SEP 8, 2026 · PREPRINT
Do Reviewers Still Reward Lexical Complexity? A Frozen-Rater Study of Preference Drift in 124K ICLR Reviews
arXiv
A novel observational study using computational methods to infer reviewer preference drift from 8 years of peer review data, with clever design but no direct measurement of causal mechanism and applicability limited to one venue and evaluation model.
PreliminaryJiabin Zheng
Reported
Human coefficient on non-domain l…+0.142
Human coefficient on non-domain l…-0.015
Frozen rater coefficient on non-d…+0.080