Wikipedia:Notability/RFC:Notability of free open source software

This is a proposal to rethink the notability guidelines for free open-source software (FOSS) under active development, with public repositories, and with a sizable developer community and user base. In particular, sources which would otherwise not be considered reliable (e.g. technology blogs not as popular as the Linux Journal) nevertheless indicate notability, because their apparent lack of reliability is coupled with the intrinsic verifiability of FOSS (claims can be tested by downloading the source code).

Preamble
There has been a recent intensification in the trend of deleting software-related articles. This can be seen as a useful enterprise, attempting to declutter Wikipedia of dead projects or software with very little information to go about it.

Among those deletions there have been cases of articles on active free open source software projects, with large user bases, being deleted on grounds of lack of notability (e.g. Foswiki, MojoMojo).

While notability is certainly desired for any article, the situation of open-source software presents some particular facets. FOSS benefits from instant verifiability, in that anyone can download the code and check the claims in the article. Sure, an editor writing in detail about their discoveries this way would constitute original research. On the other hand, fact checking in comparison tables simply requires referencing the software's documentation, or live demos (example: Comparison of JavaScript frameworks).

Often, the size of developer base, and automatically-generated statistics about the project longevity and activity can be found on sites such as Ohloh (example for Foswiki - https://www.ohloh.net/projects/foswiki) or GitHub (example for MojoMojo - http://github.com/marcusramberg/mojomojo/). Most such software is not the "subject of multiple, reliable, independent, non-trivial, published works", and most can never be. Discounting web reviews and blogs as references is disconnected from reality in the case of the "paperless encyclopedia".

Problems with the AfD process for FOSS articles
Given that open source software developers generally are not registered Wikipedia users, the AfD discussion of FOSS articles suffers from specific problems.

The ban on canvassing
The ban on canvassing for votes and the "Not a vote" mantra are used to keep people who are knowledgeable but don't regularly edit Wikipedia from participating. While vote canvassing is a reason to ignore a headcount, it should not be a reason to ignore the arguments and evidence from the newcomers. Arguments and evidence are just as good whether they come from canvassing or not.

Single-purpose accounts
FOSS developers generally do not edit Wikipedia. However, they possess valuable knowledge of a project. While single-purpose accounts generally indicate dubious credibility when they are created to intervene in an AfD of an ordinary topic, this is most often not the case with FOSS developers. As Aaron Swartz's study Who writes Wikipedia showed, the largest contributions are made by unregistered users.

Expertise and WP:COI
WikiGnomes perform an invaluable function of keeping Wikipedia clean. At the same time, they are experts in a specific range of fields. When their expertise does not include the FOSS in question, the opinions of FOSS experts should be given more weight, whether they are established Wikipedia users or not. Who else has better knowledge of an open-source project than its developers? If the article is not worded like an advertisement, Invoking WP:COI is unrealistic, especially as no commercial interests are at stake. FOSS developer Wikipedians who disclose their affiliation with a FOSS project should not be accused of COIs or of sock-puppetry. In the end, it should matter what is written and how it's supported -- not who wrote it.

Following is Open Directory Project's policy on editors with affiliations :

The goal of editors should be to produce useful resources for the web public. We do not bar editors with business affiliations, since those editors with their own sites usually know their competition and related sites better than anyone. This knowledge can be ideal for helping build an authoritative directory. However, we will not tolerate editors who only add their own sites, or maliciously interfere with others' listings in the directory.

Reliable sources
The requirement for third party reliable sources for software projects done over the Internet is broken. People don't generally write books or even cnn.com articles about free software projects except for the very biggest ones.

Another problem with WP:RS is that the verifiability of the claims made about the FOSS is not under question - anyone can download the source code and check the facts, and in many cases there are directly accessible online demos. Unreliable sources are not used to establish facts, but notability. For instance, if Rush Limbaugh talks for weeks about a software project, it's probably notable.

Common sense
In the case of Foswiki, the AfD discussion involved a free software fork of a project that was notable (Twiki) but has been taken proprietary, where all but two developers moved to the free project. While obviously a forked project doesn't inherit notability every time, if common sense were used, the Foswiki fork would be deemed notable by any standard.

WP:EGO
Ego should not play a role in AfDs, but it can be observed that the more supporters and keep votes a FOSS article has, the more editors vote for its deletion (which suggests canvassing for deletion). Some FOSS projects generated very short AfD discussions, with less than 4 votes: DidiWiki, Noösphere, Ruwiki, SnipSnap, Wiclear, WWWiki. Other articles generated heated debate: Foswiki, MojoMojo. This very fact is an indication of, if not notability, engagement, and a passionate user base.

NPOV
It should be noted that none of the AfD nominators of the Foswiki and MojoMojo articles voted in the AfDs of any of the other articles mentioned in the previous heading. Also, only two of the dozens of deleters at Foswiki and MojoMojo, voted at all in the AfDs for the other wikis. This casts doubts on the neutrality of the deletion process.

Improving Wikipedia
While this may conflict with deletionist views, removing articles about large active FOSS projects, or removing software entries from comparison lists in general does not help improve Wikipedia.

Alternatives to deletion
Since most FOSS developers are not familiar with Wikipedia policies, the first time they find out about the guidelines for deletion is when they are used on their article. At that point, it's often too late to do anything about it.

Instead of marking articles for deletion, editors should consider following WP:PRESERVE and using the refimprove and unreferenced templates.

Conclusion
While maintaining articles on FOSS software may be desirable, at the same time Wikipedia is not a directory. However, the types of reliable sources allowed by Wikipedia typically do not mention open source software, which leads to a systemic bias against it. The aim of this proposal is to rethink the class of sources accepted for notability so that they align better with the nature of sources dealing with free open source software.