Skip to main content

Investigate: IMDikator's API, for the two IMDi indicators bosettingstall doesn't cover

Whether IMDikator's backend can be reverse-engineered to serve imdi-innvandringsgrunn-kjonn (immigration reasons by gender) and imdi-landbakgrunn (country of origin) — the two IMDi indicators a 2026-09-29 outside-consumer report flagged as missing that imdi-bosetting does not cover.

IMPLEMENTATION RULES: Before implementing this plan, read and follow:

Status: Backlog​

Goal: Determine whether IMDikator's API can be reverse-engineered well enough to ingest imdi-innvandringsgrunn-kjonn and imdi-landbakgrunn — or whether a different source needs to be found for them.

Last Updated: 2026-10-01

Origin: Split out of PLAN-009-imdi-bosetting.md, which deliberately did not fold these two indicators in despite the parent investigation's own INVESTIGATE-new-norwegian-public-sources.md [Q42] recommending exactly that — see that PLAN's Implementation Notes for the full reasoning. Also tracked as open items [Q42] (scope) and [Q43] (overlap with fhi-innvandrere) in the parent investigation, which stays in backlog/ until this ships too. Independently corroborated by INVESTIGATE-samfunnspuls-replacement-gaps.md [Q4] — a different outside consumer (an agent rebuilding samfunnspuls.rodekors.no) flagged the same two IMDi indicators as missing, counting "IMDi (2)" among 15 ungraphed Samfunnspuls panels.

Questions to Answer​

  1. [Q1] Does IMDikator's backend (app-simapi-prod.azurewebsites.net) actually carry imdi-innvandringsgrunn-kjonn and imdi-landbakgrunn data at all? Not confirmed — only the tool's visible scope (population composition, immigration reasons, world regions, residence duration) was read from its UI, which reads as closer to these two indicators than to bosettingstall, but no API response has actually been inspected.
  2. [Q2] What are the real endpoint(s), request shape and response shape? The bundle that calls the API is loaded dynamically (confirmed: static analysis of the page's JS found only a script injector, no endpoint strings) — this needs a headless browser to observe actual network traffic, which a prior pass explicitly deferred as out of proportion for PLAN-009.
  3. [Q3] Does it need authentication, a key, or is it open like imdi-bosetting's static HTML?
  4. [Q4] Overlap with fhi-innvandrere — the parent investigation's own [Q43] notes FHI typically lags IMDi by one cycle; worth checking whether the methodology gap between the two is meaningful enough to justify a second, IMDi-specific ingest of overlapping ground.
  5. [Q5] If IMDikator doesn't pan out, is there any other public distribution for these two indicators — data.norge.no re-check, a direct IMDi statistics page in the same static-HTML shape as bosettingstall, or similar?

Current State​

What's already confirmed, from PLAN-009-imdi-bosetting.md's Phase 1.2 research (2026-10-01):

  • IMDikator is real and live, at arkiv.imdi.no/statistikk/.
  • Backed by a dedicated API host, app-simapi-prod.azurewebsites.net (found via the page's data-api-host attribute).
  • Currently updated — data as recent as January 2025 was seen on the page.
  • No Swagger/OpenAPI docs found at the usual paths (checked, 404).
  • The data.norge.no catalogue record that points toward it (org 987879696, "Statistikk om innvandring og integrering") is 6 years stale (modified: 2020-05-26) — not trustworthy on its own, consistent with catalogue staleness already found wrong twice this session (NAV, Bufdir).
  • imdi-bosetting itself is live, shipped, and verified (urb-agents#1799) — proof IMDi as a publisher is reachable and that Terje's licence authorization ("IMDI is ok. we can use it.", 2026-10-01) exists for this publisher. ⚠️ That authorization was given for bosettingstall specifically — treat it as likely to extend to other IMDi indicators, not as already re-confirmed for them. Flag for Terje if this investigation reaches an implementation decision.

What's not yet done: nobody has pointed a headless browser (Playwright/Puppeteer) at arkiv.imdi.no/statistikk/, driven its UI, and captured the actual XHR/fetch calls it makes. That is the next concrete step, not more static analysis.

Options​

Option A: Reverse-engineer IMDikator with a headless browser​

Pros:

  • If it works, likely the richest, most current source for both indicators.
  • One mechanism serves both missing indicators at once.

Cons:

  • No documented API — purely observational reverse-engineering, which can break silently on any IMDikator frontend change (no contract, no versioning promise).
  • Unknown effort until someone actually drives it — could be a 30-minute Playwright session or a multi-day slog depending on how the app is built (SPA state management, auth tokens, pagination).
  • [Q1] (does it even carry this data) is unconfirmed, so the investigation could end in "no."

Option B: Find a different public distribution​

Pros:

  • Could turn out to be as simple as imdi-bosetting itself (static HTML, no reverse-engineering).

Cons:

  • The obvious lead (data.norge.no) already pointed at IMDikator and nowhere else — a fresh search needs a reason to expect a different result this time, not just another query.

Option C: Accept the gap, rely on fhi-innvandrere for the overlapping ground​

Pros:

  • Zero new ingest work.
  • fhi-innvandrere already exists and may cover enough of the same signal.

Cons:

  • Doesn't answer the consumer report's actual request — a Samfunnspuls-shaped UI wants IMDi's own framing (immigration reason, country of origin), not FHI's.
  • Doesn't resolve [Q43]'s open "is the methodology gap meaningful" question, it just avoids it.

Recommendation​

None yet — this is exactly the gap this file exists to close. [Q2] (observe real network traffic with a headless browser) is the one concrete next action; everything else is downstream of what that returns.

Next Steps​

  • Drive arkiv.imdi.no/statistikk/ with a headless browser, capture real XHR/fetch calls against app-simapi-prod.azurewebsites.net.
  • If a usable API surface is found: create PLAN-011-imdi-innvandringsgrunn-landbakgrunn.md (or split into two PLANs if the two indicators turn out to need different handling).
  • If not: re-check data.norge.no and IMDi's own site for any other distribution ([Q5]), or close this out documenting why neither indicator can be self-served — same shape as redcross-branches' bespoke-extract conclusion in INVESTIGATE-samfunnspuls-replacement-gaps.md [Q8].