Les données ouvertes, l'avis libre. · Open data, your own opinion.
An independent, bilingual (🇫🇷/🇬🇧) open-data observatory that maps the real neighborhood footprint of low-cost halal fried-chicken chains in France — and lets every citizen reach their own conclusion.
In April 2026, a Master Poulet chicken shop opened in the heart of Saint-Ouen, almost under the town hall's windows. Three days later the mayor closed it and set concrete blocks at the door; a court ruled the blocking an illegal violation of the freedom of enterprise; the city replied with giant flowerpots. The "guerre du poulet" went national — gentrification vs. popular demand, "junk food" vs. accessibility, class, halal, real estate, safety.
The debate overflowed with confident figures — on prices, "food deserts", crime, even money laundering — yet almost none were sourced, and several openly stigmatized working-class neighborhoods.
This project turns that noise into something useful: a transparent, reproducible, source-cited tool built only on open public data, where every claim is labelled documented / emerging / contested — so you can make up your own mind instead of picking a side handed to you.
🧭 The rule of the house: never present a correlation as a cause, nor a rumor as a fact.
|
Toggle layers across Greater Paris — property prices (€/m²), fast-food density, target chicken chains, and municipal crime (shown only behind an ecological-fallacy warning). Click any commune or outlet. |
A two-sided case study: the timeline, the price trend, a focused map, a "draw your guess then compare" moment, and a pour / objections debate panel — each argument tagged with its evidence level. |
Type any French address → the life & death of the fast-food shops around you: how many, which brands, which closed or went bankrupt, and how long they last (a survival curve vs. your commune & France), with a downloadable share card. National SIRENE survival data + live BODACC bankruptcy lookups. Honest by design: closures are normal small-business churn, independents are never named, and revenue is shown only where public.
Plus a two-commune comparison, a transparent Methods & limits page, a "money-laundering: proven vs. rumour" explainer, a sources registry, and a full English mirror.
Scope: Greater Paris — Paris (75) · Hauts-de-Seine (92) · Seine-Saint-Denis (93) · Val-de-Marne (94).
| Layer | Source | Status |
|---|---|---|
| 🏠 Property prices (median €/m², 2021–2024) | DVF géolocalisé (Etalab/DGFiP) | ✅ real — 337k sales |
| 🍔 Fast-food density | OpenStreetMap amenity=fast_food |
✅ real — 4 999 POIs (non-exhaustive) |
| 🍗 Target chicken/crousti outlets | OSM brand match + reference seed | ✅ real — 64 outlets |
| 🚨 Recorded crime (commune, 2016–2025) | SSMSI (Ministère de l'Intérieur) | ✅ real (municipal only, gated) |
| 🏘️ Fast-food survival, closures & lifespan (national) | SIRENE (INSEE) | ✅ real — 393k shops, 233k already closed |
| ⚖️ Bankruptcies (live, on-demand) | BODACC (DILA) | ✅ real — queried by SIREN in the browser |
| 🗺️ Boundaries & basemap | IGN Géoplateforme | ✅ real |
| 💶 Income / poverty (Filosofi) · SIRENE 56.10C | INSEE | ⏳ national version |
✔️ A sanity check is baked into the pipeline: Saint-Ouen's median lands at 6 399 €/m² (2024), matching the press dossier's ">6 300 €/m²".
This is what separates an observatory from a hot take:
- Evidence labels on every number —
documenté/émergent/contesté, with a source link on every figure. - Reverse causation, up front — cheap food locates where rents are already low; the cause usually precedes the shop.
- Crime stays commune-level, behind a gate — context, never a cause attributable to a shop or a population.
- Statistical-secrecy values stay
null, never0— a suppressed cell reads "non diffusé". - No money-laundering map, no identity framing — the cash-business risk is real in general but unproven for these brands, and we say exactly that.
- RGPD-safe — only non-nominative parcel/establishment data, no cookies, no trackers.
Public open data ──▶ Node + DuckDB pipeline ──▶ src/data/*.json ──▶ Observable Framework
(DVF, OSM, SSMSI, (download, clean, (flat GeoJSON/JSON (static site: MapLibre
IGN, geo.api) spatial-join, publish) the site consumes) maps + Plot charts)
No server, no database at runtime, no tracker — a fully static, cheap-to-host site.
Pipeline stages (pipeline/): 30_geo (boundaries) · 10_dvf (prices) · 20_chains (fast-food + brands) · 32_crime (SSMSI) · 40_summary (KPIs). Each is reproducible from public URLs.
npm install
npm run pipeline # rebuild src/data/ from public sources (Node + DuckDB)
npm run dev # Observable Framework preview → http://localhost:3000
npm run build # static build → dist/
node test/smoke.mjs # headless smoke test (uses system Chrome) → "11/11 pages clean"Deploy dist/ to any static host (Cloudflare Pages / Netlify / GitHub Pages). See CLAUDE.md for architecture and data-source gotchas, src/methodes.md for methodology, and DESIGN_BRIEF.md for the redesign roadmap.
- v1 — national · INSEE Filosofi income/poverty (IRIS), SIRENE 56.10C legal universe, France-wide scale-up (PMTiles).
- v2 — inferential · event-study (difference-in-differences) around a chain opening, with confidence intervals and reverse-causation warnings — never on the home page.
- Design & growth · scrollytelling, mobile-first map, social share-cards, "your neighborhood" deep links (see
DESIGN_BRIEF.md).
Code: MIT. Data: Licence Ouverte / Etalab 2.0 (OpenStreetMap: ODbL).
Contains data from Etalab/DGFiP (DVF), INSEE (SIRENE, census), SSMSI — Ministère de l'Intérieur (crime), IGN (boundaries, basemap), and © OpenStreetMap contributors.
⚖️ Independent observatory — affiliated with no chain and with no politician. The pro-chain support dossier is cited only as one party's position, never as the site's voice. Figures are dated by their source vintage; corrections welcome.