Norway (NO) — internal licensing notes

Source research, 2026-08-29

Clean GO, and a genuine official upgrade over GeoNames is available.

The better source — Bring / Posten Norge postnummerregister

https://www.bring.no/postnummerregister-ansi.txt

Publisher: Bring (Posten Norge AS), Norway's national postal operator. Licence: Norwegian Licence for Open Government Data (NLOD) 2.0. Verified directly against the licence text (SPDX / data.norge.no) on 2026-08-29 — commercial use and resale ARE permitted (an earlier search summary claimed the opposite; checked the actual licence text and that claim is wrong). No share-alike. Attribution required, default form: "Contains data under the Norwegian licence for Open Government data (NLOD) distributed by [name of licensor]." NLOD is explicitly listed as compatible with CC BY and OGL — same shape of licence as Oman's NCSI OGL already used elsewhere in this catalogue.

Postal-area boundary polygons (postkretser) are separately published by Kartverket (the national mapping authority) via data.norge.no, also under Norwegian open-government terms — not needed for this schema (no geometry column) but useful if boundary data is ever wanted.

Not yet fetched or parsed this pass — this is a research pass only. The file is a flat ANSI text table (postal code, place, municipality, county), should be straightforward to parse.

The fallback — GeoNames NO postal export

https://download.geonames.org/export/zip/NO.zip   # 5,132 rows

Licence CC BY 4.0, verified against export/zip/readme.txt on 2026-08-29 — no per-country caveat for Norway. Admin data is complete: admin_name1 (fylke/county) filled on all 5,132 rows, 15 distinct (matches Norway's current 15-fylke structure); admin_name2 (kommune/municipality) filled on all 5,132 rows, 357 distinct. No admin_name3. Zero primary-key collisions on (postal_code, place_name) — all 5,132 postal codes are already distinct. Clean data, no defects found.

Since the official Bring source is directly downloadable, openly licensed, and more authoritative, it should be preferred as the base if a rebuild happens — GeoNames remains a clean fallback either way.

licence_status (proposed — needs human sign-off before shipping)

{
  "channel": "postal-codes-r2",
  "source_url": "https://www.bring.no/postnummerregister-ansi.txt",
  "licence_id": "custom:nlod-2.0",
  "collected_at": "2026-08-29",
  "decision": "compatible",
  "decided_by": "claude-session-2026-08-29",
  "decided_at": "2026-08-29"
}

Proposed tier

B ($49). Single official source normalised in place. No second source consolidation needed — GeoNames' admin tiers already match what Bring would provide, so this doesn't reach tier A's multi-source-consolidation bar even if both are checked against each other for a quality pass.

Verdict

Build it — this is one of the cleaner cases in the batch. No licensing obstacle, complete 2-tier admin coverage, no data defects found in either candidate source. Preference: parse Bring's official file as the base (stronger provenance than GeoNames for a paid product), cross-check row count and admin values against GeoNames.

Build — 2026-08-30 (local only, not published)

Built via datasets/no/scripts/build_postal_codes_product.py, parsing Bring's official postnummerregister-ansi.txt directly (latin-1, tab-delimited, 5,122 rows, no duplicates on postal_code).

Real finding: Bring's file carries kommune (municipality) but no fylke (county) column at all. Derived fylke from the first two digits of kommune_kode — this IS Norway's own official fylkesnummer by design, not an assumption; verified by cross-checking all 17 distinct 2-digit prefixes found in the live data against no.wikipedia.org/wiki/Fylkesnummer's current list, exact 1:1 match (10x, 11x, 15x, 18x, 21x Svalbard, 22x Jan Mayen, 31x, 32x, 33x, 34x, 39x, 40x, 42x, 46x, 50x, 55x, 56x).

Second real finding, caught before hardcoding a wrong ISO code: Norway's fylke structure changed 2024-01-01 (Viken split into Ostfold/Akershus/Buskerud; Vestfold og Telemark split into Vestfold/Telemark; Troms og Finnmark split into Troms/Finnmark), but ISO 3166-2:NO's own register has not been updated to reflect this — confirmed via WebSearch 2026-08-30, quoting a source stating the ISO standard "matches the county codes assigned by the Norwegian government from 2020 to 2024." A first WebFetch of the English Wikipedia ISO page returned exactly that stale 2020-2024 structure (11 counties including merged "Viken", "Vestfold og Telemark", "Troms og Finnmark") and would have produced wrong codes if trusted uncritically. Resolution, matching this project's existing Angola precedent for an ISO register lagging a real administrative reform: use Norway's own official fylkesnummer for all 17 values (NO-03 … NO-56), and disclose in ATTRIBUTION.txt/ README.md that 7 of them (Ostfold, Akershus, Buskerud, Vestfold, Telemark, Troms, Finnmark — 1,527 of 5,122 rows) do not yet have an individual ISO 3166-2:NO entry, rather than silently reusing their stale merged predecessor code or leaving the column empty.

Results:

No coordinates in this release — Bring's register doesn't carry them (same as GeoNames' NO export lacking this feature; neither source has it).

Frictionless validation: valid: True.

Status: built and validated locally only. Nothing published to R2, nothing committed to git. datasets/no/README.md and datapackage.yml updated in place (tier B, premium_source_key: no/no.zip, real sources/licenses, replacing the "Multiple sources" placeholder).