Wikipedia:Reliable sources for software articles

The Wikipedia articles should rely on and reference reliable sources. Being an essay, this document provides a commentary on application of corresponding policy to the articles about software. Software includes all code or programming meant to be operated by a computer or dedicated computing device such as a game console.

Sources normally considered reliable
The sources of unquestionable reliability simply do not exist; though generally the following criteria define the source that would likely be seen as reliable by the majority of editors:
 * The books, scientific papers and mass-media articles by the authors who are both generally accepted as the experts in the field and are independent from the topic of the article.
 * The books, mass-media articles and blogs, published by the people and organizations with a reputation for accuracy and fact-checking.
 * The articles in the online media, which is generally accepted as a reliable source of software-related information.

Adequateness of sources
The choice of sources should heavily rely on the material they are supposed to support.

User communities
The games' fan sites and user communities can be a valuable source of the in-depth information about the software. Still, such sources ordinarily lack any editorial control and may contain the false information, so generally they should not be cited until the statement they are supposed to prove is not cited elsewhere; even in that case such information should be removed from the article if anyone expresses doubts about its inaccuracy.

Never cite forum posts and the writings by unknown authors. Avoid quoting such sites or reusing quotes from there.

Directories and download sites
The sites that are supposed to inform users about the existence of software or provide downloads (such as Softpedia or SourceForge) can only prove the availability and release dates of the software, though primary sources are still preferable in determining these facts. The publishers' descriptions on such sites should be considered yet less reliable than the primary sources. The editors' reviews in such sources should be used carefully. A software repository may describe the software's feature set, edited independently of the software's own publisher.

Online media and blogs
The decision to cite an online media should be based on its reputation for fact-accuracy and depth of coverage, which should match the claim's impact on the article. The important claims should be backed with solid, undisputable reputation.

Nevertheless, if the source has a known, widely acknowledged bias, the same rules as for the primary sources should be applied to the material that is related to such bias.

It is also important to make difference between the blog, hosted by reputable mass media source, and the mass media source itself: being a blog post, the cited source may or may not be subject to editorial overview and/or reputation of the source for fact checking, independence of the top and other properties of publisher, which make it reliable.

The less important are the claims, the less strict rules should apply. E.g., the claims about product X being criticized for reason Y could be supported by virtually any source, including the uncontroversial Slashdot posts, reputable blogs (e.g. Lifehacker) and local newspapers.

Depth of coverage
The sources specializing in the field of computing and software should be generally considered more reliable than the general public sources. The editor of Washington Post is likely to possess less knowledge of the software topic than the editor of Ars Technica. Thus the description of the features, algorithms or implementations of the software should be supported with the sources known for better understanding of such details.

Mailing lists and conferences
The mailing list and developer conferences (like FOSDEM) closely related to the subject are a valuable source of community-reviewed information from the people and organizations with the in-depth knowledge of the subject. Though, reading the whole thread of a mailing list discussion (or the reviews of the conference) before actually citing the material may save a lot of editors' time and effort.