Zohar Barak's RR-CWM mechanism breaks deterministic barrier with 1.27 ratio

Zohar Barak's new paper 'Facility Location Mechanism Design -- Breaking The Deterministic Barrier' tackles a classic problem in algorithmic game theory: placing a single facility to minimize total travel distance for agents who may strategically misreport their locations. The key innovation is RR-CWM, a randomized strategyproof mechanism that achieves an expected approximation ratio of 4/π (about 1.27) for 2D Euclidean space. This strictly beats the best possible deterministic mechanism (ratio √2 ≈ 1.41), closing a long-standing open problem. In higher dimensions (d), the expected ratio stays between 1.41 and 1.547, showing strong performance regardless of space dimensionality.

Beyond the core result, Barak demonstrates that RR-CWM can be combined with machine learning predictions for improved performance in the learning-augmented setting, achieving better consistency-robustness trade-offs than prior work (Agrawal et al. 2022, Barak et al. 2024). The paper also establishes a matching lower bound of 4/π for the Generalized Random Dictator (GRD) class of mechanisms—meaning no mechanism that only picks an agent's reported location can outperform RR-CWM. For higher dimensions, the GRD lower bound is √2 - O(1/d). These results have implications for any application where truthful reporting and near-optimal facility placement are critical.