The FOSS fragmentation problem

I could not find someone that briefly described this problem, so I’ll just do a very quick job at doing so. I take it that this is a common observation that I was just unable to find the right term for. And yes, I appreciate the metaness of this. :)

The FOSS fragmentation problem: everybody writes their own solutions (standards), but because it requires a lot of work to get any major project done right and humans consistently underestimate how long it takes to do something right (planning fallacy), there’s a plethora of half-done solutions. It’s a kind of coordination problem: if people would ‘just’ work together, we would have fewer but better solutions. There are two reasons why they generally don’t: first, people are almost always working on FOSS projects in their spare time, so there’s no boss that can order them around to focus. Second, FOSS attracts people who are prone to lone wolfery/independently minded. Larger projects with such persons tend to split.

Never claim general phenomenons without listing many examples. Okay.

Example: Linux distributions

There are at least 290 Linux distributions on DistroWatch. The familial relationship between distros has been visually summarized like this (via Wikipedia):

Think this is complicated? You can find similar figures for singly families too.

Example: Wiki systems

Given that the above example image comes form Wikipedia, we might wonder… how many Wiki systems are there? Well, WikiMatrix is a searchable list and has 141. It’s totally not comprehensive. Searching just a little bit more, I was able to find an out of date list of 20 Python-based systems. An exhaustive search of Github (and BitBucket, and SourceForge, and GitLab, and …) would turn up many more.

(I could not find any family chart. Use your imagination.)

Example: forum software

There a sister-site to the above, ForumMatrix, which lists 67 forums. And there’s a list of 19 Django-based forums, and there’s even other Python-based solutions too (e.g. FlashBB).

More examples?

You get the idea, but:

The problem extends to other domains

The problem is most commonly seen with FOSS projects, but occurs in any domain where it’s easy to set up your own version by copying from others or recreating from scratch. The primary other example I’m thinking of is, not coincidentally, licenses for FOSS projects. Wikipedia lists 54 licenses.

Rejoiner: but some of these areas are actually dominated by a few players of good quality

They are, e.g. browsers have just a few top players (but due to mobile units, this seems to be fragmenting further). However, the point remains in weaker form: if people working or who worked on all the competitors that approx. no one uses had instead been focused on, say, 5 or 10 projects, these projects would have been a lot better.