LinuxLinks has kindly given me the space to publish a semi-regular blog where I share my thoughts on open source, the Linux ecosystem, Windows, and other topics. It’s a place for my opinions and personal perspectives. As you might expect, everything I write here reflects my own views and prejudices.
One of Linux’s greatest strengths is choice. It is also one of its easiest excuses.
There is nothing wrong with creating another Linux distribution. Free software gives us the freedom to take existing work, change it, package it differently and release the result. Plenty of excellent distributions started that way. But having the freedom to create one doesn’t mean every new distribution has a convincing reason to exist.
LinuxLinks maintains a large list of actively developed distributions, and one thing becomes obvious fairly quickly: Debian, Arch, Ubuntu, Fedora, Void and a handful of others sit underneath a remarkable chunk of the Linux ecosystem. That’s not a criticism in itself. Kali gives Debian a very specific security focus, Tails has a clearly defined privacy model, while Bazzite turns Fedora’s atomic technology into something aimed squarely at gaming. Other projects tackle accessibility, education, embedded hardware or communities poorly served elsewhere.
They can all answer the important question: why isn’t the parent distribution enough?
A new wallpaper, a different application selection and a few extra repositories aren’t especially persuasive answers. If the main distinction could be reproduced with a post-install script, I’m not convinced it needs to become an operating system.
Running a distribution also brings responsibilities. Installation media needs testing, packages need updating, security fixes need delivering, documentation needs maintaining and infrastructure needs keeping online. Users are trusting the project with their operating system, not experimenting with another text editor.
I’m certainly not arguing for a Linux monoculture. Alpine, NixOS, Gentoo and Void show how valuable genuinely different ideas can be, and experimental distributions can be fantastic learning projects. But an experiment isn’t automatically something I’d recommend for someone’s main computer.
Perhaps Linux doesn’t have too many distributions. Perhaps it has too many projects calling themselves distributions before they’ve answered one simple sentence:
You should use our distribution instead of the one it is based on because…
If the answer is basically a theme, some preinstalled packages and somebody else’s repositories, stop right there.
Please read our Comment Policy before commenting.
Oh man! You’re going to get persecuted for saying this.
I don’t really agree with the premise. The sheer number of Linux distributions is often presented as fragmentation, but I see it more as experimentation. Most derivatives cost mainstream users nothing, and successful ideas can filter back into larger projects. If a distribution only changes defaults, desktop choices or package selection, that can still be worthwhile if it gives a particular group of users a better starting point.
The real test should be whether people find a distribution useful, not whether it introduces enough technical novelty to justify its existence. Projects will appear, evolve and sometimes disappear, but that has always been part of the Linux ecosystem. I’d rather have too much choice and duplication than impose an expectation that every new distribution must first prove that its parent is inadequate.
I, too, disagree with this premise but for a totally different reason: most people either forget, don’t realize, or ignore, that, officially, “Linux” just the kernel, only. An operating system, OTOH, is the kernel + userland + all the other stuff on top. So, when you think of it that way, what we have isn’t 300 different versions of Linux: what we have instead are 300 different OS’s which happen to use the same kernel. There is no fragmentation. We just happen to use the word “Linux” as a shorthand for Linux-based OS’s.
That’s a semantic dodge, not a rebuttal. Nobody thinks there are 300 different Linux kernels.
Calling them “300 different OSs using the same kernel” doesn’t make fragmentation disappear – it describes it. Hundreds of projects duplicating packaging, repositories, installers, testing and maintenance is exactly the fragmentation the article is discussing.
What I find more frustrating than the number of distributions is watching a genuinely competent fork unravel. Manjaro is a good example. For years it had a strong identity, with useful custom tools such as Pamac, Manjaro Settings Manager and MHWD, and it made Arch-based Linux far more approachable.
That makes its decline especially disappointing. The project now feels far less focused and much more unsettled than it once did. The real problem isn’t simply that Linux has too many distributions; it’s that good, established projects can lose their way after years of valuable work.
Yeah, their small band of developers all jumped ship.
Yeah, it’s a shadow of its former glory. Best it give up tbh.
Pretty reasonable opinion on the whole. Linus Torvalds said last year that Linux distro fragmentation is excessive enough to be a significant problem.
He’s not wrong. There are too many vanity projects. All distributions struggle for maintainers and reviewers, even something the size of Debian. More distributions just spreads a small pool of competent and willing people more thinly, so the argument that it does no harm to other projects is incorrect.
I agree 100% with the sentiment.
I think we need to define fragmentation. If you have Ubuntu, Mint, Debian, Red Hat, Fedora, Arch and some other core distros representing 90% of the user space, and 10,000 other distros representing the remaining 10%… is this fragmentation?
Let’s look at real numbers, the percentage of users. Without these numbers, is the fragmentation discussion even a serious one?
That defines fragmentation purely by user share. Linus Torvalds himself has complained about distribution infighting and fragmentation holding back the Linux desktop. The issue is duplicated effort, packaging, testing and support – not whether every distro has an equal number of users.
How many android phones are on the market, surely they should also be limited.
All produced by competing commercial companies, all trying to make money. What’s the relevance of that to Linux distributions?
No relevance at all really.
I wonder how many of these distros could accomplish their goals by existing as a meta-package for whatever distro serves as their upstream.
Quite a few. If the main differences are package selection, configuration, themes and a few extra repositories, a meta-package or setup tool could achieve much the same thing without creating another distribution to maintain.
That’s really the point of my article: a separate distro makes more sense when it is solving a problem that genuinely requires changes below that level.