# EM-0036 — Four-case How We Know library

- Object ID: `em:documentation:sha256:98f11be079ee8b5b5e4434bd8b4223909919a17afe4dd44478d09a1e001d2f8d`
- Kind: `documentation`
- Repository path: [`docs/execution-plans/EM-0036.md`](https://github.com/yoheinakajima/epistemedia/blob/f92846570180dfa4511263f8ba98ecd18f7772c9/docs/execution-plans/EM-0036.md)
- Content digest: `0294d6f86c04a17f9a17e46438c7326a6cce0d5437ac0772cefbe33ded73bda5`

**Also filed under:** [Disclosure and Public Projection](https://epistemedia.org/topics/disclosure/), [Epistemedia](https://epistemedia.org/topics/epistemedia/), [Epistemic Mesh Protocol](https://epistemedia.org/topics/epistemic-mesh/), [Sovereign Realm Federation](https://epistemedia.org/topics/federation/), [Autonomous Governance](https://epistemedia.org/topics/governance/), [Knowledge Objects](https://epistemedia.org/topics/knowledge-objects/), [Human and Agent Interfaces](https://epistemedia.org/topics/public-interfaces/), [Releases and Reproducibility](https://epistemedia.org/topics/releases/), [Research Program](https://epistemedia.org/topics/research-program/), [Security and Adversarial Robustness](https://epistemedia.org/topics/security/)

## Source content

# EM-0036 — Four-case How We Know library

## Objective

Admit the exact independently reviewed Case 003 and Case 004 dossiers into the public application
library without changing accepted research bytes or forcing either case into an older case's count
grammar.

## Bound inputs

- Case 003 dossier: 224,792 bytes, SHA-256
  `32c4457b3823237b2f988a26d51b2f6222af8060e662993524aff1c1a5d79e5d`.
- Case 003 independent receipt: 13,568 bytes, SHA-256
  `f35f8c093778ad1cdeafc57746575f1a82d12fe6e6fbf0d1d4a546e3c5296e1e`.
- Case 004 dossier: 176,885 bytes, SHA-256
  `44e7bf407091d7665d4f1ab2aabc285255364e790e2d17520913f9ba9c57b418`.
- Case 004 independent receipt: 71,345 bytes, SHA-256
  `0bb9184660deb3ef83addbcdb7be61d49f234026d6d484747b51628a4f8f087f`.

## Implementation

1. Add exact manifests for Cases 003 and 004 with dossier, receipt, reviewer, head, tree, count,
   view, practical-reading, and lexicon bindings.
2. Add one strict bounded-proposition adapter that derives public views from the accepted dossier,
   rejects receipt drift and unsupported reviewer state, and preserves per-case ledgers.
3. Extend static, Markdown, review, share-card, REST, MCP, CLI, discovery, sitemap, and homepage
   surfaces to four deterministically ordered cases.
4. Add adversarial identity and receipt tests, semantic closure tests, complete parity tests, and
   desktop/mobile no-JavaScript accessibility checks.

## Boundaries

The task does not modify research packets, promote the reversible adapter into a normative schema,
activate hosted API or MCP services, add a future case, or claim deployment before Pages and live
route read-back. Case 003 is a bounded historical score-and-comparison case; Case 004 is a bounded
history and propagation-lineage case, not a universal law of communication.

## Author validation

- Exact Case 003 and Case 004 dossier and independent-receipt identities load fail closed.
- Focused adapter, adversarial, parity, discovery, and interface tests pass.
- Full `PYTHONPATH=src make check PYTHON=.venv/bin/python` passes accepted-input validation, the
  complete test suite, disclosure audit, independent deterministic rebuild, and unchanged source
  state.
- Local desktop 1440×900 and mobile 390×844 checks observe one H1, zero scripts, no horizontal
  overflow, visible policy divergence, a 3px focus outline, and click expansion. The browser driver
  focused native `summary` controls but did not dispatch Enter activation; exact-head independent
  review and post-deployment read-back must retain this bounded keystroke observation rather than
  infer a pass.

## Independent-review correction

The first exact-head review correctly rejected the generic human projection because its featured
relations displayed a generic note and source cards while leaving the resolved proposition,
assertion, lineage, typed dependence edges, calculation register, and count-member provenance only
in JSON. The forward correction now resolves span-to-proposition relations fail closed, renders the
accepted proposition and identity chain, exposes calculation equations, inputs, resolved cells,
dependencies, uncertainty, results, and exact source receipts, and retains each ledger member's
object and basis work/edition/span. Adversarial projection tests assert these records in both HTML
and Markdown. Accepted research dossiers and their independent receipts remain byte-identical.