New method finds fair ways to split items using noisy data
Researchers crack a core fairness puzzle by figuring out how many questions you need to ask.
Deep Dive
Researchers have developed a method to fairly divide items between two people when their preferences can only be measured with error. They determined the precise number of noisy queries needed to find an envy-free split, scaling with the number of items and the fairness gap. The algorithm uses simple thresholds, runs in polynomial time, and their theoretical bounds prove its efficiency is optimal, even when allowing unlimited computation.
Why It Matters
This provides a practical, mathematically sound framework for fair division in real-world systems with imperfect information.