The bet
Every location-based app sits on the same layer: geocoding, map rendering, routing, search. Mapzen bet that this layer should be open source, not owned by a handful of large mapping companies, and that giving it away would seed a far larger ecosystem than keeping it proprietary ever could.
Founded in New York in 2013 and funded by Samsung, Mapzen built and open-sourced the full stack. Pelias handled geocoding, Tangram rendered vector tiles and Valhalla did routing. Alongside them came a steady release of open map and points-of-interest data drawn from OpenStreetMap and other public sources.
Peak
For several years Mapzen was one of the most productive open-source shops in the geospatial world. Its tools became foundations for civic mapping projects, transit apps and researchers who couldn't or wouldn't pay for a commercial mapping API.
Samsung backed it entirely, so Mapzen never had to raise its own funding rounds. That let it release ambitious open infrastructure with no pressure to make money from any single piece of it.
What worked
The open-source strategy did what it was meant to do for the ecosystem, if not for the company. Pelias, Tangram and Valhalla are still used and maintained today, years after Mapzen closed, because the code was well built and never sat behind a paywall. Civic and civic-adjacent projects that needed serious mapping infrastructure, and had no budget for commercial APIs, got a durable public good out of it.
What failed — the mechanism
Every other platform in this catalogue eventually has to answer who pays. Mapzen never did, because one corporate parent paid for everything. That's comfortable until the parent's priorities move.
Samsung's strategic interest in owning a mapping stack apparently did. Mapzen closed abruptly at the end of January 2018, with no public explanation beyond a short farewell message. No funding crisis was reported. No failed pivot or falling usage was cited. It came down to a single backer's decision that Mapzen was no longer worth funding.
Mapzen did wind down responsibly. It transferred legal ownership of its code to the Linux Foundation, and its user community founded Nextzen to keep the services running.
What Agora takes from it
Mapzen sits a little outside this catalogue's usual shape. It built open geospatial infrastructure, and it was never a channel between citizens and government. The funding lesson applies directly all the same. A single backer, however generous, is no business model. A project can be technically excellent, widely used and well run, and still close the moment it stops fitting one funder's strategy, corporate or philanthropic.
Agora's PBC structure and its own product revenue exist so the platform's survival is never one funder's discretionary call. Mapzen's code outlived its funder's interest. The company didn't.
Common questions
- What was Mapzen?
- An open-source mapping infrastructure company wholly backed by Samsung, whose tools are still used today, and which closed in 2018 with no explanation once its one corporate backer's interest moved on.
- Why did Mapzen shut down?
- Mapzen is recorded here as shut down. We file it under the pattern we call The business-model void, which names the recurring way this kind of platform comes apart.
- When did Mapzen operate?
- ~5 years (2013-2018).
- Mapzen — Wikipediasecondary
- Thank you for being so awesome and so open, Mapzen. RIP. — Geoawesomepostmortem · 2018-01
- GIS company Mapzen to shut down but users can still avail open-source data — Geospatial Worldsecondary
- Samsung do (quietly) own a mapping company: Mapzen — Hacker News discussionsecondary
- AmericaSpeaksThe gold standard of large-scale in-person deliberation, which convened 180,000 citizens over two decades and closed as soon as the grants that paid for it dried up.
- BlockboardA neighborhood app for reporting potholes and watching out for your block, bought by Klout (a company built to score social-media influence) and functionally ended as a civic tool the day the acquisition closed.
- BrigadeSean Parker's social network for political positions, with an estimated $40–50 million behind it, that never found enough users, or anyone to pay.