Towards Designing Usable Languages

Edit, compile . . . edit, compile. This tireless cycle dates back to the 1960s, when the cost of editing and compiling was substantial.

Despite this being a long-standing interaction between human and computer, we have only recently linked the observable behavior of a novice programmer with affective states. That's a fancy way of saying, "We can detect when students are frustrated." Our short-term goal in this study is to support the learner with sensible interventions based on automated observation of their interactions with the compiler and their environment. Longer-term, we hope our work helps provide a human-centered foundation for the design of languages and their environments.

To this end, our language design target is ambitious: parallel languages for robotic control. We've built a small virtual machine to support message-passing parallel languages (the Transterpreter) and have begun exploring its use on small, microcontroller-based mobile robotics platforms. In part, we felt this was good engineering: begin by exploring, understanding, and reusing well-tested and formally verified languages with a rich 20-year history. Also, we thought more people might use our tools if they could play with them on robots made out of shiny yellow plastic. In short, ours is a story about people trying to do some cool stuff at the intersection of usability research and the design and implementation of parallel programming languages.