The original version of this article, by Aaron Lovell, appeared on the Wilson Center’s Science and Technology Innovation Program Commons Lab blog.
On June 28, officials from USAID met at the Wilson Center
to discuss a recent experiment in using crowdsourcing to help clean up and map data on development loans. The agency had 117,000 records on private loans made possible by their Development Credit Authority, which theoretically could be mapped and made available to the public. The problem? Location data for the loans was not standard and was difficult to parse. USAID decided to turn to “the crowd” and recruit interested volunteers from the Standby Task Force
– two online volunteer communities – to help clean the data, making it available for additional analysis.
Shadrock Roberts, Stephanie Grosser, and D. Ben Swartley – all with USAID – discussed the results of the project. They were able to clean up the information at virtually no cost to the government, noting that they hope it will be an example for further collaboration between government and engaged volunteers.
“Our hope is that the case study will provide others in government with information and guidance to move forward with their own crowdsourcing projects. Whether the intent is opening data, increased engagement, or improved services, agencies must embrace new technologies that can bring citizens closer to their government,” the officials wrote in a recently released case study focused on the exercise. While the exercise resulted in high-quality output, Roberts et al. did provide advice for other agencies considering similar work. Grosser stressed the need to “crawl, walk, [then] run” – that is, start small with a few hundred records and then expand on that, as the system is refined.
But USAID sees crowdsourcing becoming a key part of improving government data. “We need to be working as hard to release relevant data we already have as we are to create it,” Roberts concluded, when discussing how the experience could apply more generally. “The crowd is willing to do research, data mining, and data cleanup.”