Preventive measures towards vandalism on OpenStreetMap have long been non-existent, and it takes less than a minute to sign up on OSM with an email address and make an edit — accurate or egregious. With more than 3 million edits a day, data errors introduced into OSM create a substandard map experience to anyone using OSM as a basemap. At Mapbox, hundreds of millions of users view the map everyday, and data quality is crucial to our products.

This talk will focus on the approaches Mapbox has taken over the past years to ensure data quality and what we learned with each iteration. We will talk about the tools we have built, and present a new unit of change for OSM edits that, in terms of volume, is more efficient than using changesets and features. We have taken on the challenge of reviewing every edit on OSM in the shortest time possible while continuously ingesting feature updates. We will also talk about how we are sharing the reviewed data with the community, and how community can benefit from our latest review pipeline.