Skip to content

Software Quality

About this project and how it is checked

How the content here is checked, and what that checking does not cover.

This site surveys the ways of finding out whether software works — one page per method, from types and static analysis through tests and fuzzing to monitoring and mathematical proof. Methods is the catalog, the axes the classification behind it.

How claims are checked

  • Citations are checked against a primary source. A work is read, or its identity confirmed against a bibliographic index, before it is cited. Where the primary text could not be reached — a paywalled standard, a claim confirmed only from a secondary source — the entry records that instead of dropping the caveat.
  • Peer-reviewed work is the default source. Journal and conference literature carries most of the evidence here. A claim resting on an industrial case study, a preprint, or a tool's own documentation is labeled as such rather than presented as a finding.
  • Evidence is labeled by strength. A measured result is reported as a finding, thin support is marked as a plain note, and folklore is named as folklore — see Testing folklore and Design folklore.
  • Adoption is not evidence. Popularity, longevity, and authority don't stand in for a measured result. Where a practice has no empirical support, the page says so rather than arguing from how many teams follow it.
  • Every method is described in the same frame. The same axes and the same questions on every page, so two methods can be compared instead of each being described on its own terms.

What this does not guarantee

Working copy

This is a working copy (preprint) — possibly containing errors. Pages vary in maturity, and nothing here has been through review by a subject expert.

The author makes no claim to expertise in every area covered here. The aim is narrower: to organize what's known and hold each claim to a stated check.

Author

stereobooster has worked as a developer for 20 years, in fintech, travel, marketing technology, and learning technology. Code is at github.com/stereobooster and writing at stereobooster.com.

Citing this work

Nothing here carries a version number, so a citation records the date the page was read.

@misc{stereobooster2026,
  author = {stereobooster},
  title = {{Software Quality: a field guide to knowing whether your software works}},
  year = {2026},
  url = {https://quality.stereobooster.com/},
  note = {Working copy. Accessed <date>.}
}

A single page takes its own title and URL in place of the site's — title = {{Mutation testing}}, url = {https://quality.stereobooster.com/mutation-testing/}.

The content is CC BY-SA 4.0: attribution, a link to the license, and the same license on anything derived from it.

Corrections

A correction, a study that changes a conclusion, or a method missing from the catalog belongs in the issue tracker. That repository holds the Claude Code skills distilled from this knowledge base; its issues are the channel for the content here as well.

Support this work

This knowledge base is unfunded and free to read. Support goes through thanks.dev.

Referenced by