Open Data Manifesto
Why Feed America publishes its 327,000-location food-assistance directory as open data under Creative Commons BY 4.0 — and what becomes possible when civic-tech infrastructure refuses to gate.
Feed America publishes the entire directory of 327,000+ verified food-assistance locations as open data under Creative Commons BY 4.0. Anyone — commercial, non-commercial, individual, organizational — can use the data with attribution. This is unusual for a US nonprofit operating a directory at this scale. This document explains why we made that choice and what we ask in return.
Five principles
1. Federal data is already public. So is ours.
USDA Food and Nutrition Service publishes the SNAP retailer database. HRSA publishes the FQHC locator. State WIC agencies publish clinic registries. NCES publishes school-meal data. All of it is public-domain federal data. The work we add is integration, verification, and standardization — not gating.
2. Gating costs more than it earns.
Most large food-help directories charge for API access, charge pantries to be listed, restrict bulk export, require account registration, or sell visitor data. Each gating mechanism has a business rationale — but each acts as friction that prevents the data from reaching the families that need it most.
3. Open standards multiply impact.
By implementing HSDS 3.0 Open Referral conformantly, our data flows automatically into 211 networks, Findhelp/Aunt Bertha, United Way 211, Unite Us, and any HSDS-compatible consumer. One implementation, infinite consumers. Bilateral data-sharing agreements would have capped us at 5-10 partners; open standards reach 100+.
4. AI assistants are the new search engine.
By 2025, AI assistants had become a meaningful surface for "where can I find food help?" queries. ChatGPT, Claude, Perplexity, Bing Copilot — each was answering food-help questions with varying accuracy. The fix required structured, authoritative tools the AI could call directly. We built the MCP server in 2025 and the entity-graph.jsonld in 2026 because AI-discovery infrastructure is becoming what robots.txt + sitemap.xml were for traditional search engines.
5. Discovery friction is the bottleneck.
If the goal is reducing hunger, the marginal calorie comes from a community pantry, a SNAP-EBT purchase, a school meal, a WIC voucher. Feed America doesn't operate any of those. What we do is reduce discovery friction: making it 60 seconds easier for a hungry family to find which pantry is open today within 5 miles, instead of 60 minutes of phone-tag with stale listings. Open data scales discovery-friction reduction because anyone can build on top of it.
What becomes possible
The compounding effects of open data + open API + open standards:
- 211 dispatchers can route hungry callers to nearby pantries with verified hours, without our staff needing to negotiate with each state's 211 program.
- Hospital discharge planners screening patients for food insecurity can pull our HSDS feed directly. No DPA, no per-site licensing, no integration meetings.
- AI assistants can answer "where can I find a food pantry near 77065" with structured, sourced answers.
- Academic researchers at universities can cite our dataset (CC BY 4.0 is journal-compatible) and build food-desert mapping, intervention effectiveness analyses, and hunger-rate correlation studies.
- Civic-tech projects — hackathons, grad-student capstones, open-source libraries — don't have to start over.
What we ask in return
If you've built something on top of Feed America data and we don't know about it, please email partners@feedam.org. We track downstream usage to prioritize improvements.
The harder thing
Publishing data openly is the easy part. The harder thing is keeping it accurate. Open data without verification is just a spreadsheet of pantries that may or may not exist. Feed America invests heavily in:
- Federal-primary ingestion with quarterly refresh cycles
- Operator-claim flow at /pantry — pantry operators add live "open today" status
- User feedback loops — wrong-info reports auto-deactivate stale records
- Hours staleness decay — 180-day timer prevents zombie listings
- Geocode + phone validation — automated quality checks on every record
The verification work is what makes the open data useful. The CC BY 4.0 license is what makes it scalable. Together, they're the leverage.
Funding
Feed America operates without staff payroll. ~98% of donor dollars goes to platform infrastructure (Cloudflare Workers + D1 database serving 327,000+ verified locations), data verification, community partnerships, and open-data publishing. ≥85% program-spending ratio commitment. Annual financials at /transparency.
If open data is what you value, donate to keep this going.
License
Feed America directory data: Creative Commons BY 4.0. Federal-source data: public domain (re-published in standardized formats). Editorial content (articles, press releases, founder essays): © Feed America with the same CC BY 4.0 license.
— Sharika Parkes
Founder, Feed America (EIN 92-1761881)
Houston, TX · 2026-05-04