Facebook has created a map of Earth that can pinpoint any man-made structures to a resolution of 5 meters, to help company's Internet provision efforts
- Facebook used satellite-based data and government census information to map the earth's entire human population. Thanks: @mattrosoff
Context & Ripple Effects
This 2017 map is the seed of a geospatial program Facebook spent the next two years expanding: it fused satellite imagery with government census data into a single 5-meter-resolution picture of where humans actually live, built explicitly to decide where its internet-provision efforts should go. By 2019 the same pipeline had matured into AI-generated population-density maps covering most of Africa and Map With AI, tooling that maps previously unmapped places in the developing world.
The reason it matters beyond connectivity is what happened next: Facebook began releasing these datasets externally, including [[a:941879|population-density, movement, and network-coverage maps handed to nonprofits and universities tracking infectious-disease spread]]. A dataset built as internal infrastructure became a public-facing asset — and a private company now holds one of the most granular pictures of global human settlement.
First-order effects
- Facebook gains a targeting layer for its own connectivity build-out: infrastructure can be routed to unmapped or underserved settlements identified by the satellite-plus-census fusion rather than by ground surveys.
- Governments supplying census data become upstream partners, effectively contributing public statistics to a corporate mapping effort whose primary beneficiary is Facebook's own expansion.
Second-order effects
- Once released, the maps become free planning inputs for nonprofits, universities, and presumably network operators — lowering the cost of infrastructure and epidemiology work that previously depended on sparse official censuses.
- Commercial satellite-imagery providers gain a large institutional customer whose demand for global coverage validates imagery as a mass-market input rather than a niche government procurement.
Third-order effects
- If the pattern holds, platform companies become de facto geospatial infrastructure providers: census-grade knowledge of human settlement accumulates inside firms whose incentives are growth, not cartography, raising questions about who governs the world's most detailed population maps.
- The reuse of government census data inside a commercial product sits squarely on the public-data permission boundary — the same tension that will recur as more public datasets feed private AI and mapping systems.
The trend: Platform companies are converting satellite imagery and public census data into proprietary geospatial infrastructure — first to target their own expansion, then increasingly released as influence-building public goods.