Ask HN: Is the proliferation of Linux distros harming the Linux ecosystem? | Hacker News Reader

Ask HN: Is the proliferation of Linux distros harming the Linux ecosystem?

26 pointsbehnamoh33 comments

I've used Windows, macOS, and GNU-Linux. Now I use Windows on desktop, but I have this temptation to try Linux again, mainly out of curiosity. But there are tens if not hundreds of Linux "flavors", and a lot of them have so much in common (e.g., Ubuntu is based on Debian), and yet, there are major differences among them, such as how they provide updates (e.g., rolling release) or the hardware that they target.

In a sense, one can't call all of them _one_ OS, whereas you would call all versions of Windows basically one OS that has been consistently updated throughout the years.

I wonder if so many Linux distros is actually Linux's Achilles' heel. IMO, it'd be better if these sparse efforts would get focused on _one_ solid, no-frets, no-hassle, Linux OS that ordinary people could trust to be stable and reliable for years to come.

View on HN

> IMO, it'd be better if these sparse efforts would get focused on _one_ solid, no-frets, no-hassle, Linux OS that ordinary people could trust to be stable and reliable for years to come.

I think that's conceptually a misunderstanding of how open source development works. If five people work on five different distributions, then that's generally because these people have different ideas of how they want things to behave and are motivated to work on realizing these ideas. If you would try to "focus" these "sparse efforts", you wouldn't end up with five people cooperating on one common distribution, you would end up with five people no longer contributing at all since the common system no longe represents their personal motivation for working on it in the first place. (That is if you would be able to magically stop the old distributions)

Now if this is a good or a bad thing depends on your point of view. Is it harming "mass adoption"? Probably. Is "mass adoption" actually a reasonable goal for an open source ecosystem? Probably not. After all non contributing users don't really provide much value to the projects they are using, especially if they still need support. (I'm not trying to say that open source projects are or should be hostile towards users, just that increasing adoption is not necessarily a priority.)

I agree with you, but would like to add another "nuance":

Personally, I entertain the theory that a large number of people agreeing what an OS should be and do also isn't necessarily a good thing, even if it happened. That's how you end up with Windows: It's the greatest common denominator, and for a large group, that's really "low".

For example: I'm very much a fan of how Nix and NixOS deal with the configuration and maintenance of a system. But I'm sure that the Nix way wouldn't have survived a more streamlined distribution landscape. In fact, Nix also works on BSD and OSX, but -- at least to my mind -- NixOS would never have happened there.

And that's understandable. It's still ever-shifting and full of rough edges. It breaks with many established conventions and old "best practices". It's not for everyone. But it still represents one of the greatest paradigm shifts in system configuration I've witnessed so far. In a landscape where you get hammered into the conventional mold, such radical approaches could never blossom and prove their upsides.

Again, look at Windows. There's still no established package manager for non-geeks. For home users, downloading and installing software from all over the internet is still the norm -- and where it isn't, it has been replaced through utilities attempting to lock you in (Microsoft Store). Conformity is a great way to enforce stagnation.

Obviously, things are better in BSD-land, but I'd argue that that's due to the smaller user base, too. Plus there are also at least very specialized BSD distros, too (pfSense, for example).

Sometimes I think people care more about what distribution something is, or is based on, or uses, than really matters. For example: as a distro user, I generally don't care what language something is implemented in. If it's in the distro's repositories; I can use it and it is reasonably well integrated and supported regardless of language.

Similarly, a lot of the time it doesn't really matter what distro something is based on, because I can just consume it in a container or VM and don't need to care.

What does matter are properties such as supportability, the availability and ease of security updates, and so on. So if you ship a container that requires something obscure that I'm not sure security updates will be readily available for, then that matters. But just the fact that it's "not my distro" doesn't really matter.

It's also self defeating to use an obscure distro if you're going to need help and the people you're going to ask for help don't know it.

Each distro works for its user base. The Linux ecosystem is and will remain relatively crap because it isn't commercial. MacOS is great, even with a ton of underlying Open source code because as part of commercialization it does not rely on github issues and pull requests for UX feedback nor does it depend on users asking for features to develop them. Tons of money and time that cannot be paralleled even by commercial Linux that isn't sufficiently well funded. It's not just money and man power either, law of diminishing returns and all. You need the whole pipeline scaled to a competitive level.

Each Linux distro is an OS i n its own right and targets specific audiences. Many have and continue to try and fail to unify and create a distro that is like MacOS or Windows without thinking about what it took for those OS to be consumer ready.

Are you willing to pay hundress of dollars for a distro? If not, then see which distro comes close to meeting your specifc needs and accept it will never have mass adoption. Work to help solve bugs and improve UX in your distro of choice as well.

The incentive to develop Linux distros vary wildly but they are all needs the creators and supporters agree upon. While the incentive behind MacOS and Windows are mass adoption for the sake of profit.

Maybe one day there will be a for profit Linux distro that is will funnel profits to upstream devs and has actual adoption as a goal instead of what current users and devs think is best. Maybe that year is 2022, the year of the Linux Desktop!

Think of open source development the way you think about evolution. It's mostly untargeted, random mutation where 20 Solutions to the same problem pop up and one or two survive in the long run.

It's just random people doing random things and making the same mistakes over and over, it's horribly inefficient but incredibly resilient as a social mechanism. Distributions are an attempt to orchestrate the resulting chaos into something coherent and usable, and since they are open source in nature, they suffer from the very same inefficiencies (and resilience)

> IMO, it'd be better if these sparse efforts would get focused on _one_ solid, no-frets, no-hassle, Linux OS that ordinary people could trust to be stable and reliable for years to come.

You want BSD, not linux.

Why?

The BSD philosophy is to provide a whole cohesive system that has been designed to work together. Linux is just a kernel and components bolted together.

BSD is what I want, but Linux is what I need. (I do not expect this to change any time soon.)

On the other hand, MacOS is a BSD, is it not?

MacOS is an Apple GUI built on top of a BSD foundation. That works out better and cheaper for Apple, as the BSD foundation is open-source and maintained by thousands of developers. Apple merely has to supply a few hundreds of developers to construct and maintain their GUI desktop.

Many of us have suggested that Microsoft also 'do an Apple' and construct the Windows desktop over a BSD or Linux foundation. It would be cheaper for them, and they could concentrate all their developers on producing a really good consistent Windows instead of the rather spotty, uneven productions they have released over the years.

I have heard that Windows kernel is not bad, actually. (I doubt that its architecture or the time MS devs continue to spend on it would have any impact on the quality or consistency of the UI.)

It supports an order of magnitude less hardware than Linux.

Which BSD?

Linux is a different beast, and the big variety of choices you get is a special thing about Linux and not a detriment.

If you're not sure what to use, just use Ubuntu. It's a great default distro.

Ubuntu, to quote, "values your privacy," just like what Windows does. The straight-up Debian, on the other hand, does not have much to say about this, for some reason.

Sort of a nitpick, but: "Distros" don't really exist. Someone chose standard component A, standard component B, standard component C, and put them all on a CD-ROM and called it "Ubuntu". Someone else took A, B, and D, and called it "Debian". Fedora uses A, D, and Q. Distros that use almost the same components and are hard to tell apart (Kubuntu vs KDE Neon).

(So the real question might be "Is the proliferation of standard Linux components harming the Linux ecosystem?")

> IMO, it'd be better if these sparse efforts would get focused on _one_ solid, no-frets, no-hassle, Linux OS that ordinary people could trust to be stable and reliable for years to come

I don't think it's possible to make one Linux version that appeals to everyone. "no-frets, no-hassle" comes with tradeoffs. So does "solid".

But even if you did try to make a beginner-friendly Linux distro, you wouldn't have the support of the majority of the community, because most Linux users are perfectly happy with an existing distro. Even people who try to convert their parents and grandparents to Linux, seem perfectly fine to throw them into Ubuntu and call it a day.

One OS distro (or at least a small number of compatible versions) to rule them all is a major advantage of proprietary systems like Windows/macOS/iOS/etc.. Another advantage is a large staff of paid developers.

The closest equivalent desktop Linux distro in terms of usability, poularity, and support might be Ubuntu, but Ubuntu's design decisions don't necessarily satisfy all use cases and users, and there is nothing preventing them from developing their own Linux-based OS distributions.

For better or for worse, Windows binaries are the de facto standard format for commercial PC games and productivity software. Fortunately many of them can run on Linux via Proton/WINE.

I think the Achilles' heel of OSS are actually the crucial-but-seriously-understaffed libraries, with bus factor [0] of 1 or 2.

For example, openssl [1] is a crucial piece of code with very few full-time developers, especially if you consider the total installed base thereof.

There is, of course, a XKCD for this problem as well: [2]

[0] https://en.wikipedia.org/wiki/Bus_factor

[1] https://en.wikipedia.org/wiki/OpenSSL

[2] https://xkcd.com/2347/

One of the main problems is that Linux is just the kernel, not an operating system (FSF has always been vocal about this with their GNU/Linux branding).

A minimal Linux system requires many GNU utilities and software from other projects to give you a basic command prompt with network connectivity. A Desktop system requires GNOME, KDE, etc.

Contrast this with BSD where a default install gives you a base system. There are BSD distros, but they only focus on orchestration and higher level components, not the base system.

Is the proliferation of car brands harming the motor-vehicle ecosystem?

Surely it would be better to restrict sales to two car brands and only allow two models within those two brands.

(That's a ridiculous suggestion, isn't it? About as ridiculous as restricting the number of Linux distros. As with cars, people want things that suit THEIR needs better, not what everybody else in the world THINKS they should use.)

The beauty of Linux is that distros can be as individual (Ferrari) or as widespread (F150) as the cars on the road are. So some people like Windows (Ford) or MacOS (GM), so what? The choice of Operating systems can be as broad as the choice of cars.

I started with Mandrake and Red Hat, moved to Slackware, then Gentoo, Ubuntu, Mint, and landed on Fedora about 12 years ago. That's my "solid, no-frets, no-hassle, Linux OS". But a lot of people want Ubuntu, Arch, or one of their variants, because they solve some problem that the user perceives, better than one of the other distros.

The variety of choice that we have available is one of the greatest strengths of OSes built around the Linux kernel.

We had a very similar development. In my case it was Mandrake, RedHat, Gentoo, openSUSE and now I oscillate between Fedora and Ubuntu.

It is my experience that while Fedora generally works really well, if something goes wrong, it is seriously wrong and cannot be fixed until the great devs decide to do so. In my case, it was a recurring X11 glitch that returned more than once in several Fedora releases. (I reported it to their bug tracking system.)

On the other hand, Ubuntu has such a massive user base that any problem I encountered already seemed to have a working workaround.

> it'd be better if these sparse efforts would get focused on _one_ solid, no-frets, no-hassle, Linux OS that ordinary people could trust to be stable

The problem is that we in the Linux community can't admit there are problems and then fix them. If there is a problem, chances are the best way to fix it is by forking.

On the other hand, "forking" Debian looks more like sport than anything else...

You are really arguing against a core tenet of Open Source.

One big strength of open source is having source available so others can use and improve and extend it. You can never have two many clones, copies or workalikes.

Now you're suggesting this might be a weakness?

A core tenet of Open Source is that forking is allowed. It's not necessarily the goal that there's a ton of forks that don't try to merge back together eventually.

Maybe not the goal --- but it certainly does nothing to discourage it either.

The kernel is like the ICE of cars. There is so much more than the kernel going on in a distro, that it's amazing there are less mainstream distro's than there are car manufacturers.

Realistically I don't think we could ever force everyone to work on the one right thing so if there were fewer distros it wouldn't make the remaining ones better.

The proliferation of distros _is_ the ecosystem.

Considering the needs of the market are vast and varied it makes sense that there are so many distros.

Anything except Ubuntu or RedHat is a statistical anomaly anyways.

A Linux is a Linux is a Linux... All GNU/Linuxes are same. That is, if you want to do something meaningful with it, it does not really matter which flavor you choose. You can even build a Linux "from scratch" [0] and it will do just fine.

[0] https://www.linuxfromscratch.org

Reply on news.ycombinator.com