nth Labs — writing Product QA for people who ship fast: what breaks, what lies, and how to catch both. https://nthlabs.dev/feed.xml 2026-08-28T00:00:00Z nth Labs Software production is booming. Who is checking it? https://nthlabs.dev/blog/who-checks-the-code 2026-08-28T00:00:00Z Repositories, commits and merged pull requests all rose by a fifth or more in a year, and the code being produced is measurably less maintained than it was. Five studies, read side by side. A confirmed bug becomes a standing check https://nthlabs.dev/blog/a-confirmed-bug-becomes-a-standing-check 2026-08-25T00:00:00Z Every tool can generate tests. The hard part is knowing which of them could ever fail. A re-check is not counted as protection until it has been seen to fail. 36 out of 36 is not 100% https://nthlabs.dev/blog/thirty-six-out-of-thirty-six-is-not-100-percent 2026-08-21T00:00:00Z The confidence interval everyone reaches for returns negative rates, rates above 100%, and zero uncertainty from a perfect score. Exact coverage curves, and the one-line fix. Testing behind a login, without your password https://nthlabs.dev/blog/testing-behind-a-login-without-your-password 2026-08-18T00:00:00Z Hand a run the session your own browser already has. SSO and 2FA are no obstacle, because nothing here ever sees a password or a login form. Which accessibility issues can be automated? https://nthlabs.dev/blog/which-accessibility-issues-can-be-automated 2026-08-14T00:00:00Z Automated checkers ship around a hundred rules. Counted against the spec they land on 29 success criteria, and 29 of those rules cover one criterion between them. What a machine can and cannot settle, counted from axe-core 4.12.1. How the persona panel picks its testers https://nthlabs.dev/blog/how-the-persona-panel-picks-its-testers 2026-08-11T00:00:00Z Six simulated testers, drawn from ten thousand records, matched to the audience you describe. How the draw works, what it refuses to model, and what it is not. You are not your user, in five numbers https://nthlabs.dev/blog/you-are-not-your-user 2026-08-07T00:00:00Z In a 10,000-record research dataset of simulated users, 20.8% are modelled as trusting. 27.3% write code. 12.2% are digital natives. Charts, denominators, and what it means for what you build. How to tell a website was built by AI https://nthlabs.dev/blog/how-to-tell-a-website-was-built-by-ai 2026-08-04T00:00:00Z There is a house style now, and once you have seen it you cannot stop seeing it. Seven things anyone can spot without opening developer tools.