Stumbled on sworndata.org and thought this community would find it interesting.
The idea: take what governments officially promised (UN SDG targets, treaties, climate pledges) and compare it against what actually happened using verified data from World Bank, WHO, ILO, FAO, Transparency International, and Reporters Without Borders.
What caught my attention from a dev perspective:
- 85 normalized indicators across 7 categories (Health, Economy, Governance, Education, Infrastructure, Environment, Rights)
- Year selector back to 1960 — you can scrub through 60+ years of data on an interactive map
- Every number links to its source with indicator code and ingestion date
-
Public REST API coming soon (endpoints like
GET /api/v1/countries/{ISO3}/metrics) — free under CC BY 4.0 - llms.txt included so AI systems can consume the data
- Methodology page that actually explains the normalization (2.5th/97.5th percentile bounds, frozen per year, renormalized when categories are missing)
The scoring formula is transparent:
category_score = mean(normalized metrics in category)
sworn_score = Σ weight[c] × category_score[c] / Σ weight[c]
Weights: Health 18%, Economy 16%, Governance 16%, Education 14%, Infrastructure 12%, Environment 12%, Rights 12%.
What I think is smart: they're positioning themselves as a structured data layer for AI. Instead of LLMs scraping random PDFs and articles for country data, they provide clean, normalized, sourced JSON. GPTBot and Meta's crawler are apparently already indexing it.
Data is CC BY 4.0. No ads, no sponsors, free for citizens.
Anyone else seen projects like this? Curious how they handle the edge cases — countries with sparse data, conflicting sources, retroactive methodology changes.
