Most US zoning codes exist only as unstructured PDF or paper documents that cannot be programmatically queried — so nobody can answer 'what can I build here?' without hiring a land-use attorney
housinghousing0 views
A property owner, developer, or housing advocate who wants to answer the simple question 'what can I build on this parcel?' must navigate a zoning code that, in most US jurisdictions, exists as a scanned PDF, a decades-old Word document uploaded to a municipal website, or — in some towns — only as a physical binder in a clerk's office. There is no API, no structured data, no machine-readable format. The National Zoning Atlas project, run by researchers at Cornell, has been attempting to digitize zoning rules across the country and has documented that many jurisdictions still rely on hand-drawn maps that do not align with GIS parcel data. Researchers must 'decipher tangled geospatial files, digitize decades-old, hand-drawn maps,' and manually redraw misdrawn layers including 'unintentional overlaps and leftover slivers.'
The practical consequence is that determining what is buildable on a specific lot requires a professional — typically a land-use attorney billing $300-$600/hour or a zoning consultant — to manually cross-reference the zoning map, the text of the ordinance, any overlay districts, the comprehensive plan, and recent amendments that may not yet be codified. This process takes days to weeks and costs $2,000-$10,000 per parcel analysis. For a small developer evaluating 20 potential sites, that is $40,000-$200,000 in feasibility costs before any design work begins. This cost is a rounding error for large institutional developers but a dealbreaker for the small-scale builders who would otherwise produce missing-middle housing like duplexes and fourplexes.
Zoning codes remain unstructured because municipalities lack the budget, technical staff, and institutional incentive to digitize them. The code was written by lawyers for lawyers, amended piecemeal over decades, and maintained by planning departments that are chronically understaffed. There is no federal or state mandate requiring machine-readable zoning data. Startups like Gridics and Zoneomics are attempting to build this layer commercially, and the National Zoning Atlas is doing it academically, but as of 2026 the Atlas is still missing up-to-date electronic zoning data for more than a dozen cities and towns even in states where the project is active. The result is that the single most important regulatory constraint on housing — what the zoning code allows — remains locked inside documents that only specialists can interpret.
Evidence
National Zoning Atlas: methodology describing digitization challenges — https://www.zoningatlas.org/how | National Zoning Atlas: about — https://www.zoningatlas.org/about | enCodePlus (Feb 2026): 'AI-Ready Zoning and Planning Documents for Cities' — https://www.encodeplus.com/2026/02/ai-ready-zoning-planning-documents/ | Zoneomics: 'Top 5 Challenges with Zoning' — https://www.zoneomics.com/blog/top-5-challenges-with-zoning