apol commited on
Commit
122b842
·
verified ·
1 Parent(s): 9ba7c7e

Fix Hub Viewer conversion fan-out

Browse files

Expose one stable research_view configuration while preserving all 26 CSV/Parquet tables. Keep the frozen v0.4.0 release immutable and separate historical-package verification from mutable-main parity.

CHANGELOG.md CHANGED
@@ -1,5 +1,11 @@
1
  # Changelog
2
 
 
 
 
 
 
 
3
  ## 0.4.0 — 2026-08-15
4
 
5
  - Defined the release unambiguously as a purposive, exploratory evidence index: not proof of agency transfer, not a prevalence sample, and not a causal estimate of election effects.
 
1
  # Changelog
2
 
3
+ ## Unreleased — Viewer metadata hotfix
4
+
5
+ - Limited the mutable `main` dataset card to one interactive `research_view` configuration after Hugging Face's 26-way conversion fan-out repeatedly hit shared-branch locks. All 26 canonical CSV and Parquet tables remain unchanged and directly downloadable.
6
+ - Replaced links to retired Data Studio configurations with frozen file links and added a direct loader recipe for any table.
7
+ - Kept the `v0.4.0` tag and its ZIP, workbook and checksums immutable; release verification now distinguishes the frozen package from later documentation-only changes on `main`.
8
+
9
  ## 0.4.0 — 2026-08-15
10
 
11
  - Defined the release unambiguously as a purposive, exploratory evidence index: not proof of agency transfer, not a prevalence sample, and not a causal estimate of election effects.
README.md CHANGED
@@ -3,8 +3,6 @@ license: cc-by-4.0
3
  language:
4
  - en
5
  pretty_name: AI, Elections and Agency Transfer Evidence Index
6
- size_categories:
7
- - 1K<n<10K
8
  tags:
9
  - elections
10
  - political-disinformation
@@ -16,62 +14,14 @@ configs:
16
  - config_name: research_view
17
  default: true
18
  data_files: [{split: data, path: parquet/research_view.parquet}]
19
- - config_name: case_catalog
20
- data_files: [{split: data, path: parquet/case_catalog.parquet}]
21
- - config_name: pathways
22
- data_files: [{split: data, path: parquet/pathways.parquet}]
23
- - config_name: observations
24
- data_files: [{split: data, path: parquet/observations.parquet}]
25
- - config_name: model_evaluations
26
- data_files: [{split: data, path: parquet/model_evaluations.parquet}]
27
- - config_name: cases
28
- data_files: [{split: data, path: parquet/cases.parquet}]
29
- - config_name: claims
30
- data_files: [{split: data, path: parquet/claims.parquet}]
31
- - config_name: sources
32
- data_files: [{split: data, path: parquet/sources.parquet}]
33
- - config_name: claim_evidence
34
- data_files: [{split: data, path: parquet/claim_evidence.parquet}]
35
- - config_name: analytic_record_claims
36
- data_files: [{split: data, path: parquet/analytic_record_claims.parquet}]
37
- - config_name: events
38
- data_files: [{split: data, path: parquet/events.parquet}]
39
- - config_name: case_sources
40
- data_files: [{split: data, path: parquet/case_sources.parquet}]
41
- - config_name: case_actors
42
- data_files: [{split: data, path: parquet/case_actors.parquet}]
43
- - config_name: technology_uses
44
- data_files: [{split: data, path: parquet/technology_uses.parquet}]
45
- - config_name: content_items
46
- data_files: [{split: data, path: parquet/content_items.parquet}]
47
- - config_name: case_elections
48
- data_files: [{split: data, path: parquet/case_elections.parquet}]
49
- - config_name: election_sources
50
- data_files: [{split: data, path: parquet/election_sources.parquet}]
51
- - config_name: sampling_frame
52
- data_files: [{split: data, path: parquet/sampling_frame.parquet}]
53
- - config_name: coverage_summary
54
- data_files: [{split: data, path: parquet/coverage_summary.parquet}]
55
- - config_name: candidates
56
- data_files: [{split: data, path: parquet/candidates.parquet}]
57
- - config_name: watchlist
58
- data_files: [{split: data, path: parquet/watchlist.parquet}]
59
- - config_name: official_elections
60
- data_files: [{split: data, path: parquet/official_elections.parquet}]
61
- - config_name: official_election_metrics
62
- data_files: [{split: data, path: parquet/official_election_metrics.parquet}]
63
- - config_name: official_turnout
64
- data_files: [{split: data, path: parquet/official_turnout.parquet}]
65
- - config_name: official_results
66
- data_files: [{split: data, path: parquet/official_results.parquet}]
67
- - config_name: official_data_sources
68
- data_files: [{split: data, path: parquet/official_data_sources.parquet}]
69
  ---
70
 
71
  # AI, Elections and Agency Transfer Evidence Index
72
 
73
  **Version 0.4.0 · released 15 August 2026 · research cutoff 12 August 2026**
74
 
 
 
75
  This is a **purposive, exploratory evidence index**. It makes claims, sources, counterevidence, uncertainty, mechanisms, and selected measurements easier to inspect. It does **not** empirically prove the agency-transfer hypothesis, estimate how common election manipulation is, or identify votes or election outcomes caused by AI.
76
 
77
  The repository slug is inherited. The canonical title is the one above: the index also contains preparedness files, boundary comparisons, screening leads, official electoral context, and empirical model studies. Inclusion is not a finding that every record is manipulative or harmful.
@@ -82,7 +32,7 @@ The repository slug is inherited. The canonical title is the one above: the inde
82
  - [Read the Spanish quick guide](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/docs/guia-rapida-es.md) — concise interpretation guide.
83
  - [Read the methodology](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/docs/methodology.md) — units, coding and limitations.
84
  - [Read the v0.4.0 red-team report](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/docs/red-team-v0.4.0.md) — findings, fixes and residual limits.
85
- - [Inspect the claim–evidence table](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/viewer/claim_evidence/data) — trace claims to source records.
86
  - [Download the complete v0.4.0 package](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/resolve/v0.4.0/release/agency-transfer-election-cases-v0.4.0.zip?download=true) — CSV, Parquet, documentation, ledgers, scripts and workbook.
87
  - [Download the v0.4.0 Excel workbook](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/resolve/v0.4.0/release/agency-transfer-election-evidence-index-v0.4.0.xlsx?download=true) — navigation and QA convenience artifact.
88
  - [Verify release checksums](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/release/SHA256SUMS) — hashes for the frozen ZIP and workbook.
@@ -111,9 +61,9 @@ These counts describe the generated v0.4.0 build and must be refreshed if that b
111
 
112
  The strongest pathway conclusion is deliberately limited: 4 records show an observed control mechanism, 2 show an observed attempted transfer, 2 remain insufficient, and 2 are preparedness records rather than incidents. **No record codes observed agency change or observed agency preservation/extension.** These categories describe evidence ceilings, not severity scores.
113
 
114
- ## The 26 configurations
115
 
116
- | Use | Configurations |
117
  |---|---|
118
  | Orient and scope | `research_view`, `case_catalog`, `coverage_summary`, `sampling_frame` |
119
  | Claims and evidence | `cases`, `claims`, `sources`, `claim_evidence`, `analytic_record_claims` |
@@ -122,7 +72,9 @@ The strongest pathway conclusion is deliberately limited: 4 records show an obse
122
  | Discovery | `candidates`, `watchlist` |
123
  | Empirical capability evidence | `model_evaluations` |
124
 
125
- Hugging Face Data Studio reads the native Parquet files in `parquet/`. Equivalent source tables are in `data/`; relational types, keys and foreign keys are declared in [`data/datapackage.json`](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/data/datapackage.json).
 
 
126
 
127
  Files under `research/` are transparent build inputs and historical working ledgers, not canonical v0.4.0 tables. Some retain superseded draft annotations; see [`research/README.md`](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/research/README.md) before using them.
128
 
@@ -141,6 +93,16 @@ view = load_dataset(
141
  )
142
  ```
143
 
 
 
 
 
 
 
 
 
 
 
144
  ### pandas
145
 
146
  ```python
@@ -229,11 +191,15 @@ After the externally generated workbook is frozen in `release/`, maintainers run
229
 
230
  ```bash
231
  npm run build:package
232
- npm run verify:package
233
  npm run build:release-manifest
234
  npm run validate
235
  ```
236
 
 
 
 
 
237
  `npm run audit:links` is a separate, time-dependent network check; it is not evidence that a blocked or rate-limited page is absent. Within the declared runtime and inputs, the deterministic scripts rebuild and check the tabular and packaged artifacts they cover. That remains narrower than end-to-end source reproducibility: external pages are not frozen, 47 pinpoints cover only a factual premise or part of their claim, 20 relations have no verified pinpoint, and automated QA does not validate source truth.
238
 
239
  The Excel workbook is a convenience artifact, not the canonical analytic source. Its builder requires the `@oai/artifact-tool` runtime, which is not part of the public `package.json`; use CSV or Parquet unless that environment is available. The ZIP and workbook should be verified against the release manifest, not assumed equivalent merely because their filenames contain the same version.
 
3
  language:
4
  - en
5
  pretty_name: AI, Elections and Agency Transfer Evidence Index
 
 
6
  tags:
7
  - elections
8
  - political-disinformation
 
14
  - config_name: research_view
15
  default: true
16
  data_files: [{split: data, path: parquet/research_view.parquet}]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
17
  ---
18
 
19
  # AI, Elections and Agency Transfer Evidence Index
20
 
21
  **Version 0.4.0 · released 15 August 2026 · research cutoff 12 August 2026**
22
 
23
+ **Viewer hotfix on `main`, 15 August 2026:** Data Studio now exposes one orientation view. The frozen `v0.4.0` tag, its 26 tables, ZIP, workbook and checksums are unchanged.
24
+
25
  This is a **purposive, exploratory evidence index**. It makes claims, sources, counterevidence, uncertainty, mechanisms, and selected measurements easier to inspect. It does **not** empirically prove the agency-transfer hypothesis, estimate how common election manipulation is, or identify votes or election outcomes caused by AI.
26
 
27
  The repository slug is inherited. The canonical title is the one above: the index also contains preparedness files, boundary comparisons, screening leads, official electoral context, and empirical model studies. Inclusion is not a finding that every record is manipulative or harmful.
 
32
  - [Read the Spanish quick guide](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/docs/guia-rapida-es.md) — concise interpretation guide.
33
  - [Read the methodology](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/docs/methodology.md) — units, coding and limitations.
34
  - [Read the v0.4.0 red-team report](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/docs/red-team-v0.4.0.md) — findings, fixes and residual limits.
35
+ - [Inspect the claim–evidence table](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/data/claim_evidence.csv) — trace claims to source records.
36
  - [Download the complete v0.4.0 package](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/resolve/v0.4.0/release/agency-transfer-election-cases-v0.4.0.zip?download=true) — CSV, Parquet, documentation, ledgers, scripts and workbook.
37
  - [Download the v0.4.0 Excel workbook](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/resolve/v0.4.0/release/agency-transfer-election-evidence-index-v0.4.0.xlsx?download=true) — navigation and QA convenience artifact.
38
  - [Verify release checksums](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/release/SHA256SUMS) — hashes for the frozen ZIP and workbook.
 
61
 
62
  The strongest pathway conclusion is deliberately limited: 4 records show an observed control mechanism, 2 show an observed attempted transfer, 2 remain insufficient, and 2 are preparedness records rather than incidents. **No record codes observed agency change or observed agency preservation/extension.** These categories describe evidence ceilings, not severity scores.
63
 
64
+ ## The 26 tables
65
 
66
+ | Use | Table names |
67
  |---|---|
68
  | Orient and scope | `research_view`, `case_catalog`, `coverage_summary`, `sampling_frame` |
69
  | Claims and evidence | `cases`, `claims`, `sources`, `claim_evidence`, `analytic_record_claims` |
 
72
  | Discovery | `candidates`, `watchlist` |
73
  | Empirical capability evidence | `model_evaluations` |
74
 
75
+ All 26 canonical tables are available as CSV in [`data/`](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/tree/v0.4.0/data) and as native Parquet in [`parquet/`](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/tree/v0.4.0/parquet). Relational types, keys and foreign keys are declared in [`data/datapackage.json`](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/data/datapackage.json).
76
+
77
+ Only `research_view` is declared as an interactive Data Studio configuration. This keeps the public preview small and avoids conversion-lock contention among 26 unrelated schemas. It does not remove or merge any table; load the other 25 directly from their CSV or Parquet files.
78
 
79
  Files under `research/` are transparent build inputs and historical working ledgers, not canonical v0.4.0 tables. Some retain superseded draft annotations; see [`research/README.md`](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/main/research/README.md) before using them.
80
 
 
93
  )
94
  ```
95
 
96
+ To load any of the 26 tables directly, replace `claim_evidence` with a table name from the map above:
97
+
98
+ ```python
99
+ from datasets import load_dataset
100
+
101
+ table = "claim_evidence"
102
+ url = f"https://huggingface.co/datasets/apol/ai-election-manipulation-cases/resolve/v0.4.0/parquet/{table}.parquet"
103
+ records = load_dataset("parquet", data_files={"data": url}, split="data")
104
+ ```
105
+
106
  ### pandas
107
 
108
  ```python
 
191
 
192
  ```bash
193
  npm run build:package
194
+ npm run verify:package:checkout
195
  npm run build:release-manifest
196
  npm run validate
197
  ```
198
 
199
+ `npm run verify:package` verifies the frozen ZIP internally against its embedded manifest and checksums. Use `npm run verify:package:checkout` only while cutting a release to additionally require byte-for-byte parity with the current checkout; later documentation-only changes on mutable `main` do not rewrite an existing release.
200
+
201
+ Do not run `build:package` to overwrite an already tagged version from mutable `main`; bump the version and cut a new tag first.
202
+
203
  `npm run audit:links` is a separate, time-dependent network check; it is not evidence that a blocked or rate-limited page is absent. Within the declared runtime and inputs, the deterministic scripts rebuild and check the tabular and packaged artifacts they cover. That remains narrower than end-to-end source reproducibility: external pages are not frozen, 47 pinpoints cover only a factual premise or part of their claim, 20 relations have no verified pinpoint, and automated QA does not validate source truth.
204
 
205
  The Excel workbook is a convenience artifact, not the canonical analytic source. Its builder requires the `@oai/artifact-tool` runtime, which is not part of the public `package.json`; use CSV or Parquet unless that environment is available. The ZIP and workbook should be verified against the release manifest, not assumed equivalent merely because their filenames contain the same version.
docs/guia-rapida-es.md CHANGED
@@ -9,16 +9,16 @@ Los datos y campos codificados están en inglés. Esta guía y algunos metadatos
9
  ## Empiece aquí
10
 
11
  1. [Vista de investigación](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/viewer/research_view/data): 10 registros resumidos.
12
- 2. [Catálogo completo](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/viewer/case_catalog/data): separa núcleo, candidatos y estudios de modelos.
13
- 3. [Vías analíticas](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/viewer/pathways/data): mecanismo, control y estatus epistémico.
14
- 4. [Afirmaciones](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/viewer/claims/data) y [relaciones de evidencia](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/viewer/claim_evidence/data): base para cualquier uso factual.
15
- 5. [Observaciones](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/viewer/observations/data): cantidades seleccionadas con unidad, denominador y estatus causal.
16
 
17
  El proyecto forma parte de la [colección Agency Transfer](https://huggingface.co/collections/apol/agency-transfer).
18
 
19
- ## Mapa de las 26 configuraciones
20
 
21
- | Uso | Configuraciones |
22
  |---|---|
23
  | Orientación y cobertura | `research_view`, `case_catalog`, `coverage_summary`, `sampling_frame` |
24
  | Afirmaciones y evidencia | `cases`, `claims`, `sources`, `claim_evidence`, `analytic_record_claims` |
@@ -27,6 +27,8 @@ El proyecto forma parte de la [colección Agency Transfer](https://huggingface.c
27
  | Descubrimiento | `candidates`, `watchlist` |
28
  | Estudios de modelos | `model_evaluations` |
29
 
 
 
30
  ## Cargar una versión fija
31
 
32
  ```python
@@ -40,6 +42,16 @@ vista = load_dataset(
40
  )
41
  ```
42
 
 
 
 
 
 
 
 
 
 
 
43
  Con pandas o Polars:
44
 
45
  ```python
 
9
  ## Empiece aquí
10
 
11
  1. [Vista de investigación](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/viewer/research_view/data): 10 registros resumidos.
12
+ 2. [Catálogo completo](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/data/case_catalog.csv): separa núcleo, candidatos y estudios de modelos.
13
+ 3. [Vías analíticas](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/data/pathways.csv): mecanismo, control y estatus epistémico.
14
+ 4. [Afirmaciones](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/data/claims.csv) y [relaciones de evidencia](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/data/claim_evidence.csv): base para cualquier uso factual.
15
+ 5. [Observaciones](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/blob/v0.4.0/data/observations.csv): cantidades seleccionadas con unidad, denominador y estatus causal.
16
 
17
  El proyecto forma parte de la [colección Agency Transfer](https://huggingface.co/collections/apol/agency-transfer).
18
 
19
+ ## Mapa de las 26 tablas
20
 
21
+ | Uso | Tablas |
22
  |---|---|
23
  | Orientación y cobertura | `research_view`, `case_catalog`, `coverage_summary`, `sampling_frame` |
24
  | Afirmaciones y evidencia | `cases`, `claims`, `sources`, `claim_evidence`, `analytic_record_claims` |
 
27
  | Descubrimiento | `candidates`, `watchlist` |
28
  | Estudios de modelos | `model_evaluations` |
29
 
30
+ Data Studio muestra de forma interactiva solo `research_view`. Las 26 tablas completas siguen disponibles en [`data/`](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/tree/v0.4.0/data) como CSV y en [`parquet/`](https://huggingface.co/datasets/apol/ai-election-manipulation-cases/tree/v0.4.0/parquet) como Parquet. Esta separación evita que 26 esquemas distintos compitan durante la conversión del Viewer.
31
+
32
  ## Cargar una versión fija
33
 
34
  ```python
 
42
  )
43
  ```
44
 
45
+ Para cargar cualquier tabla por nombre:
46
+
47
+ ```python
48
+ from datasets import load_dataset
49
+
50
+ tabla = "claim_evidence"
51
+ url = f"https://huggingface.co/datasets/apol/ai-election-manipulation-cases/resolve/v0.4.0/parquet/{tabla}.parquet"
52
+ datos = load_dataset("parquet", data_files={"data": url}, split="data")
53
+ ```
54
+
55
  Con pandas o Polars:
56
 
57
  ```python
docs/red-team-v0.4.0.md CHANGED
@@ -1,5 +1,7 @@
1
  # Red-team report for v0.4.0
2
 
 
 
3
  ## Verdict
4
 
5
  Version 0.4.0 is an exploratory evidence index, not an empirical test of the agency-transfer hypothesis. The red-team found that v0.3.1 was structurally useful but its central pathway field could make inferred control shifts look observed. The release was therefore recoded rather than patched cosmetically.
@@ -20,7 +22,7 @@ Version 0.4.0 is an exploratory evidence index, not an empirical test of the age
20
  | The validator could pass stale audits, a forged Parquet manifest or silent type coercion | The v0.4 validator compares stored and live audits, verifies all 26 CSV/Parquet schemas and hashes, rejects invalid coercion, and has isolated mutation tests. |
21
  | Legacy scripts could overwrite current outputs | The old `audit-data.mjs` entry point is now a read-only compatibility wrapper; the v0.4 auditor retains its historical `audit-v03.mjs` filename, and the obsolete workbook entry point fails before writing. |
22
  | Public version identity and packaged metadata drifted | v0.4 aligns README, CFF, package, datapackage, Parquet manifest, workbook and ZIP; package members and release artifacts receive explicit SHA-256 manifests. |
23
- | Dataset-card onboarding and link health were weak | Added clickable entry points, a 26-config map, English-language metadata, a Spanish guide, pinned Datasets/pandas/Polars recipes and corrected known dead links. |
24
 
25
  ## What remains unresolved
26
 
@@ -41,6 +43,6 @@ A publishable v0.4.0 build must satisfy all of the following:
41
  5. a visually reviewed workbook whose QA sheet reports the same audit result;
42
  6. a safe ZIP whose member hashes match both its internal manifest and the checked-out files;
43
  7. external SHA-256 checksums for the ZIP and workbook;
44
- 8. a fixed Hub revision, public downloads and anonymous Data Studio checks for all 26 configurations.
45
 
46
  This report records model-assisted red-team work. It is not a peer review, source-by-source adjudication, legal opinion or substitute for independent replication.
 
1
  # Red-team report for v0.4.0
2
 
3
+ > Post-release Viewer note, 15 August 2026: the original 26-configuration Data Studio fan-out repeatedly hit shared conversion locks. The mutable `main` card now exposes only `research_view`; all frozen v0.4.0 CSV, Parquet and release artifacts remain unchanged. Acceptance criterion 8 below reflects that operational correction.
4
+
5
  ## Verdict
6
 
7
  Version 0.4.0 is an exploratory evidence index, not an empirical test of the agency-transfer hypothesis. The red-team found that v0.3.1 was structurally useful but its central pathway field could make inferred control shifts look observed. The release was therefore recoded rather than patched cosmetically.
 
22
  | The validator could pass stale audits, a forged Parquet manifest or silent type coercion | The v0.4 validator compares stored and live audits, verifies all 26 CSV/Parquet schemas and hashes, rejects invalid coercion, and has isolated mutation tests. |
23
  | Legacy scripts could overwrite current outputs | The old `audit-data.mjs` entry point is now a read-only compatibility wrapper; the v0.4 auditor retains its historical `audit-v03.mjs` filename, and the obsolete workbook entry point fails before writing. |
24
  | Public version identity and packaged metadata drifted | v0.4 aligns README, CFF, package, datapackage, Parquet manifest, workbook and ZIP; package members and release artifacts receive explicit SHA-256 manifests. |
25
+ | Dataset-card onboarding and link health were weak | Added clickable entry points, a 26-table map, English-language metadata, a Spanish guide, pinned Datasets/pandas/Polars recipes and corrected known dead links. Data Studio exposes one orientation view; all tables remain directly loadable. |
26
 
27
  ## What remains unresolved
28
 
 
43
  5. a visually reviewed workbook whose QA sheet reports the same audit result;
44
  6. a safe ZIP whose member hashes match both its internal manifest and the checked-out files;
45
  7. external SHA-256 checksums for the ZIP and workbook;
46
+ 8. a fixed Hub revision, public downloads for all 26 tables and an anonymous Data Studio check for the `research_view` orientation configuration.
47
 
48
  This report records model-assisted red-team work. It is not a peer review, source-by-source adjudication, legal opinion or substitute for independent replication.
package.json CHANGED
@@ -15,6 +15,7 @@
15
  "build:parquet": "node scripts/build-parquet.mjs",
16
  "build:package": "node scripts/build-package.mjs",
17
  "verify:package": "node scripts/verify-package.mjs",
 
18
  "build:release-manifest": "node scripts/build-release-manifest.mjs",
19
  "audit:links": "node scripts/audit-links.mjs",
20
  "test:hardening": "node scripts/test-hardening.mjs"
 
15
  "build:parquet": "node scripts/build-parquet.mjs",
16
  "build:package": "node scripts/build-package.mjs",
17
  "verify:package": "node scripts/verify-package.mjs",
18
+ "verify:package:checkout": "node scripts/verify-package.mjs --checkout-parity",
19
  "build:release-manifest": "node scripts/build-release-manifest.mjs",
20
  "audit:links": "node scripts/audit-links.mjs",
21
  "test:hardening": "node scripts/test-hardening.mjs"
scripts/build-release-manifest.mjs CHANGED
@@ -11,12 +11,15 @@
11
  import fs from "node:fs/promises";
12
  import path from "node:path";
13
  import { createHash } from "node:crypto";
 
14
  import { fileURLToPath } from "node:url";
15
 
16
  const RELEASE_VERSION = "0.4.0";
17
  const here = path.dirname(fileURLToPath(import.meta.url));
18
  const root = path.resolve(here, "..");
19
  const releaseDir = path.join(root, "release");
 
 
20
  const requested = process.argv.slice(2);
21
  const releasePattern = new RegExp(`-v${RELEASE_VERSION.replaceAll(".", "\\.")}\\.(?:zip|xlsx)$`);
22
  const packageFiles = ["README.md", "CITATION.cff", "package.json", "data/datapackage.json"];
@@ -33,6 +36,18 @@ function safeArtifactName(value) {
33
  return name;
34
  }
35
 
 
 
 
 
 
 
 
 
 
 
 
 
36
  let artifactNames;
37
  if (requested.length) {
38
  artifactNames = requested.map(safeArtifactName);
@@ -43,6 +58,9 @@ artifactNames = [...new Set(artifactNames)].sort();
43
  if (!artifactNames.length) {
44
  throw new Error(`No v${RELEASE_VERSION} ZIP/XLSX artifacts found in release/. Build them before checksums.`);
45
  }
 
 
 
46
 
47
  const artifacts = [];
48
  for (const name of artifactNames) {
@@ -51,8 +69,9 @@ for (const name of artifactNames) {
51
  }
52
 
53
  const packageFileEntries = [];
 
54
  for (const name of packageFiles) {
55
- const buffer = await fs.readFile(path.join(root, name));
56
  packageFileEntries.push({ name, bytes: buffer.byteLength, sha256: sha256(buffer) });
57
  }
58
 
 
11
  import fs from "node:fs/promises";
12
  import path from "node:path";
13
  import { createHash } from "node:crypto";
14
+ import { spawnSync } from "node:child_process";
15
  import { fileURLToPath } from "node:url";
16
 
17
  const RELEASE_VERSION = "0.4.0";
18
  const here = path.dirname(fileURLToPath(import.meta.url));
19
  const root = path.resolve(here, "..");
20
  const releaseDir = path.join(root, "release");
21
+ const packageRoot = `agency-transfer-election-cases-v${RELEASE_VERSION}`;
22
+ const packageZipName = `${packageRoot}.zip`;
23
  const requested = process.argv.slice(2);
24
  const releasePattern = new RegExp(`-v${RELEASE_VERSION.replaceAll(".", "\\.")}\\.(?:zip|xlsx)$`);
25
  const packageFiles = ["README.md", "CITATION.cff", "package.json", "data/datapackage.json"];
 
36
  return name;
37
  }
38
 
39
+ function readZipMember(zipPath, member) {
40
+ const result = spawnSync("unzip", ["-p", zipPath, member], {
41
+ encoding: null,
42
+ maxBuffer: 128 * 1024 * 1024
43
+ });
44
+ if (result.error) throw result.error;
45
+ if (result.status !== 0) {
46
+ throw new Error(`Could not read ${member} from ${path.basename(zipPath)}: ${String(result.stderr || "").trim()}`);
47
+ }
48
+ return Buffer.from(result.stdout);
49
+ }
50
+
51
  let artifactNames;
52
  if (requested.length) {
53
  artifactNames = requested.map(safeArtifactName);
 
58
  if (!artifactNames.length) {
59
  throw new Error(`No v${RELEASE_VERSION} ZIP/XLSX artifacts found in release/. Build them before checksums.`);
60
  }
61
+ if (!artifactNames.includes(packageZipName)) {
62
+ throw new Error(`The frozen package ${packageZipName} is required to build release package-file checksums.`);
63
+ }
64
 
65
  const artifacts = [];
66
  for (const name of artifactNames) {
 
69
  }
70
 
71
  const packageFileEntries = [];
72
+ const packageZipPath = path.join(releaseDir, packageZipName);
73
  for (const name of packageFiles) {
74
+ const buffer = readZipMember(packageZipPath, `${packageRoot}/${name}`);
75
  packageFileEntries.push({ name, bytes: buffer.byteLength, sha256: sha256(buffer) });
76
  }
77
 
scripts/test-hardening.mjs CHANGED
@@ -167,10 +167,43 @@ try {
167
  assert(packageRebuild.status === 0 && firstZipHash === secondZipHash, "Repeated package build was not byte-deterministic");
168
  const packageVerification = run(repo, "verify-package.mjs");
169
  assert(packageVerification.status === 0, `Package member verification failed.\n${packageVerification.stdout}\n${packageVerification.stderr}`);
 
 
170
  const manifestBuild = run(repo, "build-release-manifest.mjs");
171
  assert(manifestBuild.status === 0, `Release manifest fixture build failed.\n${manifestBuild.stdout}\n${manifestBuild.stderr}`);
172
  const cleanValidation = run(repo, "validate-data.mjs");
173
  assert(cleanValidation.status === 0, `Matching release manifest fixture did not validate.\n${cleanValidation.stdout}\n${cleanValidation.stderr}`);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
174
  await fs.appendFile(path.join(releaseDir, zipName), "tampered\n");
175
  const tamperedValidation = run(repo, "validate-data.mjs");
176
  const codes = errorCodes(tamperedValidation);
@@ -181,7 +214,8 @@ try {
181
  results.push({
182
  mutation: "release_artifact_after_manifest", status: "rejected",
183
  code: "RELEASE_ARTIFACT_HASH_MISMATCH", deterministic_package_rebuild: true,
184
- package_member_verification: "pass"
 
185
  });
186
  }
187
 
 
167
  assert(packageRebuild.status === 0 && firstZipHash === secondZipHash, "Repeated package build was not byte-deterministic");
168
  const packageVerification = run(repo, "verify-package.mjs");
169
  assert(packageVerification.status === 0, `Package member verification failed.\n${packageVerification.stdout}\n${packageVerification.stderr}`);
170
+ const initialCheckoutParity = run(repo, "verify-package.mjs", ["--checkout-parity"]);
171
+ assert(initialCheckoutParity.status === 0, `Fresh package must match its checkout.\n${initialCheckoutParity.stdout}\n${initialCheckoutParity.stderr}`);
172
  const manifestBuild = run(repo, "build-release-manifest.mjs");
173
  assert(manifestBuild.status === 0, `Release manifest fixture build failed.\n${manifestBuild.stdout}\n${manifestBuild.stderr}`);
174
  const cleanValidation = run(repo, "validate-data.mjs");
175
  assert(cleanValidation.status === 0, `Matching release manifest fixture did not validate.\n${cleanValidation.stdout}\n${cleanValidation.stderr}`);
176
+ const releaseManifestPath = path.join(releaseDir, "release-manifest.json");
177
+ const cleanReleaseManifest = await fs.readFile(releaseManifestPath, "utf8");
178
+ const duplicateManifest = JSON.parse(cleanReleaseManifest);
179
+ duplicateManifest.package_files[1] = { ...duplicateManifest.package_files[0] };
180
+ await fs.writeFile(releaseManifestPath, `${JSON.stringify(duplicateManifest, null, 2)}\n`);
181
+ const duplicateValidation = run(repo, "validate-data.mjs");
182
+ assert(
183
+ duplicateValidation.status !== 0 && errorCodes(duplicateValidation).includes("RELEASE_PACKAGE_FILE_DUPLICATE"),
184
+ "Duplicate release package-file entries were not rejected"
185
+ );
186
+ await fs.writeFile(releaseManifestPath, cleanReleaseManifest);
187
+ const manifestBeforeCheckoutChange = sha256(await fs.readFile(releaseManifestPath));
188
+ await fs.appendFile(path.join(repo, "README.md"), "\nPost-release mutable-main documentation fixture.\n");
189
+ const frozenVerification = run(repo, "verify-package.mjs");
190
+ assert(frozenVerification.status === 0, `Mutable checkout incorrectly invalidated the frozen package.\n${frozenVerification.stdout}\n${frozenVerification.stderr}`);
191
+ const changedCheckoutParity = run(repo, "verify-package.mjs", ["--checkout-parity"]);
192
+ assert(
193
+ changedCheckoutParity.status !== 0 && /differs from current checkout/i.test(`${changedCheckoutParity.stdout}\n${changedCheckoutParity.stderr}`),
194
+ "Explicit checkout-parity verification did not reject mutable-main drift"
195
+ );
196
+ const manifestRebuild = run(repo, "build-release-manifest.mjs");
197
+ const manifestAfterCheckoutChange = sha256(await fs.readFile(releaseManifestPath));
198
+ assert(
199
+ manifestRebuild.status === 0 && manifestBeforeCheckoutChange === manifestAfterCheckoutChange,
200
+ "Release manifest changed when only the mutable checkout diverged from the frozen ZIP"
201
+ );
202
+ const mutableMainValidation = run(repo, "validate-data.mjs");
203
+ assert(
204
+ mutableMainValidation.status === 0,
205
+ `Mutable-main documentation incorrectly invalidated the frozen release.\n${mutableMainValidation.stdout}\n${mutableMainValidation.stderr}`
206
+ );
207
  await fs.appendFile(path.join(releaseDir, zipName), "tampered\n");
208
  const tamperedValidation = run(repo, "validate-data.mjs");
209
  const codes = errorCodes(tamperedValidation);
 
214
  results.push({
215
  mutation: "release_artifact_after_manifest", status: "rejected",
216
  code: "RELEASE_ARTIFACT_HASH_MISMATCH", deterministic_package_rebuild: true,
217
+ package_member_verification: "pass", mutable_main_release_verification: "pass",
218
+ explicit_checkout_parity_after_drift: "rejected", duplicate_package_file_entry: "rejected"
219
  });
220
  }
221
 
scripts/validate-data.mjs CHANGED
@@ -76,6 +76,18 @@ function cleanOutput(value) {
76
  return String(value || "").trim().replace(/\s+/g, " ").slice(0, 500);
77
  }
78
 
 
 
 
 
 
 
 
 
 
 
 
 
79
  function readJson(filepath, label, code) {
80
  if (!fs.existsSync(filepath)) {
81
  fileError(`${code}_MISSING`, label, `${label} is missing`);
@@ -398,19 +410,34 @@ if (!sourceOnly) {
398
  if (actualSums !== canonicalSums) fileError("RELEASE_SUMS_MISMATCH", "release/SHA256SUMS", "SHA256SUMS differs from sorted manifest artifacts");
399
  }
400
  const expectedPackageFiles = ["README.md", "CITATION.cff", "package.json", "data/datapackage.json"];
 
 
 
401
  if (!Array.isArray(manifest.package_files)) fileError("RELEASE_PACKAGE_FILES_INVALID", "release/release-manifest.json", "package_files must be an array");
402
  else {
403
  const names = manifest.package_files.map((item) => item?.name);
 
 
 
 
 
 
404
  for (const name of expectedPackageFiles) if (!names.includes(name)) fileError("RELEASE_PACKAGE_FILE_MISSING", name, "Release manifest lacks package metadata hash");
 
 
 
405
  for (const item of manifest.package_files) {
406
  if (!expectedPackageFiles.includes(item?.name)) { fileError("RELEASE_PACKAGE_FILE_UNEXPECTED", String(item?.name), "Unexpected package-file entry"); continue; }
407
- const filepath = pathInside(packageRoot, item.name);
408
- const buffer = filepath && fs.existsSync(filepath) ? fs.readFileSync(filepath) : null;
409
- if (!buffer) fileError("RELEASE_PACKAGE_FILE_ABSENT", item.name, "Package file is missing");
410
- else {
411
- if (item.bytes !== buffer.byteLength) fileError("RELEASE_PACKAGE_FILE_BYTE_COUNT_MISMATCH", item.name, "Package file byte count differs");
412
- if (item.sha256 !== sha256(buffer)) fileError("RELEASE_PACKAGE_FILE_HASH_MISMATCH", item.name, "Package file differs from release manifest SHA-256");
 
413
  }
 
 
414
  }
415
  }
416
  if (manifest.reproducibility?.workbook !== "checksum_only_external_artifact_runtime_not_in_public_dependency_graph") {
 
76
  return String(value || "").trim().replace(/\s+/g, " ").slice(0, 500);
77
  }
78
 
79
+ function readZipMember(zipPath, member) {
80
+ const result = spawnSync("unzip", ["-p", zipPath, member], {
81
+ encoding: null,
82
+ maxBuffer: 128 * 1024 * 1024
83
+ });
84
+ if (result.error) throw result.error;
85
+ if (result.status !== 0) {
86
+ throw new Error(`unzip could not read ${member}: ${String(result.stderr || "").trim()}`);
87
+ }
88
+ return Buffer.from(result.stdout);
89
+ }
90
+
91
  function readJson(filepath, label, code) {
92
  if (!fs.existsSync(filepath)) {
93
  fileError(`${code}_MISSING`, label, `${label} is missing`);
 
410
  if (actualSums !== canonicalSums) fileError("RELEASE_SUMS_MISMATCH", "release/SHA256SUMS", "SHA256SUMS differs from sorted manifest artifacts");
411
  }
412
  const expectedPackageFiles = ["README.md", "CITATION.cff", "package.json", "data/datapackage.json"];
413
+ const frozenPackageRoot = `agency-transfer-election-cases-v${RELEASE_VERSION}`;
414
+ const frozenPackageName = `${frozenPackageRoot}.zip`;
415
+ const frozenPackagePath = path.join(releaseDir, frozenPackageName);
416
  if (!Array.isArray(manifest.package_files)) fileError("RELEASE_PACKAGE_FILES_INVALID", "release/release-manifest.json", "package_files must be an array");
417
  else {
418
  const names = manifest.package_files.map((item) => item?.name);
419
+ if (new Set(names).size !== names.length) {
420
+ fileError("RELEASE_PACKAGE_FILE_DUPLICATE", "release/release-manifest.json", "Duplicate package-file entry");
421
+ }
422
+ if (JSON.stringify(names) !== JSON.stringify(expectedPackageFiles)) {
423
+ fileError("RELEASE_PACKAGE_FILE_ORDER_INVALID", "release/release-manifest.json", "Package-file entries must exactly match the declared order");
424
+ }
425
  for (const name of expectedPackageFiles) if (!names.includes(name)) fileError("RELEASE_PACKAGE_FILE_MISSING", name, "Release manifest lacks package metadata hash");
426
+ if (!fs.existsSync(frozenPackagePath)) {
427
+ fileError("RELEASE_PACKAGE_ARCHIVE_MISSING", frozenPackageName, "Frozen package ZIP is missing");
428
+ }
429
  for (const item of manifest.package_files) {
430
  if (!expectedPackageFiles.includes(item?.name)) { fileError("RELEASE_PACKAGE_FILE_UNEXPECTED", String(item?.name), "Unexpected package-file entry"); continue; }
431
+ if (!fs.existsSync(frozenPackagePath)) continue;
432
+ let buffer;
433
+ try {
434
+ buffer = readZipMember(frozenPackagePath, `${frozenPackageRoot}/${item.name}`);
435
+ } catch (caught) {
436
+ fileError("RELEASE_PACKAGE_FILE_ABSENT", item.name, `Frozen package member is missing or unreadable: ${cleanOutput(caught.message)}`);
437
+ continue;
438
  }
439
+ if (item.bytes !== buffer.byteLength) fileError("RELEASE_PACKAGE_FILE_BYTE_COUNT_MISMATCH", item.name, "Frozen package member byte count differs");
440
+ if (item.sha256 !== sha256(buffer)) fileError("RELEASE_PACKAGE_FILE_HASH_MISMATCH", item.name, "Frozen package member differs from release manifest SHA-256");
441
  }
442
  }
443
  if (manifest.reproducibility?.workbook !== "checksum_only_external_artifact_runtime_not_in_public_dependency_graph") {
scripts/verify-package.mjs CHANGED
@@ -1,6 +1,6 @@
1
  #!/usr/bin/env node
2
 
3
- /** Verify ZIP safety, member hashes, checksums and parity with the checkout. */
4
 
5
  import fs from "node:fs";
6
  import path from "node:path";
@@ -13,7 +13,13 @@ const here = path.dirname(fileURLToPath(import.meta.url));
13
  const root = path.resolve(here, "..");
14
  const packageRoot = `agency-transfer-election-cases-v${VERSION}`;
15
  const workbookName = `agency-transfer-election-evidence-index-v${VERSION}.xlsx`;
16
- const zipPath = path.resolve(process.argv[2] || path.join(root, "release", `${packageRoot}.zip`));
 
 
 
 
 
 
17
 
18
  function sha256(buffer) {
19
  return createHash("sha256").update(buffer).digest("hex");
@@ -67,11 +73,13 @@ for (const entry of manifest.files) {
67
  const member = `${packageRoot}/${entry.path}`;
68
  const buffer = Buffer.from(unzip(["-p", zipPath, member], true));
69
  if (buffer.byteLength !== entry.bytes || sha256(buffer) !== entry.sha256) throw new Error(`ZIP member differs from manifest: ${entry.path}`);
70
- const checkoutPath = entry.path.startsWith("workbook/")
71
- ? path.join(root, "release", workbookName)
72
- : path.join(root, entry.path);
73
- if (!fs.existsSync(checkoutPath) || sha256(fs.readFileSync(checkoutPath)) !== entry.sha256) {
74
- throw new Error(`ZIP member differs from current checkout/release artifact: ${entry.path}`);
 
 
75
  }
76
  }
77
  const expectedSums = `${manifest.files.map((entry) => `${entry.sha256} ${entry.path}`).join("\n")}\n`;
@@ -86,5 +94,6 @@ process.stdout.write(`${JSON.stringify({
86
  bytes: zipBuffer.byteLength,
87
  sha256: sha256(zipBuffer),
88
  members: fileMembers.length,
89
- payload_files: manifest.files.length
 
90
  })}\n`);
 
1
  #!/usr/bin/env node
2
 
3
+ /** Verify frozen ZIP integrity, with optional parity against the mutable checkout. */
4
 
5
  import fs from "node:fs";
6
  import path from "node:path";
 
13
  const root = path.resolve(here, "..");
14
  const packageRoot = `agency-transfer-election-cases-v${VERSION}`;
15
  const workbookName = `agency-transfer-election-evidence-index-v${VERSION}.xlsx`;
16
+ const cliArgs = process.argv.slice(2);
17
+ const checkoutParity = cliArgs.includes("--checkout-parity");
18
+ const unknownOptions = cliArgs.filter((arg) => arg.startsWith("--") && arg !== "--checkout-parity");
19
+ const positional = cliArgs.filter((arg) => !arg.startsWith("--"));
20
+ if (unknownOptions.length) throw new Error(`Unknown option: ${unknownOptions.join(", ")}`);
21
+ if (positional.length > 1) throw new Error("Expected at most one ZIP path");
22
+ const zipPath = path.resolve(positional[0] || path.join(root, "release", `${packageRoot}.zip`));
23
 
24
  function sha256(buffer) {
25
  return createHash("sha256").update(buffer).digest("hex");
 
73
  const member = `${packageRoot}/${entry.path}`;
74
  const buffer = Buffer.from(unzip(["-p", zipPath, member], true));
75
  if (buffer.byteLength !== entry.bytes || sha256(buffer) !== entry.sha256) throw new Error(`ZIP member differs from manifest: ${entry.path}`);
76
+ if (checkoutParity) {
77
+ const checkoutPath = entry.path.startsWith("workbook/")
78
+ ? path.join(root, "release", workbookName)
79
+ : path.join(root, entry.path);
80
+ if (!fs.existsSync(checkoutPath) || sha256(fs.readFileSync(checkoutPath)) !== entry.sha256) {
81
+ throw new Error(`ZIP member differs from current checkout/release artifact: ${entry.path}`);
82
+ }
83
  }
84
  }
85
  const expectedSums = `${manifest.files.map((entry) => `${entry.sha256} ${entry.path}`).join("\n")}\n`;
 
94
  bytes: zipBuffer.byteLength,
95
  sha256: sha256(zipBuffer),
96
  members: fileMembers.length,
97
+ payload_files: manifest.files.length,
98
+ checkout_parity: checkoutParity ? "pass" : "not_checked"
99
  })}\n`);