The public knowledge base — how this tree is organized

Written normative pages + generated reference data, produced by the method documented under method/. New here? Read the overview first.

What this tree is

The public-facing reference core: every entity, its mechanics, its taxonomy class, its lineage and liquidity structure — generated from the data layer, never hand-curated into existence. A versioned research log (the provenance back-end) holds the errata history and working notes behind every supersession; this tree always reflects current-state.

Layout

• method/ — provenance + coverage policy (public Method pages)
• taxonomy/ — the semantics-derived taxonomy: dimension and class definitions, the machine rules, and how to reproduce them
• mechanics/ — written, normative mechanics pages: the five minter generations + cross-cutting topics (registry/tx.origin, _hu ladder, claim/publish, multipliers). The portal appends generated census statistics to these pages at build time — written text and generated numbers never live in the same artifact.
• deep/ — deep-dive tier: distillations for Tier-1 core entities (reserves, math family, Keys Of Ong, TeddyBear, Atropa, the FDIC/POOR spines). The README states the selection criterion; pDAI/Maker core is explicitly out (external systems stay at timeline rows + registry-page depth). Same generated-panel pattern as mechanics.
• overview.md — the orientation page: what the system is, the actors, how the pieces fit, the era windows, and a glossary. The natural entry point for a newcomer.
• reference/ — first-class written reference pages: shared infrastructure (MATH/RNG/math family), liquidity structure, admin surface (what the owner keys actually gate). Token and contract entity pages are generated portal output (/tokens/{address}, /contracts/{address}), not files in this tree.

Machine contracts

• Taxonomy fields are derived per-token from the census + registry data.
• The claims registry is generated from the same data layer as the pages.
• Shared infra state (MATH family) is generated as the one bundle both this portal and the AFFECTION portal consume.

Editorial rules

1. Normative content = [src] / [chain] only, each statement with a proof pointer and pinned block where relevant.
2. [ext] / [irc] / [ext-owner] never normative; community material lives in a labeled Community-record layer only.
3. Coverage is completeness-by-construction (see method/coverage.md).
4. Nothing from the quarantine layer ships here, ever.
5. Actors: functional attribution only (deployer/creator fields, computed role tables). No person pages, no identity claims.