Improving the Security of United States Elections with Robust Optimization
提出一种基于鲁棒优化的低成本方法，改进已有百年历史的投票机逻辑与准确性测试（LAT），确保能检测出导致选票跨候选人交换的配置错误，并在密歇根州2022年11月选举中仅增加1.2%成本，自2023年夏季起已在真实选举中试点成功。
The preservation of democracy hinges on the reality and perception that votes in elections are counted properly. In the paper “Improving the Security of United States Elections with Robust Optimization,” Crimmins, Halderman, and Sturt provide a low-cost approach to reducing the security risks of voting machines that are used to scan ballots and count votes. Their approach consists of applying robust optimization to a century-old testing procedure called logic and accuracy testing (LAT), which is performed by election officials on each voting machine before each election. The authors show that their robust optimization approach is guaranteed to detect any misconfiguration of voting machines that would cause votes to be swapped across candidates. Applying their approach to Michigan’s November 2022 election, the authors show that their approach to LAT would have only required a 1.2% increase in cost to election officials compared with current practice. Their approach, which is forthcoming in the Societal Impact section, has been successfully piloted in real-world elections by the Michigan Bureau of Elections since Summer 2023 as a cost-efficient way to enhance election security and public trust in election outcomes.