The field of machine learning is moving towards a more nuanced understanding of model performance and decision-making. Researchers are developing new metrics and methods to evaluate model performance, taking into account factors such as human disagreement and class label hierarchies. This shift is driven by the need for more reliable and trustworthy model evaluations, particularly in high-stakes applications such as healthcare. Noteworthy papers in this area include: Forest vs Tree, which investigates the trade-off between the number of items and the number of responses per item needed for reliable machine learning evaluation. Honest and Reliable Evaluation and Expert Equivalence Testing of Automated Neonatal Seizure Detection, which proposes best practices for evaluating machine learning models for neonatal seizure detection. Sensitivity of Stability, which presents a comprehensive theoretical and empirical analysis of replicability in transfer learning.