Foursquare Releases Its Places Data as Open Source

Foursquare has announced a new contribution to the open-source community. The location-based service company is making its foundational open data set, Open Source Places, generally available. Containing more than 100 million global places of interest (POI), it can be used commercially under the Apache 2.0 license framework.

“As we enter a new frontier of technology innovation that connects computer systems to the physical world (also known as Spatial Computing), a mission-critical layer is helping these systems understand their context — in part or in whole — relative to physical places,” Foursquare Chief Executive Gary Little writes. “This makes having a comprehensive and accurate understanding of the surrounding world foundational to the technology, otherwise your autonomous system may navigate to the wrong drop off point or your augmented optical experience may hallucinate resulting in poor customer experiences and so on.”

Building a Better Location Mouse Trap

Originally a repository for all the location data Foursquare gleaned from user checking, Places became challenging to maintain. The company states that “while effective in capturing nuanced insights,” there were technical and capital resource limitations. It claimed a “substantial backlog” in location edits and user activity that needed approval, wreaking havoc on update timelines. To do it effectively, human-AI collaboration was required.

That led to the development of Foursquare’s Places Engine, an operating system that leverages humans and artificial intelligence to “actively curate a database representative of the current world.”

Foursquare Places contains 22 core attributes, from the POI basics (e.g., name, address, region, city, country, website, social media handles, and latitude and longitude) to category.

The Message to Superusers

A crucial component of Places has been Foursquare’s human contributors. These so-called “Superusers” essentially operated like Reddit moderators, helping police edits to businesses. The more dedicated users gained more prominent status and had greater autonomy.

In an email to this community, of which I am a part, Little and Foursquare’s Chief Technology Officer Vikram Gundeti, expressed their appreciation for the work Superusers have done for the platform and said, “the years of careful curation and maintenance by the Superuser community will now have an even broader impact, helping developers, businesses, and communities worldwide build better location-based experiences.”

They describe Superusers as playing a central role in the company’s effort to make Places the “foundation for countless innovations in how people discover, experience, and interact with the world around them.”

“As stewards of this data, you’ll be helping to power not just Foursquare’s applications but potentially thousands of services and applications that will shape how people experience places in the future.”

The company plans to eliminate its “Superuser” branding and replace it with “Placemaker” in recognition of these users as “the guardians and curators of our open Places dataset.”

Foursquare also teased that it will launch a redesigned Superuser portal in December with improved interfaces and “efficient workflows” for curating real-world data.

A Gesture Back to the Open Source Community

By making Places open-source, Foursquare is helping any company looking for a geospatial component for its technology. Of course, this will be invaluable to AI vendors. For Foursquare, it’s one way to give back to a community from which it has significantly benefited.

“In our software development, we have been able to more efficiently innovate by leveraging generalized open source capabilities like Kafka for data streaming, PyTorch for machine learning, and Apache Iceberg for data storage. Our geospatial capabilities are enhanced through specialized tools like Uber’s H3 indexing system, Apache Sedona’s spatial framework, and OpenStreetMap’s (“OSM”) comprehensive mapping data,” Little states. “Unfortunately, in geospatial, location, and mapping software, the data layer remains largely the provenance of large-scale proprietary systems. The walled-garden nature of the data layer greatly hampers the industry’s ability to go from strict specialization to generalized adoption, and it is in the general adoption layer that the real value to customers exists.”

Developers would also welcome open-sourced geospatial data, as they would no longer rely on mapping information from Big Tech firms like Google. Foursquare’s location dataset has some heft. It has been used to predict retail foot traffic, help Coca-Cola inform sales teams on new store openings, enable Uber customers to get to specific locations on a map, and more. Places could also power BeeBot, an audio guide app developed by Foursquare’s co-founder Dennis Crowley.

Innovating on Swarm Means More Places Data

Today’s announcement comes weeks before Foursquare sunsets its City Guide app as the company shifts its focus to its Swarm app. It seems more valuable data comes from checking in instead of a Yelp-like service. By dedicating additional resources to Swarm, the reward will be that Foursquare will harvest more location data and offer it to developers. And the cycle continues, creating a boon for its enterprise strategy.

When will an updated Swarm app be made available? Foursquare didn’t provide any specific date, but it said it was coming “very soon.”

Featured Image: AI-generated image of a city map with multiple location pins scattered throughout. Image credit: Adobe Firefly