Skip to content

Latest commit

 

History

History
48 lines (43 loc) · 3.1 KB

COLLECTION_COUNTS.md

File metadata and controls

48 lines (43 loc) · 3.1 KB

Collection counts

data/exhibit-trust.json is the existing authoritative exhibit identity source. Its collection and interactive fields now also own collection totals; no parallel exhibit manifest is maintained. Trust classification is a separate field and does not determine collection membership.

  • Total exhibits: every unique registered exhibit ID/path, including people, institutions, history, culture, mathematics and cryptographic constructions.
  • Cipher and history exhibits (cipher-history): the main collection, including ciphers, machines, people, places and cultural subjects. This is not a count of encryption algorithms. New historical exhibits append IDs after §164; earlier IDs stay fixed.
  • Foundations exhibits (foundations): mathematical exhibits in the annex, §141–157. Its landing page is not an exhibit.
  • Protocols and constructions (protocols): §158–160, ECDSA, Shamir's Secret Sharing and Zero-Knowledge Proofs. The label includes signatures and secret sharing rather than implying all three are encryption ciphers or protocols.
  • Registered engines: keys in window.CipherEngines, evaluated without a DOM. Engines and exhibits are different units; some exhibits have bespoke math or visualization code outside this registry.
  • Interactive exhibits: entries marked interactive: true with an on-page working demo or controllable visualization. This includes the Rosetta and Dorabella viewers, mathematical and protocol interactives, and the historical demos. Static explanatory pages and navigation controls do not qualify. It is a count of exhibits with an interactive, not of buttons, widgets or reversible encryption algorithms. The cipher/history subset is reported separately. The previous main-collection demo count omitted working visualizations. Review also identified the Bible Code ELS controls and the annex’s PNT and entropy sliders; these count as interactives, not encryption engines.

Run npm run build:counts to render current totals into owned static surfaces: count slots, meta descriptions, roster headings, search fallback, README current inventory sections, and global “Exhibit X of Y” chips. node scripts/build-counts.js --check fails for stale output. Search entries carry canonical IDs and collection membership instead of counting descriptive kinds. tests/test-counts.js checks registrations, output freshness, idempotence and number preservation. Historical prose, dated release summaries, corpus record accounting and archived audit results are outside this generator's ownership.

Order after changing registrations: counts → trust labels → search/artifact metadata → minified bundles and service-worker version. Full validation remains npm test, npm run test:engines, npm run lint:html and corpus replay.

The comparison table is a curated subset and includes historical/cultural context. comparisonExhibits counts its actual registered rows, validates every row against the same collection identities, and generates its heading and metadata. Undated specifications remain undated and sort after dated entries.