sgit newsroom v0.1.29 · snapshot 2026-09-24

Reading room · influences.sgit.ai · llms-full

On this page

Reading room / influences.sgit.ai / llms-full.txt · section 33 of 38

03 — Site Architecture: influences.sgit.ai

Version v0.33.62 · 25 August 2026 Base The house pattern (copy pki.sgit.ai): llms.txt + /llms-full.txt, /documents/ raw markdown, markdown twin at every URL, /admin/comms.html with N1…/T1… threads, /shipped/, versioned everything, participant page.


1. URL scheme

/                          the thesis + the three-tier map (counts generated)
/register/<slug>/          one influence entry (full format, 7 blocks)
/register/<slug>/trace/    the trace table alone, as data (md + json twins)
/tiers/                    TRACED / STATED / DISCOVERED, with definitions
/map/                      the influence graph (influences → principles → estate features)
/format/                   the register format itself, documented (from 01__)
/library/                  the union of all Block-7 wider libraries, one page per person/topic
/documents/                raw markdown of everything
/admin/comms.html          questions to the founder (see §4)

Slugs are people or topics, not works: /register/bret-victor/, /register/wardley/, /register/semantic-web/, /register/flow/, /register/luhmann/, /register/neil-peart/… The anchor work is metadata inside the entry, because anchors can be superseded (a briefing document may reveal the real anchor was a different talk).

2. Data model

influences__seed.json (this pack) is the initial vault content. Each record: slug, tier, kind, status, anchor, principle, corpus-evidence[] (path + note, regenerable), sibling-site, nests-under, briefing-status, research-plan. The site is a projection of this data — same architecture as standards.sgit.ai: the register lives in a vault, the site renders it. Trace tables are per-entry JSON so /map/ can be generated, not drawn.

Generated, not claimed: tier counts, corpus-evidence file counts, and the /map/ graph are all derivable from the seed JSON + a corpus scan. Publish the scan script's identity and date beside every number.

3. Tier movement is the site's changelog

The interesting events on this site are transitions: STATED → TRACED (a briefing document arrived and was integrated), DISCOVERED → TRACED (the founder confirmed), a trace row flipping absent → implemented (the estate built a gap). Each transition is a dated entry on /shipped/. A provenance site whose provenance changes silently would be self-refuting.

4. The comms channel has a special role here

On other sites /admin/comms.html carries build questions. Here it is also the briefing-document request queue: one thread per STATED entry (N1: Neil Peart — awaiting briefing; N2: Kevin Kelly — …), so the founder's stated plan ("in time I will create a detailed briefing document about each one") has a visible, addressable inbox. When a briefing lands, the thread closes and the entry's tier moves.

5. Infographics

The founder asked for infographics of people and ideas. Two rules from 04__: likenesses are original artwork only (never scraped photos); idea-infographics (the golden circle, the flow channel, evolution axes) are redrawn in house style with attribution, never copied. Every infographic's source data is a markdown/JSON twin — the house rule that a picture is a projection of data, not an original.

6. What this site does NOT do

No hosting of anchor works (talks, book excerpts, essays) — links only, per 04__. No hagiography — entries carry disagreements (the Semantic Web entry is the model). No rankings — influences are a graph, not a top-ten. And no silent scope creep into siblings: if an entry's material grows past resonance-and-trace, the material moves to the sibling site and the entry keeps the pointer.


This document is released under the Creative Commons Attribution 4.0 International licence (CC BY 4.0).


source: /briefs/04__boundaries-and-licensing.md