Despite its reputation as humanity's most reliable truth-seeking method, science may systematically optimize for the wrong variables. A new paper published on arXiv argues that scientific discovery should be examined as an optimization problem—and the results are troubling. Rather than converging on fundamental truths, research trajectories appear locked into local minima, constrained by path dependence, institutional reward structures, and funding mechanisms that prioritize publishability over accuracy. The authors contend that once a scientific community commits to a particular theoretical framework or methodology, switching costs become prohibitively high, even when alternative approaches might yield deeper insights. This lock-in effect applies across disciplines: a researcher's career trajectory depends on citations within an established paradigm, creating perverse incentives that favor incremental confirmation over paradigm-shifting discovery. The implications are profound—entire fields may be optimizing toward dead ends while better explanations languish unexplored simply because they require abandoning established reputational investments.