Interacting with massive numbers of student solutions

When teaching programming or hardware design, it is pedagogically valuable for students to generate examples of functions, circuits, or system designs. Teachers can be overwhelmed by these types of student submissions when running large residential or recently released massive online courses. The underlying distribution of student solutions submitted in response to a particular assignment may be complex, but the newly available volume of student solutions represents a denser sampling of that distribution. Working with large datasets of students' solutions, I am building systems with user interfaces that allow teachers to explore the variety of their students' correct and incorrect solutions. Forum posts, grading rubrics, and automatic graders can be based on student solution data, and turn massive engineering and computer science classrooms into useful insight and feedback for teachers. In the development process, I hope to describe essential design principles for such systems.