0012: Open-data license for TaxpayerScope’s published datasets

Context

DEBT-02 (.planning/REQUIREMENTS.md) carries a v1.0 tech-debt item forward into v1.1: the Dataset JSON-LD license field on both the county overview page and /data/ was, at v1.0 close, a descriptive CreativeWork pointing at /methodology/#independence — a plain “free public access” statement, not a named open-data license such as CC0 or CC-BY. STATE.md’s Deferred Items table records this exactly: “Content decision | Dataset JSON-LD license is a descriptive CreativeWork, not a named open-data license (e.g. CC0/CC-BY) — pending maintainer’s formal licence choice | open — scheduled v1.1 Phase 11 (DEBT-02)”. A named license is a legal choice for the maintainer, never agent-chosen (the same T-03-26 posture every other permanent, public-facing decision in this directory takes) — 11-CONTEXT.md D-15 records that the maintainer has now made that choice. This record puts the choice itself, and nothing else, in writing: it changes no published figure, so it carries no methodology_version bump (active_methodology_version() requires every active record to agree, and this record declares methodology_version: 4, the version already active via ADR-0010/ADR-0011).

Decision

TaxpayerScope’s compiled datasets are dedicated to the public domain under the CC0 1.0 Universal public-domain dedication. “Compiled datasets” means every file TaxpayerScope itself derives, normalizes, or re-exports: the published /data/v1/ JSON contract, the downloads/v1/ CSV/JSON/Parquet files, and the two Dataset JSON-LD nodes (the county overview page’s countyDatasetNode and /data/‘s dataPageDatasetNode) in apps/web/src/lib/seo.ts. It does not relicense the government source files TaxpayerScope mirrors at /sources/v1/ — those keep their own publishers’ terms, which docs/sources/edr-portal-terms.md already records as publishing no data license of their own.

Concretely:

1. Both Dataset JSON-LD nodes’ license field is the CC0 URL string (https://creativecommons.org/publicdomain/zero/1.0/), replacing the prior INDEPENDENCE_LICENSE CreativeWork object. apps/web/src/lib/seo.ts exports OPEN_DATA_LICENSE_URL and OPEN_DATA_LICENSE_NAME as the single source both nodes and the /data/ page read from, so the ADR, the structured data and the visible page can never drift apart independently (T-11-03).
2. /data/ states the licence in visible text, linked rel="license", naming the compiled- datasets-only scope and the government source files’ own separate terms (plan 11-02 Task 2).
3. A later plan (11-14) adds an optional license field to DownloadsIndex and DownloadDataset, plus Parquet key-value metadata, carrying the same CC0 URL. CSV columns are unchanged — ADR-0011’s column lists are a test-compared contract — because the licence is covered by the downloads index.json and the /data/ page, not a new per-row CSV column.

Rationale

Considered options:

• CC0 1.0 Universal (chosen). A public-domain dedication: no attribution requirement, no share-alike requirement, no conditions at all beyond the dedication itself. Matches TaxpayerScope’s own positioning — a transparency platform republishing public government records should not attach its own new restriction on top of data that was already a public record before TaxpayerScope touched it. The rejected alternative below would have added exactly such a restriction.
• CC-BY 4.0 (rejected). Requires downstream attribution to TaxpayerScope specifically. Cost of choosing it: a derivative work (a journalist’s chart, a researcher’s republished table) would need to credit TaxpayerScope by name to stay compliant — a condition the underlying government source files themselves never imposed, and one PRD §44 (independence; data presentation never serving the platform’s own interest) weighs against for a dataset that is itself a re-export of records the public already owned.
• Status quo — descriptive CreativeWork, no named license (rejected). Cost of keeping it: DEBT-02 stays open indefinitely, and neither a search engine nor a human reader can tell from the JSON-LD or the page what rights, if any, attach to reuse — the exact gap this record closes.

Source Evidence

• .planning/phases/11-data-lineage-and-publish-path-hardening/11-CONTEXT.md, D-15 and “Specific Ideas” (dated 2026-10-04): the maintainer’s recorded choice, quoted verbatim in Decision sign-off below.
• docs/sources/edr-portal-terms.md: confirms edr.state.fl.us (the source host for every Florida EDR file this project mirrors) publishes no robots.txt restriction, no stated rate limit, no attribution requirement and no redistribution restriction on the underlying files — and, separately, states no data license of its own, which is why this record’s scope line (compiled_datasets_only) is necessary: TaxpayerScope cannot and does not grant a license over files it did not author.
• .planning/STATE.md, Deferred Items table, row: “Content decision | Dataset JSON-LD license is a descriptive CreativeWork, not a named open-data license (e.g. CC0/CC-BY) — pending maintainer’s formal licence choice | open — scheduled v1.1 Phase 11 (DEBT-02) | 2026-09-28 | v1 (Phase 07)” — the exact blocker this record resolves.

Machine-readable lines

The three lines below are parsed directly by pipeline/tests/publish/test_open_data_license.py and pin the values plan 11-02 and plan 11-14 compare their own code against.

open_data_license_name: CC0 1.0 Universal open_data_license_url: https://creativecommons.org/publicdomain/zero/1.0/ open_data_license_scope: compiled_datasets_only

Decision sign-off

Maintainer’s choice, recorded verbatim in .planning/phases/11-data-lineage-and-publish-path-hardening/11-CONTEXT.md (D-15, “Specific Ideas”, 2026-10-04):

“D-15 (licence — maintainer’s choice): CC0 1.0 Universal (public-domain dedication).”

“The licence is CC0 1.0 (chosen by the maintainer, 2026-10-04). The legal choice is made, which removes the STATE.md DEBT-02 blocker.”

Recorded: 2026-10-04T23:46:41Z (the UTC instant this ADR was written, 11-02-PLAN.md Task 1). This is the sole authority for the open-data licence this record and plan 11-14 implement — T-03-26’s mitigation: a permanent, public-facing legal choice is never made by the agent alone.

Methodology Version

No bump was taken by this record. The licence choice changes no published figure — every /data/v1/ field, every downloads column and every computed number is unchanged by attaching a named license to the compiled output. This record was written at methodology_version: 4 (the version already active via ADR-0010/ADR-0011), and active_methodology_version() requires every active record to agree — this record agrees by declaring the same version, with no coordinated revision needed on 0001–0011.

Revision (2026-10-04)

Initial record: 11-02-PLAN.md Task 1 wrote this record as status: active directly (not proposed, since the maintainer’s choice was already made and recorded before this plan executed — D-15), with the three machine-readable lines, the Decision sign-off quoting 11-CONTEXT.md D-15 and Specific Ideas verbatim, and methodology_version: 4 unchanged. This revision recorded by 11-02-PLAN.md Task 1.