Abstract
Between 2017 and 2020, a team of researchers from the George Washington University collaborated with NASA and Freelancer.com to design and launch the "Astrobee Challenge Series," a large-scale field experiment that aimed to generate data to characterize the relationship among how a technical problem is formulated and who is able and willing to solve, and the quality of solutions they generate. The core experimental manipulation was of the architecture of the problem posed; the typical open innovation process was instrumented to collect unusually rich data but otherwise untouched. In all, 17 individual contests were run over a period of 12 months. Over the course of the challenge series, we tracked a population of 16,249 potential solvers, of which 6,219 initiated solving, and a subset of 147 unique solvers submitted 263 judgeable solutions. The resultant dataset is unique because it captures demographic and expertise data on the full population of potential solvers and links their activity to their solving processes and solution outcomes. Moreover, in addition to winning designs (the typical basis of analysis), it captures design outcomes for all submitted design artifacts allowing analysis of the variety of solutions to the same problem. This data explainer documents the research design and implementation process and provides a detailed explanation of each data record, carefully characterizing potential limitations associated with research design choices. This data should be useful for researchers interested in studying the design and innovation process, particularly those focused on novelty, variety, feasibility of solutions or expertise, diversity and capability of solvers.