Sources and posts

A list of posts and the sources they draw on.

Sources

AI agents “MUST NOT add Signed-off-by tags”; the human submitter signs off, “taking full responsibility for the contribution”

Linux kernel docs, AI Coding Assistants

Accessed
Sample and type
project documentation

“34% of organizations that experienced an incident in the past year could not actually make that determination” (whether AI-generated code contributed)

GitLab, AI code research (press release)

Published
Sample and type
1,528 developers and technology buyers across six countries, conducted by The Harris Poll, self-reported
Disclosure
GitLab sells a DevSecOps platform with AI features.

73% identify review volume, senior-reviewer burden, or both as a top-two problem; 38% already let some agent PRs merge with sampled or no human review

Span, Coding Agents in Practice: Q4 2026

Published
Sample and type
survey of over 100 engineering leaders, self-reported
Disclosure
Span is a vendor of engineering-intelligence software.

Developers and leaders named reviewing and validating AI-generated code as the top delivery bottleneck; only 3.7% of leaders say existing processes are sufficient to maintain quality and governance as agents take on more work

Qodo, State of AI Code Quality Report 2026

Published
Sample and type
500 U.S. developers + 300 engineering leaders, self-reported
Disclosure
Qodo is a vendor of AI code-review tools.

About and methodology

Most figures on this page come from self-reported surveys. GitHub, Span, Qodo and GitLab sell related software, and Genpact sells AI services. The GitHub figure is platform data from GitHub. The Linux kernel entry is project documentation. Figures are quoted as published. Original links are unchanged.

Ledger last checked 17 Nov 2026.

Our first Instagram Reel and TikTok video (the review bottleneck) use an AI-generated voice and AI-generated illustrations. The ownership videos use Martin’s own voice, real screenshots, and stock footage by Pavel Danilyuk on Pexels.