You're mostly right, but this is incorrect. MS was the largest kernel contributor for a few days 10 years ago, but overall companies like Intel, Google, Oracle and IBM contribute a lot more.
You can conquer the world, dominate every space and yet there's thus little voice inside saying "but you're not good enough. See over there, in that one space, you're a weakling. "
Welcome to the human condition.
What stung before was that people were forced to use Windows because of lock-ins, software compatibility, etc. That doesn’t really exist any more. If there is one good thing about web apps, it’s that browsers are now the application run time and browsers are more cross platform than any software libraries that have come before it.
The only area I wish had a better story was the enterprise desktop space (like Windows domains or jamf on macOS) but currently I can’t justify allowing my team to run Linux because having its hard enough for small companies to manage a mixture of macOS and Windows fleets without adding a 3rd platform into the mix.
The danger that the history will repeat for browsers instead of OSs is real. The less market share for browsers other than Google Chrome or Chrome-based ones, the more likely it is, that we will be forced to use those because of incompatible web applications.
E.g. at work, using a Linux Computer, I switched to Chrome for Teams and Outlook, because of compatibility issues.
Source: https://techblog.willshouse.com/2012/01/03/most-common-user-...
What? This is very real and a key driver not to use Linux or Mac for millions of users. As soon as even a little performance is required webapps just can't compare. Look at anything graphics related especially 3d, CAD or video. You might be able to build shitty infographics for Instagram on Canva but real work still requires access to compute at the edge and most software providers in those spaces target Microsoft only.
I'd take every day a native app running locally over a web app. And I am a web developer.
Honestly, VSC is pretty decent due to all of the engineering MS put into it. Brackets and Atom both were way slower and had noticeable input lag, at least for me.
Other people have also looked into things like that: https://pavelfatin.com/typing-with-pleasure/
Personally, something like Notepad++ or Lazarus always felt the fastest to me.
Software like JetBrains IDEs, their Fleet editor, Visual Studio and even Visual Studio code all had noticeable hiccups but always seemed "fast enough". Even when the typing itself was okay, there'd be stuff like autocomplete delays/stuttering and refactoring slowdown, but nothing too annoying.
In contrast, Brackets, Atom, working through SSH, remote sessions through RDP/VNC or any device that lagged due to being underpowered always was annoying slow and disruptive.
Also, it runs on a $4 a month namecheap server.
Why is so fast? Because I came from the land of slow crms, and performance was my #1 goal. Instead of making it faster to develope, I made it faster for the end user.
Is that a worthwhile business goal? I don't know. But I know it's fast.
I propose that all the apps you use that you find good you assume are native, but possibly some of those are not.
You are still paying performance for using web apps, and even though your fingers don’t feel it as badly, your battery sure does.
This makes no sense. If webapps run without a loss in performance, then you don't need increased hardware performance in order to see the difference.
> [...] compared to native apps.
This simply isn't true. Run a well-designed webapp and a well-designed native app on the same platform. You'll see a difference in performance.
> you see more and more SPAs wrapped in a hybrid app still maintaining a 'native'look.
I'm still waiting to see even one example of this.
> Costs of developing webapps is dramatically lower.
True. But it comes at a cost to quality.
That's a funny conclusion to make. In the business world that still exists and even my kids are forced into the Microsoft eco-system by their school because the software they insist you run (for which perfectly good web alternatives exist) only runs on windows.
Everything you want to install has to be approved by Microsoft
Enterprises are moving to Workspace from 365. I hope they drive each other's prices to the bottom. :)
While I agree, it has to start somewhere. As a customer of both, I appreciate the competition from 1-2% per year[0] growth.
[0]https://blogs.gartner.com/craig-roth/2022/08/02/market-share...
We tried blowing Office 2013 away on everyone's machines after being told most users didn't think they needed it. Now, a few months later 90% of our users have asked for a 365 license.
What typically happens from what I can tell is that most businesses just end up having a mix of both 365 and Workspace.
Excel specifically seems to have the most attachment among our users.
I'm in that boat, but I don't especially mind. I run Windows because it allows me to run Fusion 360, but Windows Subsystem for Linux gives me a full Linux dev environment. A younger me would have got angry about it, but present me accepts that platform lock-in effects are very strong; fortunately for Linux, it dominates every new category of platform. I sincerely doubt that anyone will ever develop a new proprietary OS from scratch, because the business case for putting a pretty UI on Linux is overwhelming.
I'm not even necessarily blaming this on the performance difference, although that is a pertinent topic, but just the mental structure of something I'm using living in a browser tab rather than actually running on the computer.
For example WebGL 2.0 (2009 hardware), WebGPU (2015 hardware), any GPU capabilities after 2015? You ain't going to get them.
But the student has no choice
I don't think a win for Chromebooks counts as a win for Linux, except in maybe the most narrowly technical sense.
Its honestly quite interesting seeing new users from that perspective.
So in practice, no, it's not very useful. Most students are still forced to use a web browser and nothing more, else they run the risk of getting in trouble for hacking their Chromebook.
Tangent: That being said, it's hilarious how much worse the schools are at locking down Windows computers. As far as I can tell, they just bought a few commercial solutions, set them up on the Chromebooks, and paid no attention to the Windows computers. The BIOS isn't even locked. They're only restricted at the network level.
A loss for the school, but definitely a win for our education
I'm forced to use it at work. :(
One important reality, though, is that "forced" is a sticky word here. Anyone sufficiently anti-FAANG and tech-savvy enough will find a workaround, so the constraints are far less than most dominant forces across the lens of history.
But yes, I'm always surprised how many people buy things introduced to them in ads.
No they don't. There aren't any schools nearby that do not force Windows on the students and I can have my whole house on Linux it doesn't matter: if you want to participate in highschool around here then Windows it is. My personal views of MS, their crimes past and present are worth absolutely nothing. I would not be surprised if you dig deep enough that you'll find some level of corruption but even that doesn't help me, it would just lead to a long and drawn out court case the outcome of which will likely be resolved long after my kids are out of school.
So, I have no choice other than to take my kids out of school and that too - rightly so - isn't an option.
It's more than a little bit ridiculous too because there isn't anything in the highschool curriculum that per-se would require access to a windows computer. But they make it so that it just simply doesn't work otherwise, starting from windows native applications for the agenda and messaging system and ending with sending documents in proprietary formats.
But the lack of driver support still does. Linux on a modern laptop is just full of headaches. I mean, https://www.phoronix.com/review/amd-ryzen7-7840u this is typical: GNOME doesn't work, graphical glitches... In my own experience, operating MFC devices over several years is a nightmare, Bluetooth is problematic as well, some proprietary VPNs do not have Linux clients which might not be Linux's fault but if I can't log in I can't work.
Life is too short for this, sorry. Windows + WSL is the way. The right tool for the right job, Windows for drivers, WSL for, well, everything else.
So, they took a cutting edge Windows PC, slapped Linux on it, and it didn't work well.
How well did OSX do when they tried it? Clearly, OSX isn't ready for the desktop! No wonder nobody runs it!
Modern computers are complex enough that they can work well with Windows, OSX, or Linux, but they don't work well with more than one of them (unless the designers work really hard). See, for example, how to reboot: https://mjg59.livejournal.com/137313.html
If you want to run Linux on a computer, buy it preinstalled, with support. Anything else is an exercise in chasing down small glitches, as you describe. (This happens for Windows too! The difference is that the laptop vendor has a team of people (the system integration team) to fix the issues, either by modifying drivers, modifying the firmware, or having the ODM change the hardware. If you slap Linux on some random bit of Windows kit, you get to be the system integration team, but with only one of three avenues open to you to fix the glitches. And that only halfway, since you cant get the OEM or ODM to give you the time of day. And that's assuming you're a proficient kernel hacker.)
Slapping Linux on a Windows box is a mug's game. The only way to win is not to play.
It's about the hardware's support of the OS (see the firmware discussion above).
It's also about your support experience as a customer (do you have to reinstall Windows before the OEM will give you the time of day, let alone fix the hardware glitch you're hitting)
Many printers, especially the MFC ones, support Linux. For instance, Brother has drivers: https://help.brother-usa.com/app/answers/detail/a_id/52188/~...
WRT corporations, they can do what they like with their own hardware and software. And yes, a lot are short-sighted. Most won't be using Mac any time soon, because they've locked themselves in quite effectively.
it's possible of course, but invariably breaks, especially if you need to install any single third party thing.
i hate flatpak etc but it really makes u think about why people are putting effort into fat packages for desktop stuff
That was the case in 1999 too. The Linux on the Desktop dream wasn't about removing choice, but about MORE people making the choice to run Linux.
Also, market share translates to support, resources available, people and companies working there, drivers, and so on. Personal choice to use X doesn't magically offer those.
But I can't use the OS I like. I can't go in a store, pick a laptop I like, and get it with an OS I like. I can get it with an OS I don't like; and gamble that I can install on it an OS that I do like. It would have been very different if Linux market share on the desktop were in double digits.
That's a strange point to make. The whole point of this users changing their minds, and deciding that they prefer Linux.
> I can’t justify allowing my team to run Linux
So it's not about choice then?
And let's not even get started about work.
And high quality image editing is possible, though it's probably done differently from whatever you're used to. There are plenty of graphic artists making their work on Linux and bragging about it.
So "doesn't even remotely come close" is either a lie or a display of your ignorance.
I use Linux because it works for me and makes my life easier, is free/open and because I don't have to deal with Windows.
But in the most important ways - software availability and reliability - it is much worse.
But lutris as it stands now breaks everything almost every time I use it. I had steam unlaunchable for over a week because of something lutris was doing implicitly. Now running steam directly off the CLI seems to be working for 90% of the games I play (which isn't a very long list).
Only thing I doubt I will be able to figure out is stuff like fortnite or anything that uses easy-anti-cheat (since their devs give a fuck about linux and effectively believe we are all hackers trying to ruin their game).
That's all I got, don't bother with Lutris unless you have an insane amount of time to debug which component is missing.
"Windowsification" is a admittedly vague concept, but tends to cover software that is gratuitously different, not a team player in the bazaar, prone to making land grabs through "embrace and extend", and reluctant to offer configuration options that conflict with its "opinionated" stances.
As an aside, I'm really sick of the systemd / "random pile of shell scripts" false dichotomy. It's trivial to turn a python script into a system service with runit, too. Simple, stable, and focused.
All the UX work is being done in the steam client itself and is not exclusive to the Steam Deck. In fact, the SteamOS ui is now available 1:1 in the good old PC Steam client, replacing the old Big Picture mode.
The only thing that doesn’t comes directly with Steam is Gamescope (but you can install it yourself easily).
Other than that, the SteamOS is just an Arch Linux with Steam installed, KDE and a Steam wallpaper.
https://steamcommunity.com/groups/SteamClientBeta/discussion...
I'm trying really hard to come up with a scenario where optimizing the desktop experience for the average user is a bad thing.
For example, what keeps me on Windows (and WSL2) at the moment is that the freesync/variable refresh rate technology doesn't always behave as you would expect it to be, resulting in games that stutter, frame dip and eat inputs. Gnome devs have a branch pending final tests for vrr for months. KDE on Wayland is the best for what I'm asking atm but even that has its problems during gameplay.
And let's not forget the pains Nvidia's drivers cause compared to AMD on Linux for desktop. I built my desktop using AMD GPU and CPU specifically for Linux to play nice, and yet it doesn't for the only 2 games I play during my downtime. It's just sad.
Again, maybe I'm a sore loser, but it definitely feels like a lot of software is becoming increasingly hostile to me, personally because it's being optimized for a presumed average user. I have no beef with this average user and it's globally a totally correct optimization to make stuff work for them as well as possible, but it means that software makes tradeoffs that are inconvenient for me personally.
The most common use case is streamlined by removing options that clutter up the settings menu or banishing the buttons I press a lot into some tertiary menu, or niche features I rely on become unsupported and then get removed (of course with forced auto-updates, because the average user experience is improved by them) to free up developer bandwidth for the core features. And then I end up using a different off-brand window manager with the serial numbers filed off that barely works in general, but works for my specific 'workflow" exactly like I want it.
Of course, at the same time I also benefit from general improvements targeting the average user. It's only about the tradeoffs.
It's a bad thing if you're not an average user, because software aimed at the "average user" tends to be a pain in the butt for the non-average user. I would point to Windows, the entire web browser space, etc., as examples of this.
And it's not just about the average user. It's also about the compromises needed to make a product suitable for mass production. I already see the effects of that on Linux, in the form of things like systemd, snaps, flatpack, etc.
They currently have around 20% market share (as Internet servers - much higher inside business LANs), which is a very nice chunk. And I'm sure encourages them to keep working on it.
Frankly the tooling for MS server keeps it in the hunt - they really understand the concept of making things easier to use.
Extremely big doubts about this. How would anyone even know? All the hyperscale datacenters built by FAANG must be 99.9% Linux, and those DCs alone would probably be enough to make 20% impossible to hit for windows.
(Scroll down to servers)
Incidentally this is public facing Internet servers. The % -inside- enterprises us much much higher.
Tools like Active Directory allow you to scale enterprise operations, which is important for companies that have hundreds-of-thousands of desktops etc.
In any event your estimate of 99.9% is unlikely. AWS offers Windows based EC2 instances, and one presumes they're not doing that for 0.01% of the market. Azure will likely have an even higher %.
From the data though, not just guessing, I'd say it's about 20% overall.
I can believe private enterprise usage is higher, but I imagine FAANG and similar companies also have quite a lot non public Linux servers. Not trying to rebut your overall point, but I'm not sure the total non public % is much different.
We know that all the top 25 websites are 100% Linux and represent a lot of servers each.
Also a whole lot of that cited report seems to be misidentified. The top Windows site other than Microsoft properties listed is nih.gov, which looks like Drupal on Linux behind Akamai GHost -- all Linux.
I do accept that Windows is probably about 20-25% of server instances and slowly contracting. But public facing webservers and the infrastructure that directly supports them? Nothing close to this.
What stings me a bit is that Microsoft is the dominant player on the desktop and then they are so little about making the making it a good and consistent experience. Windows 95 was never amazing, but they put the work in and gave us the impression that they at least tried. Now it's 2023 and my desktop has ads, tracking and UI elements that are 30 years old, because they could never be arsed to update them.
No, not really.
You can't make things smooth and flexible at the same time, it takes too much effort. You can't support a ton of varied hardware and make everything work smoothly in every case; it again would take non-viable amounts of work.
You can have a pretty smooth desktop Linux experience if you pick a large mainstream distro, pick reliable, well-supported mainstream hardware, and pick a polished desktop environment such as KDE or that Gnome variant of Pop OS.
But if you want to tweak things, you've got to tweak things.
It feels a little bonkers that in 2023, Flatpak or something like it isn’t the default way software is installed. That calculator you just installed should not have access to your camera or microphone or file system. That weather app may need your location, but you should be able to grant access to sensitive resources always, never, or only after asking.
Or maybe Fuchsia, when it's ready for desktop.
Depending on how you define "desktop" i would argue that Linux has a very big presence on the desktop. Considering the "new desktop" is a smartphone, and Android is based on Linux, that alone would account for a very large desktop presence.
Unfortunately there is no real “transmission” between the two, pretty much only the kernel is shared, absolutely nothing in userspace. Hell, it is ridiculously difficult to even run/emulate android on a normal linux distro.
Which is a shame, because while android’s model is far from perfect, it is definitely eons ahead of the traditional linux userspace, especially on the security front.
Despite the lack of any exponential moonshots in marketshare, compatibility layers like Wine and Wine derivatives, as well as platform agnostic package management tools, have made Linux desktops much more pleasant to use compared to just 4-5 years ago. This has allowed me to almost completely de-Windows my computer usage without abandoning Windows-only applications.
I do try to shill Linux to friends from time to time as a casual tease. Haven't had much success in that regard but it's not a concrete goal of mine. Despite having all gone to the same STEM program that required Linux environments for most courses, I'm the only one using it as a daily driver outside of work among the 15-20 people I still regularly contact.
Not even a slight itch.
What do I care how many people use the thing I do? The only thing that matters to me is how good the toys I get to use are.
> See over there, in that one space, you're a weakling.
A lot more people eat fast food that gourmet food. I wonder if the people who can afford gourmet food feel like weaklings because of that. My guess is: No they don't.
Perception matters :)
There is a lot we can learn from a creature whose life purpose is eating shit all day, every day.
No. In fact, as a Linux user, I'm happy that Linux doesn't have a major share of the desktop. If it did, then I think it would have a terrible (to me) effect on the direction of Linux development, because it would be more tempting to aim for the most common denominator, and we already have good options for the most common denominator.
NOO IT DOESN'T!!
Love the can of worms you opened with that comment. It does sting, because the desktop market is optimized for proprietary software. Open source, even if superior, is always behind on a hill that is upwards both ways. Proprietary firmware, drivers, patents, vendor lock-in, etc., have become very high walls.
For the average person, they don't care what a window manager is or pulse/alsa/jack or whatever. They just want a singular consistent experience that "just works", whatever software they install, whatever hardware they plug in. And the Linux community is obsessed with the complete opposite of this.
Apparently a lot of the systemd pushback was because it was making everything "too similar and compatible with each other". Now isn't that just laughable; that's what most people want!
When explaining linux to skeptical non-techies, I tell them it's the OS running the nuclear submarines right now. Suddenly they're more confident it can handle their ecommerce dashboard.
For US nuclear subs: All of the core components and electronics use a custom RTOS; older ones built in Ada, newer ones built in C++. The "control computers" (the computers that humans interface with to tell core components what to do) aren't running anything like Ubuntu or Fedora, but instead run a heavily ruggedized and customized Linux system based on the Titanium suite from Star Lab.
For UK subs, they run a similar custom OS for the modules and electronics and (until 2017 or so, at least) run Windows XP for their control computers.
I have no idea what other nations use; but I would assume they are similarly setup.
That's not to disavow the fact that clearly the US govt trusts those Linux computers to be reliable, thus their investment in them. So it is an accomplishment, they just don't "run" anything.
That’s not to be minimized. Sounds like very much on the critical path, and definitely not some shitty infotainment sidecar. If that crashed, you couldn’t control core functions of the sub? Are there overrides?
Btw, thanks for sharing these details. I would never have learnt this. Love HN sometimes.
Sure, it's an important component of the entire system. Much as a steering wheel is an important component of a car, it still doesn't "run" the car. You need to interface, yes obviously; but if the machine crashes the sub operates merrily along like usual and any of the other control computers can continue to give it commands.
As to overrides, if you lost every single control computer and the direct interfaces (the analog control boards you see in crucial areas of naval vessels) then the overall system has a fallback procedure. It's not documented or public knowledge, but assumedly it would surface and activate an emergency beacon as the vessel is crippled (and likely damaged) at that point.
However, desktop Linux is where the product is 100% OSS. That’s the apples to apples comparison. And that’s where it can’t compete without vendors writing property code to put in front of it to make it actually work well.
Server side Linux gets a lot of support from professional programmers paid by big corps. They have improved it beyond hobbyist level, and it works pretty well. And of course cheap makes up for a lot.
Linux has the largest collection of various hardware drivers, and a rather wide community of kernel developers who know how to produce them.
Linux has a number of specialty filesystems, some pretty important for embedded development.
Linux kernel has real-time extensions, which is not entirely a commodity, AFAICT.
Also, stuff like io_uring, BPF VM inside the kernel, namespaces (see "containers"), etc.
No, I don't think you can easily replace Linux kernel with whatever BSD kernel, Darwin, VxWorks, Contiki, FreeRTOS, etc, or, well, Windows. Much of the larger embedded stuff is Linux-specific, and much of big server stuff is either Linux-only or has extensive specific support.
> Various versions of NT family operating systems have been released for a variety of processor architectures, initially IA-32, MIPS, and DEC Alpha, with PowerPC, Itanium, x86-64 and ARM supported in later releases.
> In order to prevent Intel x86-specific code from slipping into the operating system, due to developers being used to developing on x86 chips, Windows NT 3.1 was initially developed using non-x86 development systems and then ported to the x86 architecture. This work was initially based on the Intel i860-based Dazzle system and, later, the MIPS R4000-based Jazz platform. Both systems were designed internally at Microsoft.
Tell that to your BIOS or the firmware that runs on your WLAN or NVMe. Hell, tell that to the CPU itself[^1].
[^1]: https://en.m.wikipedia.org/wiki/Intel_Management_Engine
Sour grapes.
> Linux is the go to choice when you could just as well be using something else that’s really cheap, like BSD.
You actually couldn't be just as well using BSD. Corporations would love to be able to take Linux and not contribute their changes back (and that's frequently what happens when they use it internally but don't distribute it). They don't use a BSD licensed kernel in many such places (e.g., Android) because Linux is more advanced.
More advanced in what sense? There are lots of devices using BSD, so that's entirely possible.
Taking one off the top of my head that was relevant around the time Android was being developed, dynamic ticks which is important for low power operation. Linux supported this in 2007 about 5 years before FreeBSD and I think NetBSD still doesn't support it. Could not field a phone OS without that, at least not back in those days.
Here's a report about server market share:
https://w3techs.com/technologies/details/os-linux
Here's one for total device market share (quite a bit out of date, since Gartner paywalls all new reports; still it shows the general trend):
https://www.computerworld.com/article/3050931/windows-comes-...
Etc. All of which agree with you. But, it turns out, most people don't care about what OS runs their toothbrush or thin client. Only their mobile device or desktop, which is why these reports are so much more popular.
Yeah. But it's not its OS. The true OS of most Android installations is some propietary firmware blob.
Huh? This is...totally wrong. Software devs at Microsoft are supplied with Windows machines by default, a lot of software is built on Microsoft-specific technologies like C#, and most of the server footprint is Windows Server.
My team only uses MacBooks at Microsoft. It’s true that I was forced to receive a PC, but it sits in a drawer collecting dust and only gets booted up when compliance throws a tantrum.
Culture has changed, nobody cares if you use a Mac. Hell I bet half the employees are taking Teams meetings on their iPhones.
For anyone who cares about these stats using Linux is the most important thing about computing.
You can't get an app designed for Android to run on Linux Mint with ease.
The kernel is not the operating system, even if it is a key component of it. Android phones do use the Linux kernel - well, not exactly; some kind of modified version of it - but they do not use the GNU/Linux operating system. The OS is different in many ways: Filesystem hierarchy, daemons/services, graphics stack, etc.
Depends on what and why Linux. Original dream/vision selling it was a libre OS developed in the Bazaar that will give user full control over their devices against the tyranny of the corporations. You can't with straight face claim that all the billions of phones and embedded devices represent that freedom.
As about kernel, it may be BSD, Linux or maybe even NT or Plan 9, it honestly doesn't matter.
Is there a source for that?
Yes, but that's kind of a moot point.
The decades old dream was "Linux on the Desktop" (not just as some 3% market share or "works for me" - it worked for people in 1999 too), but overtaking Windows as the desktop OS for the masses).
And of course all those billion of Android devices, hide everything about what people meant and wanted from the Linux experience. It's just the kernel and some basic userland stuff, on top of which everything else is totally foreign, closed down, and tied to proprietary services under Google's control (with Samsung and co's own touches).
Linux is so irrelevant to the visible part of Android's functionality, that Google could very well change the backend to Fuchsia when that's ready, and those billion devices would be un-Linuxed within a few years, as they get old, and their owners replace them by new Fuchsia running smartphones.
It's interesting to see how over the years the user experience of Linux Desktop ia getting better, bug are being fixed, we get better software, desktop environments are getting more features e.g.
Citation needed, I think. Microsoft barely makes or doesn't even make the top 20 kernel contributors for current releases: 6.4 (#20): https://lwn.net/Articles/936113/, 6.3: https://lwn.net/Articles/929582/, 6.2: https://lwn.net/Articles/923410/, 6.1 (#19): https://lwn.net/Articles/915435/
> 20XX will be the year of Linux on the desktop
They've been saying it so long it's part of our psyche. It doesn't matter the other places Linux is running, it hasn't conquered that one place they've been trying to conquer since the start. You never get over the one who got away.
When they said it back in the 1920s it was sort of confusing, tbh.
Who is "they"?
There's are a heck of a lot of people happily building [software for] Linux systems that give precisely zero care to the success of Linux desktop.
Who is "they" here?
After so many years, they still don't have reliable "Every package even the ones not in the repos just works" status. It's mostly there, like, 99.9%, but stuff still breaks more than on Windows or Android. I suspect they never will, except for things like NixOS.
I used to think Linux was behind because of legacy reasons and game companies not being interested... but now I think Linux itself has something to do with it, and we should use that as motivation to improve.
Right now I think NixOS, or just accepting Snap packages, or waiting for Flatpak, might be the best/only chance, aside from moving to a batteries included OS like Android.
Seems like Nix really just needs GUI admin tools, everything else seems to be fairly polished.
Steam and Lutris (StarCraft 2!) work pretty well, Detroit Become Human just works. (Enable proton in Steam, and click install and play.)
(I removed snap, however, as it's just in the way, but I don't remember what was my exact problem with it.)
Linux is on more computers both today and all of the computers ever made put together than any other OS. It is on virtually all servers on the Internet, the majority OS on phones, its in many TVs and STBs and streaming sticks, its one of the few OSes ever in space, its one of the few OSes ever on Mars, and it is also the OS behind ChromeOS (which that article mistakenly breaks out as its own numbers; so by their own admission, its at least 7%).
The largest desktop OS on that list? Windows.... made by a company who makes more money off Linux than anything else, is the largest corp contributor to Linux, and has far more Linux machines internally than Windows ones, and hires software developers who do not develop for Windows.