# Paper 18 related-work ledger

Status: paper-local author ledger; not peer review

Date: 23 July 2026

This ledger records what the cited work establishes and what it cannot
establish for MDReader. All capability claims remain provisional because
`apps/mdreader/` is empty.

## Markdown syntax and source authority

| Sources | Established result used | MDReader boundary |
|---|---|---|
| `commonmark0312`, `gfm` | CommonMark defines a core Markdown syntax; GFM specifies tables, task-list items, strikethrough, and autolinks. | Neither source defines CatDB projections, stable document identity, typed overlays, semantic diff, provenance, or an MDReader parser implementation. Repository dialects still need explicit profiles and fixtures. |
| `catdbcontext` | The immutable context export proposes ordinary-file authority, profiled Markdown viewing, hybrid retrieval, source locations, and inspectable relationships. | It is a design transcript, not application code, a product result, or comparative evidence. Its competitor references are not used as implementation evidence. |

## Categorical mappings, provenance, and versioning

| Sources | Established result used | MDReader boundary |
|---|---|---|
| `spivak2012` | Schema mappings can induce functorial data-migration operations. | Paper 18 does not extend the theory and has no UI, API, Sigma, Pi, general Kan extension, or physical migration result. |
| `green2007`, `prov-o` | Provenance semirings explain input contribution to query results; PROV-O supplies entity, activity, and agent vocabulary. | The proposed record profile is narrower. No provenance executor, completeness theorem, privacy result, or export implementation exists. |
| `decibel2016`, `orpheusdb2017` | Relational systems can represent and query versioned datasets and branches. | These systems are prior art, not MDReader components. Paper 18 has no historical-query engine or SQLite revision store. |
| `prism2008` | Real database schemas evolve and require explicit change management. | The paper cites this as motivation. It does not infer stable Markdown identity or provide an executable semantic-diff engine. |

## Local storage and optional synchronization

| Sources | Established result used | MDReader boundary |
|---|---|---|
| `sqlitetransactions` | SQLite defines transactional behavior. | Transactions alone do not establish atomic logical corpus revisions, application recovery, or a current SQLite backend. |
| `sqliteforeignkeys` | SQLite documents foreign-key enforcement and its per-connection enabling requirement. | Foreign keys do not replace overlay, mapping, provenance, or logical-revision invariants. No connection policy exists in the repository. |
| `sqlitestrict` | STRICT tables narrow SQLite storage typing. | STRICT tables are a proposed storage choice, not evidence that application types or semantic obligations are enforced. |
| `sqlitefts5` | FTS5 provides a local full-text indexing facility. | FTS5 is a candidate lexical baseline. No index, query result, recovery result, or retrieval measurement exists. |
| `pglogical`, `pglogicalrestrictions`, `pglogicalconflicts` | PostgreSQL documents logical replication, schema restrictions, and conflict behavior. | These documents motivate explicit compatibility and conflict contracts. No driver, protocol, authorization model, fault harness, or synchronization result exists. |

## Retrieval

| Sources | Established result used | MDReader boundary |
|---|---|---|
| `robertson2009bm25` | BM25 is an established lexical ranking baseline. | No lexical index or ranking result exists in MDReader. |
| `malkov2018` | HNSW supports approximate nearest-neighbor search. | The paper proposes embedding candidates only. It claims no vector index, recall, latency, or semantic authority. |
| `cormack2009rrf` | Reciprocal rank fusion combines ranked lists. | Fusion is prior art. The proposed hybrid pipeline has no implementation or superiority result. |

## Visualization, usability, and accessibility

| Sources | Established result used | MDReader boundary |
|---|---|---|
| `munzner2009` | The nested model separates domain, abstraction, encoding/interaction, and algorithm validation. | It supplies an evaluation discipline, not evidence that the proposed graphs or inspectors support user tasks. |
| `iso9241` | ISO 9241-11 frames usability in terms of effectiveness, efficiency, and satisfaction in context. | Paper 18 reports no participants, task outcomes, timing, errors, or usability claim. |
| `brooke1996sus` | SUS is a compact subjective usability instrument. | SUS cannot replace task evidence. No SUS data exist. |
| `wcag22`, `aria-apg` | WCAG 2.2 defines accessibility success criteria; ARIA APG gives widget patterns. | Citations do not establish conformance. No browser UI, keyboard record, screen-reader record, reflow test, or accessibility audit exists. |

## CatDB program artifacts

| Sources | Established result used | MDReader boundary |
|---|---|---|
| `catdbmanifest` | The manifest asks whether one Markdown corpus can support multiple typed projections with visible mappings and provenance. | The target and deliverables are research contracts. They are not observed application behavior. |
| `catdbroadmap` | Slice 14 defines future server/API/MDReader gates and leaves the slice `OPEN`. | Exit commands are intentions. The package, API, corpus, equivalence fixtures, E2E suite, and review artifacts are absent. |
| `catdbslice4` | The accepted historical Slice 4 release records 29 semantic and four structural-only comparisons over its closed 33-case corpus, with zero disagreements. | This is semantic-library evidence only. The corrected manuscript reports the current later-fixture result `42/29/13/0` separately from the accepted release. Neither result establishes a parser, database, API, UI, inspector, search system, sync system, or user study. |

## Disposition

Paper 18's defensible candidate contribution is a source-preserving,
application-level acceptance contract that places Markdown projections, typed
overlays, categorical mappings, evidence status, provenance, diff, local
operation, and optional synchronization under one staged gate. Each component
has substantial prior art. The combination is a design hypothesis until a
working vertical application and direct comparisons exist.