Reproduce “From Restructuring to Stabilization: A Large-Scale Experiment on Iterative Code Readability Refactoring with Large Language Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Norman Peitek, Julia Hess, Sven Apel
The authors track how OpenAI GPT-5.1 refactors Java code across five successive passes, comparing original files with versions whose identifiers or comments were altered or removed. They measure changes between versions under general and targeted prompts. For the Meaningless variant under the general prompt, unchanged lines rise from 31% to 90%, indicating stabilization, not demonstrated improvement in readability or correctness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.