If your organization is working on an open data release and your goal is to maximize the reach and impact of your data, sometimes just releasing the data and tools isn’t enough to accomplish your goal. Derivative products — like custom maps that visualize key data — extend the reach of data even further and help reach people who will never use complex tools or know how to meaningfully manipulate raw data.
That’s why this week when the National Democratic Institute (NDI) launched an open data site for election monitors in Afghanistan, they also released 14 sets of custom map tiles created using our TileMill project to make the data more useful to end users.
The rest of the site is designed to help users combine different datasets from the past three national elections in Afghanistan into helpful visualizations that give greater insight into the election processes. For instance, the site lets users see fraud incidence overlaid on a map of security issues from the 2009 presidential election, which can help them better understand correlations between violence and fraud. Many of the datasets don’t provide obvious insights on their own, but correlations become more apparent when the datasets are combined. These visualizations are one of NDI’s key value additions to the election process that are made possible by the site.
More Than Just Data
Our team worked with NDI to create the new open data section (following the Obama Administration’s Open Government Initiative namespace protocol), which makes much of the source data visualized elsewhere on the site available for download. Like other open data releases, the major goal of this section is to empower interested organizations and individuals to run their own analysis of the data and use it in their own applications. Making the data available to others extends its reach and impact, improving transparency and creating greater efficiency among the wide group of election monitoring organizations. This is the theory with most open data projects, but in this case NDI decided to release more than just data — they also released maps and documentation to go with the data.
Why not just release raw data and let others figure out how to use it? Most of the election data on the site has a geographic component, and some of the data includes geo-specific KML files that are designed to be viewed overlaid on a map. The intent of the open data site is to make source data available for others to visualize on their own maps, but there was a major problem with this in practice.
As we worked with the Afghanistan team at NDI to plan for this project and talked with many of the organizations most likely to use the data — both on the ground in Kabul and back in Washington, D.C. — we realized that many of them didn’t have GIS capacity (either time or skills) to create complex maps online. Releasing the raw data without the maps would have made the data impractical for many of the target audience to leverage in their work. Because NDI also wanted to make the data useful for the average interested user, it became clear that we should use the open site to share some of the same custom maps we had created for NDI’s use.
Publishing custom maps with the most up to date province and district boundary lines puts end users of the data in a position to quickly build their own visualizations and applications using the core datasets that were released. To make map distribution as easy as possible, we agreed to host the maps on MapBox.com and provide them free to use with our standard SLA. To further maximize the use of the maps, we also made the tiles available for download in our new “.mbtiles” format, which combines the tiles into a single SQLite database so they can be used offline or in other applications, including offline with our Maps on a Stick tool that is being used by NGOs in the field. The work to create this new file format and make tiles practical to download and use in other applications is something we’ve been able to do along with our work on the upcoming TileMill 2.0 release.
Focus on End Users
“Open data” has become a buzzword on the web — particularly in government and humanitarian tech circles — and with that status comes some issues. There’s a perception sometimes that an open data release means just checking the right boxes (XML, RDF, “apps” contest, etc.) to be successful. Many open data initiatives don’t get to the point of explicitly thinking about how to help end users. At the end of the day, the intent of most open data projects is to improve efficiency and the use of the data, which also means supporting users with tools and other resources.
We’re really excited about how the ability to create and distribute custom maps stands to help improve the success of open geo-data projects like NDI’s, and we’ll be working more on these tools in the coming months so that it’s even easier to share custom maps and free open source mapping tools in the future.