08 — Gaps: What Must Be Written Fresh

Twelve gaps. These are the pages the corpus cannot supply, ranked by how much they block the site.

Blockers

G1 · The founder's own "meaning through connectivity" essay does not exist

The single most important page on the site, and it must be written from scratch.

The philosophy's canonical written statement (library/concepts/v0_4_0__thinking-in-graphs.md) was written for Issues-FS, by a different hand, about issue tracking, on 5 February 2026. Every founder-voice statement of the thesis in the SG/Send corpus is a fragment inside a brief about something else.

His own version lives in a LinkedIn post series that he asked to be supplied as prerequisite reading on 10 June 2026 (briefs/06/10/network-intelligence/v0.33.16__arch-brief__...subgraph-flip-verb-edges.md line 19 records the ask, and the agent's note that it was not provided). It is still absent from the corpus.

What to do: import the LinkedIn series, then write the page from his voice using the 35 quotes in 01__the-thesis.md as the skeleton. Do not lift the Issues-FS essay and present it as his. Blocks: /, /start/meaning-through-connectivity/, everything downstream.

G8 · Licence blocker on the foundational documents

All ~46 founder conceptual briefs end with "released under the Creative Commons Attribution 4.0 International licence (CC BY 4.0)" — 1,103 files repo-wide carry it, 223 of them graph-topic.

The three library/concepts/ documents carry no licence line at all and are attributed to Issues-FS. So do library/concepts/README.md, library/guides/agentic-setup/v0_4_0__role-ecosystem-guide.md and the three REFERENCE__from-issues-fs.md role files.

The site's intended centrepiece is the one thing in the corpus with unresolved publishing rights, and /start/ is built almost entirely from it. What to do: resolve licence and attribution before launch. This is the only true hard blocker in the pack.

Missing pages the corpus assumes exist

G2 · "Documents are projections of graphs" has no home document

Always invoked as already established — "the same way I talk about documents being projections of graphs" (4 Jun) — and then applied to skills, compliance standards, and legal texts. The general principle is never argued anywhere.

It is load-bearing for the entire product line: vault-per-standard, the regulation graph, skills-as-graph, and the consolidation-equals-customisation insight all rest on it. What to do: write the synthesis page. Raw material: briefs/06/04/v0.32.3__arch-brief__sg-send-skill-as-projection-of-graph-forking-ecosystem.md + briefs/07/31/canonical-act-build/v0.33.54__arch-brief__...paragraph-as-file-amendment-as-native-graph-operation...md + briefs/05/30/v0.31.9__arch-brief__sg-send-vault-per-standard-document-to-graph-artefacts.md. Target page: /depth/everything-is-a-projection/

G3 · "Why graphs at all" — the outsider's page

Every document in the corpus assumes the reader already accepts that graphs are the right substrate. There is no page that argues it to a sceptic, and none that distinguishes this use of graphs (semantics) from the two common ones (network analysis, query performance).

The corpus knows it is doing something unusual — "Not a graph database pitch" — but never explains what. What to do: write it. This is the page that stops a graph-literate reader dismissing the site in the first thirty seconds. Target page: /why-graphs/

G4 · The vocabulary page the founder explicitly asked for

Quote 34, 9 August 2026: users "don't know about semantic graphs, don't know about ontologies… we also need to explore different UIs, and different ways to name this." The document's own gloss: "naming is a design problem rather than a documentation one."

The corpus contains exactly one terminology table — concept / term / taxonomy / ontology / semantic field / concept scheme — buried inside briefs/08/06/voice-debrief/v0.33.56__...concepts-not-words...md. What to do: expand it into a real glossary, giving each term a plain-English alternative alongside the technical one. Draw the taxonomy-vs-ontology distinction from briefs/07/31/canonical-act-build/v0.33.54__arch-brief__...taxonomy-upward-ontology-outward...md. Target page: /glossary/ — a named, outstanding request from the founder.

G5 · The corpus glossary is cited but does not exist

Appendix A of the 28 July brief cites "the concepts glossary" as the authority on edge-grammar discipline and the established edge set: connected_to, observed_on, backed_by, measured_by, grants, reaches, enables, exposes, gives_rise_to, protected_by, conditional_on, defeated_by, owned_by, accepted_by, underwritten_by.

No such file exists in this repo. It lives elsewhere (RiskMandate). What to do: the website should either import it or become it. Becoming it is the better answer — an authoritative, versioned, public edge vocabulary is exactly what a graph reference site is for, and it directly serves the agent audience. Target page: /grammar/the-edge-set/

G6 · No newcomer path exists at all

library/concepts/README.md (254 words) is an import manifest, not an introduction. The three foundational docs are 3,800–5,000 words each with no summary, no diagram beyond ASCII, and no ordering. There is no 500-word version of anything. What to do: the whole /start/ section, plus a 500-word "the short version" on the front page. The teaching order in 02__concepts-index.md is the sequence.

Style blockers on excellent material

G7 · The abstract style makes the best material unreadable cold

The 12 Jul, 17 Jul, 28 Jul and 9 Aug briefs — carrying C17, C18, C20 and C22, four of the most important concepts — each open with a single bolded sentence of 400–500 words.

The ideas are outstanding. The documents are not readable cold by anyone who is not already inside the project. What to do: every Tier-1 brief from July onwards needs a written-fresh 300-word opener before publication. The highest-value single rewrite is briefs/07/12/architecture/v0.33.48__...fractal-semantic-graphs...md → /depth/a-graph-at-every-boundary/.

Concepts with no document

G9 · "The air gap" has no page

A vivid, load-bearing idea — the named, tracked place where the graph cannot reach reality — appears only as paragraphs inside two longer briefs (briefs/06/26/digital-twins-and-world-models/v0.33.35__...integration-layer-real-world-tracked-air-gaps-agent-twin.md, briefs/06/26/risk-register-and-five-whys/v0.33.35__...).

It has an excellent illustration waiting: the air-gap Wardley map (briefs/07/28/mvp-and-field-demo/v0.33.53__strategy-brief__...airgapped-register...) — "the ends are solved, the middle is people." What to do: write the page, pair it with the map, and connect it to C3 (map the gaps). A named absence beats a hidden one — that is the site's own argument applied to itself.

G10 · Time as a first-class dimension is asserted, never developed

"Time is an event, things change" (18 Jun) and thinking-in-graphs.md Part 6's versioned cross-graph edges are the entire treatment. The corpus repeatedly says the graph moves and never explains how.

Adjacent material that circles it without landing: the acceptance-interval ladder (1h/4h/2d/2w/1m/6m), repealed_from in the regulation graph, supersede-never-delete (C29), NHI 2.0's temporal permissions and "time travel", and the vault's own commit DAG — which is, after all, a working answer to "how does a graph change over time". What to do: write it, and use the vault DAG as the shipped proof.

G11 · No treatment of GraphRAG, hypergraphs, or property-graph vs RDF trade-offs

Zero corpus hits for "GraphRAG" or "hypergraph". The corpus has a strong implicit position against retrieval-by-similarity — "Knowledge is traversed, not guessed… not a similarity search that returns plausible chunks" — but never engages the named field.

A technical audience will ask on day one. There is no answer to cite. What to do: write a short, fair positioning page. It should distinguish this approach from GraphRAG honestly rather than dismissively, and it should state the property-graph vs RDF position — noting that the live regulation graph does export RDF/Turtle, so the position is "both, at different layers", not "against RDF".

G12 · Nothing links the philosophy to what actually ships

Per team/roles/librarian/reality/, the semantic-graph and ontology work is almost entirely PROPOSED. The shipped reality is the vault, the encryption, sgit, and the file-based substrate.

Without an explicit page, the site will over-claim — which would break the exact convention that makes the sibling sites credible. What to do: /shipped/. Content and the honest sentence are in 05__site-architecture.md. The good news: what does ship is genuinely interesting — a content-addressed multi-parent commit DAG with a real merge-base algorithm, a graph of graphs via .link.json, a DAG query API for untrusted apps, and 71 live nodes / 141 edges in .issues/.

Summary — the twelve fresh pages

Everything else on the site already exists in the corpus. The 19 Tier-0 documents in 07__source-manifest.csv are publishable near-as-is, subject only to G8.

== briefs/09__licensing-decision.md