GHSL mislabels huge swaths of rural Bihar as urban; researchers release better South Asia city boundaries
paulnovosad · x · 2026-09-18
Paul Novosad's team warns researchers against GHSL's global city database: it classifies huge parts of India's Bihar as urban, and its 'Hajipur megacity' is nothing of the sort.
They released a set of urban polygons that work far better in South Asia (Urban Data Commons / DDL City Boundaries), aiming to become a Penn World Tables-style global standard for city-level economic data. Context: LMIC statistical agencies have rural biases and cities are under-studied; the project aggregates satellite data, policies extracted from municipal documents with LLMs, and more.
Related event: Scholars Flag GHSL Urban Data Errors, Release Better South Asia Boundaries(3 posts)→
More from Research
- Researchers: RL training has made models' theory of mind 'terrible' — voooooogel · 2026-09-18
- The 272-page Multimodal Deep Learning book is free to read online with an AI tutor — burkov · 2026-09-18
- Therna launches Chronos RNA foundation model trained across ~50 human cell lines — LightningAI · 2026-09-18
- Arvind Narayanan's ICML Keynote: RSI Is Not the Singularity, and No Lab Milestone Kills All Jobs — random_walker · 2026-09-18
- Does RL build new representations or just make models flail? LLM researchers debate — doomslide · 2026-09-18
- Statistician argues alignment is impossible: LLM outputs are distributional, not fixed traits — gerardsans · 2026-09-18