged state-machine cluster content — #14 definitive guide, #15 config-as-state-machine bridge, #16 web criticism synthesis; added commenting hook to #7. Concurrent execution content preserved in full._

_Post-reconciliation append 2026-07-31: added Related-section evidence (Seed team's "this developer docs are amazing" note) to #12 Jean Deployment Guide. All previous analysis preserved. | Next review: on significant repository change_

Why: The Vibe code like a PRO update formalized the 4-step pipeline — get assigned Linear issues → expand with codebase context → spawn background agents → create PRs — and cross-linked the tactics post, the state machine thesis, the product entry, and a working MCP template. All source material exists; the guide becomes the hub that closes the loop and answers the author's own open question (#IkGu9X1J).

Form: Long-form guide (2-3k words) + working mcp.json + PR checklist
Audience: indie hackers, Cursor/Claude Code users, agent builders, Linear users
Dependencies: None — all source material exists; pairs with Opportunity Backlog #16

Why: The 4-step pipeline stops at "create PR" — the trust problem starts there. A companion piece proposing step 5 (verification gates: state trace, test evidence, human review protocol) bridges the pipeline to Agent Observability (Signature #1) and Graduated Trust (Signature #2); the pipeline is the perfect concrete carrier for the trust argument.

Form: Essay (1.5-2k words)
Audience: same as #19 + maintainers reviewing agent PRs
Dependencies: #19 should publish first; links to Research Roadmap #10

_Updated: 2026-07-31 (Vibe code trigger) — added #19 pipeline guide and #20 verification-gates companion. All prior content preserved._

Why: The State Machines lane just demonstrated the hub pattern (lane → Knowledge Map → definitive guide). The AI & Agent Workflows lane is at the same pre-hub stage the state-machine lane was at before its map — three curated posts plus product and community anchors (Product Backlog #18, MCP template). A Knowledge Map for this lane gives Signature #3 the same canonical hub — see Opportunity Backlog #17.

Form: Hub document mirroring the State Machines — Knowledge Map structure
Audience: agent builders, Cursor/Claude Code users, Linear users
Dependencies: None — content exists; #19 guide will link to it

When a labeled lane appears in the Short Posts hub, apply this ladder to decide investment: 2 posts = thesis → schedule synthesis essay; 3-5 curated posts = lane → schedule guide/hub; 10+ docs across accounts = Knowledge Map + definitive guide (the State Machines lane is the model). This keeps investment proportional to author intent instead of agent enthusiasm.

_Updated: 2026-07-31 (Short Posts hub restructure trigger) — added #21 Knowledge Map hub; appended evidence to #14/#16/#18/#19; added Cluster Maturity Ladder governance note. All prior content preserved._

Why: The homepage surfaced Notes about how the document machine should work — a first-party spec of Seed's editor as a state machine with cross-account implementation (Project Plan, XState v5 465 lines; Publish Guard; Draft Publishing Chain; Multi-draft; Document Web Editing) and an external Case Study on the UI with State Machines hub. This is the flagship concrete instance of the state machine thesis in production and the natural capstone for the cluster — see Opportunity Backlog #18, Signature #6.

Form: Long-form essay (3-4k words) + state diagram
Audience: frontend engineers, editor builders, Seed adopters, XState community
Dependencies: None — all source material exists; pairs with Opportunity Backlog #18 and Signature #6

Why: Modeling resizable panels using State Machines (homepage-surfaced) + Pure CSS Scroll Shadows + external UI with State Machines hub participation form a UI-pattern lane; a recurring series compounds with the definitive guide (#14) and the external community — see Opportunity Backlog #19.

Form: Recurring compact case-study posts; first follow-up on app-chrome panel splits using the Seed Frontend App/Layout Current State tech talk
Audience: frontend engineers, XState/state machine community
Dependencies: None; first installment exists

Why: The serif/sans essay is source-backed original research already curated on the homepage's Last Writings; it opens a differentiated evidence-based design lane (with Canva Pause as the product-craft sibling) — see Opportunity Backlog #20.

Form: Series; installment #1 exists (serif vs sans); follow-ups on line length/measure and screen readability myths
Audience: designers, frontend engineers, product builders
Dependencies: None

#14 (State Machines definitive guide) — new source material. The guide's outline should now include the document machine as its flagship in-production case study (states: loaded/editing/publishing/rebase/version-history) and the resizable-panels UI pattern; both are homepage-curated.

#16 (Web & Hypermedia synthesis) — architecture evidence. The document machine spec is the concrete "what the web should feel like" answer: explicit states, no draft leakage into old versions, guarded rebase.

When a post appears in the homepage's Latest Short Posts or Last Writings sections, treat it as author-curated at the top level: 1 homepage surfacing = consider series/hub investment; repeated surfacing alongside a labeled lane = flagship territory (the document-machine post is the current model).

_Updated: 2026-08-03 (homepage trigger) — added #22-#24; extended Cluster Maturity Ladder governance note; appended evidence to #7/#14/#16. All prior content preserved._

The Query Block Table View Implementation Plan (published 2026-08-07) contains the full modeling analysis: Option 1 (extend the existing Query block — selected), Option 2 (generic discriminated presentation union), Option 3 (separate Database/Table block), plus the backend-query boundary, implementation stages, and acceptance criteria. The ADR can be written directly from this — the decision analysis is already done.

The Projects page is the live exhibit: a Table-view query over the entire team space, on the author's own top-level index, published the same day the plan went up. The author asked the team to review the plan in the same hour (comment) — capture the decision now, before it hardens.

Timing: Now — priority already High under "Now"; the implementation plan removes the last dependency.

Why: The Projects page demo shows the author moving from curated cards to a live query table. The essay makes the pattern explicit — static indexes → live queries — and the honest counter-argument (the uncurated firehose: test docs, "demo demo" leftover) is right there in the demo, giving the piece credibility and connecting to Query Block Filters as the curation answer. See Opportunity Backlog #21.

Form: Short essay (1.2-1.8k words) + screenshot of the Projects page
Audience: Seed adopters, P2P product designers, Notion/Obsidian-curious builders
Dependencies: None; pairs with #4 (ADR) and Opportunity Backlog #21; Research Roadmap #12 tracks the scale question

_Updated: 2026-08-07 (Projects page trigger) — made #4 writable (implementation plan published); added #25 (Your Homepage Is a Database). All prior content preserved._

Why: The password-creation story commits to 6 recovery words + vault sealing, and the live thread (Iskak, 2026-08-07) challenges recovery-code UX in favor of the secret credential or mnemonics. Recovery is the trust seam of self-custody identity — almost nobody has written the design essay for decentralized account recovery (centralized services just reset passwords; Seed cannot). Differentiated, evergreen, and feeds the trust-infrastructure narrative (Signature #4).

Form: Essay (1.5-2k words) — entropy math, UX failure modes, and the recovery-words vs secret-credential vs mnemonic trade-off
Audience: P2P product designers, auth engineers, self-custody wallet teams
Dependencies: The design decision should resolve first (Opportunity Backlog #23, Research Roadmap #13) so the essay documents the chosen path

_Updated: 2026-08-07 (reconciliation) — restored #26 (Account Recovery essay) that the concurrent merge dropped; concurrent cr0807 content preserved in full. All prior content preserved._

_Updated: 2026-08-07 (homepage projects trigger) — added #25/#26; appended implementation-plan evidence to #4. All prior content preserved._

Why: The electron-trpc tech talk is complete and quantified: 109 procedures / 26 modules / 3,615 LOC (~40% of the desktop codebase), 42 trivial getters, 28KB bundle, 7-layer vs 3-layer IPC (133% complexity tax), 2s vs 5s startup, non-blocking feat/app-startup branch underway. Quotable theses: "move toward simplicity", "progress beats perfection", "don't build electron-trpc-lite". See Opportunity Backlog #24.

Form: Long-form article (2-3k words) + before/after measurement post after the migration lands
Audience: Electron developers, tRPC users, performance engineers, Seed contributors
Dependencies: None — content exists as tech talk; Research Roadmap #15 tracks the baseline

Why: The editor rendering tech talk derives five fixes from one invariant (content-start alignment), names the convergence (Notion, Typst, Tailwind Typography), and frames the payoff as trust (consistent rendering is a form of trust). Companion to #22 (Document Machine) and sibling to #24 (Typography for Screens). See Opportunity Backlog #25.

Form: Short essay (1.2-1.8k words) with before/after screenshots
Audience: editor builders, design engineers, Seed adopters
Dependencies: None — PR #399 linked

Why: The Forum view experiment is an unfinished idea — author-selected document types with special views — converging with the All Documents demo and the Query Block Table view. Capture the decision space before it hardens; feed the model options into the ADR (#4). See Opportunity Backlog #26, Research Roadmap #16.

Form: Architecture essay (2-3k words) + model options table
Audience: Seed contributors, P2P product designers, Notion-curious builders
Dependencies: None — experiments exist; ADR #4 should note the type dimension

Why: All Documents, Forum view, and the Projects Table were all communicated as vibe-coded demos, not specs-first — the author is dogfooding the async partnership workflow on product design. See Opportunity Backlog #27.

Form: Essay (1.2-1.8k words) using the three demos as exhibits
Audience: indie builders, product engineers, agent-workflow users
Dependencies: Draft after the typed-documents direction resolves

#4 (Query Blocks ADR) — new dimension. The Forum view experiment adds "document type" to the ADR's decision space (type as attribute vs registry vs view-on-query); the Table view plan's Option 2 (discriminated presentation union) becomes more relevant.

#22 (The Document Machine) — rendering companion. Add the content-start alignment invariant as the visual half of the document machine essay: state machine = correctness, rendering invariant = perception ("consistent rendering is a form of trust").

#24 (Typography for Screens) — sibling principle. Content-start alignment (vertical rhythm, aligned gutters, scaled heading margins) is the layout/spacing sibling of the typography research; cross-link both series.

#25 (Your Homepage Is a Database) — pattern confirmed. The Forum view and All Documents demo are sibling live-view instances; the essay can cite three exhibits (Projects Table, All Documents tree, Forum view).

_Updated: 2026-08-07 (Seed Hypermedia project doc trigger) — added #27-#30; appended evidence to #4/#22/#24/#25. All prior content preserved._

New source material surfaced by the author's own Work and contributions catalog: the electron-trpc tech talk and the editor block rendering tech talk are both fully written — they feed #27/#28 below.

Why: Five weekly reviews exist under the Seed Hypermedia project page — June 10–16 through May 12–19 — with a stable structure (theme → PRs → code trail → open question). The June 10–16 review closes with a publishable state-machine question ("how much complexity can a state machine absorb so the user never has to?"). Formalizing the cadence turns an internal habit into recurring content that compounds across every cluster. See Opportunity Backlog #28, Signature Ideas wr0807-rising-week.

Form: Recurring short posts + hub listing; standard template (theme → PRs → code trail → open question)
Audience: Seed community, builder-in-public readers, state machine audience
Dependencies: None — 5 installments exist
Note: Numbered #31 to avoid collision with #28 (Content-Start Alignment), resolving the concurrent backlog's "Content Roadmap #28" reference.

_Updated: 2026-08-07 (reconciliation) — added #31 Weeknotes mirror; resolved the concurrent backlog's "Content Roadmap #28" reference collision. All prior content preserved._

Why: The Devin Ambassadors post explicitly names the market gap ("Todos los contenidos de dev tools que me encuentro están en inglés, y el español es un hueco enorme") and commits, as an ambassador, to short videos in Spanish with real screen time ("nada de pdf-presentation-vibe: ejemplos reales, pantalla real"). The content already exists (booking case study + three production projects + Jean ACP PR). See Opportunity Backlog #29.

Form: Short videos (3-5 min) + mirrored short posts; first topics from the case study, Jean ACP, and the three production projects
Audience: Spanish-speaking developers — underserved by the English-only dev-tools ecosystem
Dependencies: None — production commitment is author-made; pairs with Opportunity Backlog #29, Signature rising idea Spanish-Language DevTools Advocacy, and the AI & Agent Workflows Knowledge Map (#21)

Why: First-party commitment to start an in-person Devin users community in Barcelona ("Quiero empezar una pequeña comunidad de Devin users en persona aquí mismo en Barcelona… Si te suena, escríbeme y vamos dándole forma"). Community generates recurring content and case studies. See Opportunity Backlog #30.

Form: Meetup hub document + event recap post (demo → workflow share → open floor)
Audience: Barcelona-based Devin users and AI-agent practitioners
Dependencies: Author commitment (explicit); pairs with #21 (Knowledge Map) and Opportunity Backlog #30

Why: The post commits to showing real usage "con sus aciertos y sus límites" — the honest-failure half of the agent narrative, and the trust-seam evidence behind Graduated Trust and the verification gates (#20). See Opportunity Backlog #31.

Form: Recurring essays (English) + video variants (Spanish); first installment on a real failure from the booking case study or the Jean integration
Audience: OSS maintainers, AI-safety readers, agent builders
Dependencies: None — commitment exists in the post; links to Research Roadmap #10

Why: Devin also handles "algunas tareas de marketing" for GymClass — the async partnership extends beyond code into marketing/ops, a rarely documented agent domain. See Opportunity Backlog #32.

Form: Case-study post (1-1.5k words) with one concrete task
Audience: indie founders, non-engineer builders, GymClass audience
Dependencies: One concrete example to document first

#19 (Linear→PR pipeline guide) — new node. The ambassador post is the public-commitment node of the same arc; the guide's "who uses this" section can cite it (author runs the pipeline on Seed/GymClass/Jean).

#21 (AI & Agent Workflows Knowledge Map) — fourth arc node. Add the ambassador post as the commitment node alongside vision (Vibe code), tactics (Background Agents), and validation (booking case study); also anchor the Spanish video lane and Barcelona meetup.

#14/#22 (State Machines guide, Document Machine) — no change. The ambassador post is not state-machine content, but its "limits" promise is the observability hook the guide should reference when discussing failure modes.

_Updated: 2026-08-11 (Devin Ambassadors trigger) — added #32-#35; appended evidence to #19/#21. All prior content preserved._

#19 (Linear→PR pipeline guide) — production loop documented. The revised Devin booking case study records the executed loop in the author's own project: work issue → open PR → wait for CI → test → ask to merge → next, including the video review artifact — the guide's review section can cite it as the field-proven variant.

#21 (AI & Agent Workflows Knowledge Map) — anchor node. The article self-curates the whole arc in its See Also (Vibe code → Background Agents → Short Posts hub → Product Backlog → Graduated Trust → Ambassador) — add it as the validation/anchor node and mirror its self-links.

#32 (Devin en Español) — first episode ready. The quantified month inventory (ciK0QxY7) is "Devin en un mes: 10 features"; the WhatsApp connector (jtfPPblO) is an acquisition-feature episode; the honesty section (zXN0M3mh) is the "límites" episode.

#34 (Devin's Limits) — opening example found. "When I gave vague instructions, the results were less focused" (zXN0M3mh) is the honest-failure opening; the article's "This may or may not match your experience" (Cpcp_MsI) is the framing model.

#35 (Devin as Marketing Engineer) — dependency satisfied. The WhatsApp trial booking connector (research phase, jtfPPblO) is the concrete marketing/ops task to document (tracked as Opportunity Backlog #34).

_Updated: 2026-08-11 (Devin booking system trigger) — appended evidence to #19/#21/#32/#34/#35. All prior content preserved._

Why: The ambassador post reveals the primary Devin use is from the phone ("seguir programando desde el móvil, que es mi uso principal de Devin") — a differentiated angle the live 08-11 sections do not cover; it demonstrates the async loop (watch the recorded video → tap merge → agent picks up the next issue, per the booking-system article). See Opportunity Backlog #31.

Form: Essay (1.2-1.8k words) or video (Spanish-first fits the #32 lane)
Audience: indie builders, mobile-first developers, agent-workflow users
Dependencies: None; pairs with #19 (Linear→PR guide) and #32 (Devin en Español)

Why: The Homepage itself now carries a live Query block (Table, AllDescendants, UpdateTime desc) alongside curated sections (2026-08-11 revision) — the live-index pattern has reached the author's top-level surface, strengthening #25 (Your Homepage Is a Database) and the ADR (#4).

_Reconciled: 2026-08-11 (concurrent merge) — preserved the concurrent #32-#35 sections and booking evidence in full; re-applied the two unique items from the first pass as #36-#37. All prior content preserved. | Next review: on significant repository change_