The Debugging Decay Index: Rethinking Debugging Strategies for Code LLMs

Muntasir Adnan, Carlos C. N. Kuhn · 2025

Read The Debugging Decay Index: Rethinking Debugging Strategies for Code LLMs by Muntasir Adnan, Carlos C. N. Kuhn (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-18403.txt.

Description

Ask a code-writing model to fix its own bugs again and again, and its skill fades fast. Muntasir Adnan and Carlos C N Kuhn measure that fade in a 2025 paper. Most models, they find, lose 60 to 80 percent of their debugging power within two or three attempts, following a predictable exponential decay.

Their Debugging Decay Index turns the fade into a number that flags when further attempts stop paying off. The remedy they test is a well-timed fresh start: abandon the stuck attempt and explore anew at the moment the index recommends. Timed restarts rescued much of the lost effectiveness. The authors offer the index as the first quantitative guide for planning how machines should iterate on code.

Catalog record

CREATORMuntasir Adnan, Carlos C. N. Kuhn
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-18403

Browse the full library · About the Tower of Babel