Evidence note

Novice Developers Produce Larger Review Overhead for Project Maintainers while Vibe Coding

Conference Paper202623rd International Conference on Mining Software Repositories (MSR 2026), Mining Challenge

What was examined?

An empirical analysis of 22,953 AI-assisted pull requests from 1,719 contributors in open-source GitHub repositories.

Contributors were grouped using an experience proxy based on historical GitHub contribution activity.

22,953 pull requests; 1,719 contributors.

Key findings

1The lower-experience-proxy group submitted 2.15 times more commits per pull request.
2The group changed 1.47 times more files per pull request.
3The group received 4.52 times more review comments.
4Resolution time was 5.16 times longer.
5Acceptance rate was 31% lower.

Why it matters. Lornets interpretation.

The speed or volume of code generation should not be treated as equivalent to downstream integration efficiency. Review, architecture fit and repository-specific constraints remain part of the engineering system.

This is the Lornets reading of the source, not a finding of the source itself.

What it does not establish

1GitHub contribution history is a proxy for experience, not a direct measurement of developer skill.
2Project review policies may affect the measured outcomes.
3The study concerns open-source pull requests rather than complete commercial applications.
4The reported multipliers must not be presented as universal maintenance multipliers.

Source

Venue: 23rd International Conference on Mining Software Repositories (MSR 2026), Mining Challenge
Evidence type: Conference Paper
Published: 2026
DOI: 10.1145/3793302.3793597
Status: Current

Read the original research