A Framework for Incentivizing Deep Fixes

Abstract

We study the problem of how to incentivize deep fixes to software bugs, where a deep fix attempts to correct the root cause of the bug instead of just suppressing it superficially. To this end we introduce a dynamic model of the software engineering ecosystem. We then solve this problem by proposing subsumption mechanisms. In a subsumption mechanism, deeper fixes can replace or subsume shallower fixes and a worker’s payoff increases if his fix subsumes other fixes. We use a solution concept known as mean field equilibrium, an approximation methodology suited to large market settings. Taking a computational approach, we simulate the dynamic model of the ecosystem with subsumption mechanisms. Our algorithm achieves convergence and thus estimates a mean field equilibrium. We further compare our mechanism to baseline mechanisms using metrics, such as percentage of bugs receiving deep fixes, rate of bugs fixed, and cost to the user. Simulation results indicate that the subsumption mechanism performs favourably versus the baseline mechanisms.