Understanding change in jurisprudence
利用基于大语言模型标注的最高法院意见数据集，研究1870至2024年间大法官法理学立场是否随职业生涯变化，发现多数人不变，约15%-20%有变化，且变化与意识形态无关，可能与同僚效应有关。
Abstract Does a legal formalist stay a formalist? Using a new measure of jurisprudential commitments based on an original dataset of annotated Supreme Court opinions leveraged through a fine-tuned large language model, we study whether, how, and why justices change their jurisprudence over the course of their careers, from 1870 to 2024. We find that most justices do not change significantly over their careers: they start how they end. However, roughly 15%–20% of justices exhibit change during their careers. Though we observe heterogeneity in the direction of jurisprudential movement over the series, locally it tends to be homogeneous: for example, we see anti-formal shifts after the judicial revolution of 1937. We investigate two mechanisms for shifting jurisprudence: we find little evidence that it is related to changes in ideology; we find suggestive evidence that it is related to peer effects and changes of the jurisprudence of other members on the Court.