Michelle M. Pettit
C++
99% — 85.5 of 86 orders sound, per qwen3.8-27b; in the top 42% of 51 graded judges
On the assumption that the model is right — qwen3.8-27b's verdicts on a court's own orders matched a later appeal 76% of the time (104 of 136), graded on a curve: A is the top fifth of the judges graded under this model, F the bottom fifth; ++ / + / − / −− within the band. How the grade is made.
The Ninth Circuit's later rulings on their dockets
affirmed 2
Each Ninth Circuit opinion that names a docket this judge wrote an order on, by its outcome; the ruling it reviewed is not always the order in the record. The order's page links the opinion.
Analysis
No analyses yet. When they exist, this shows counts of the model's findings across this judge's reviewed cases — with the court's own dispositions beside them, and only from 10 cases.