Mapbox

Company technologies

Mapbox GL JS for web, iOS, Android, React Native, Unity, Qt.

Software

Our focus for releasing software as open source is on reusable, general-purpose parts. In many cases, this means modules, rather than applications. We distribute these modules for other developers to use on websites like npm and PyPI . For instance, rasterio is a core part of our image processing pipeline that we make available for any Python programmer to use.

A minority of our software is closed-source:

• Software that connects server processes and is heavily dependent on our configuration. This is a prevalent category but tends to be very small, in terms of code.
• Software that packages up business workflows. Software highly specific to Mapbox’s product, tied into our design, and core to our competitive advantage is often kept closed. These projects are usually applications and servers composed of parts that we release as open source so that others can build from the same foundations.

Standards

Standards make modern software possible by connecting modular parts into systems. Open formats are the groundwork of open data. Mapbox is built on open standards: we adopt existing specs like GeoJSON and author new standards for new purposes, like MBTiles and GL Style Spec . We release them under permissive licenses that enable everyone to freely implement them in products and software, and to improve upon them.

Design

Design components that are reusable, general purpose, and not explicitly tied to the Mapbox brand, like Maki , make sense as open source. We also release templates that show how to use our tools, like those in Guides .

Most of our artwork is copyrighted and not placed under any open license. It’s tied to the Mapbox identity: reusing the Mapbox.com website style isn’t a use-case we want to support. The uniqueness and specificity of our style is important to how customers identify “oh, this is Mapbox”.

Data

The vast majority of Mapbox’s data is in the open. Large sources include OpenStreetMap , USGS , Landsat , Natural Earth , and OpenAddresses . We contribute to these sources, participate in their communities, and invest in tools that help improve them.

We also work extensively with proprietary sets of data that we buy: we legally can’t open these. Nor do we release heavily processed data. Processing data and delivering the final products has obvious and real costs: server infrastructure and bandwidth, as well as labor. Our ability to distill meaning out of raw material is a core part of our product.