Crowdsourcing suggestions to programming problems for dynamic web development languages

Developers increasingly consult online examples and message boards to find solutions to common programming tasks. On the web, finding solutions to debugging problems is harder than searching for working code. Prior research introduced a social recommender system, HelpMeOut, that crowdsources debugging suggestions by presenting fixes to errors that peers have applied in the past. However, HelpMeOut only worked for statically typed, compiled programming languages like Java. We investigate how suggestions can be provided for dynamic, interpreted web development languages. Our primary insight is to instrument test-driven development to collect examples of bug fixes. We present Crowd::Debug, a tool for Ruby programmers that realizes these benefits.