Holding Real-Time Distributed ML Systems to Account

Trade-offs between accuracy and efficiency are found in multiple non-computing domains, such as law and medicine, which have developed rules and heuristics to guide how to balance accuracy against efficiency in conditions of uncertainty. While accuracy-efficiency trade-offs are also commonly acknowledged in some areas of computer science - perhaps to the point of mundanity - their policy implications remain poorly examined. We argue that, since examining accuracy-efficiency trade-offs has been useful for guiding governance in other domains, explicitly framing such trade-offs in computing is similarly useful for the governance of computer systems. Specifically, we focus our discussion on accountability mechanisms for real-time distributed ML systems. Understanding the policy implications in this area is particularly urgent because such systems, which include autonomous vehicles, tend to be high-stakes and safety-critical. We describe how the trade-off takes shape in real-world distributed computing systems and then show the interplay between such systems and ML algorithms, explaining in detail how accuracy and efficiency interact in distributed ML systems. We close by making specific calls to action to facilitate holding these systems, particularly the emerging technology of real-time distributed ML systems, to account.