What desktop Linux needs to succeed in the mainstream

You might be aware of the recent Linus Tech Tips videos about switching to Linux, including one with some complaints about KDE software. For those of you who are following along, I want to let you know that we’re working on fixing the issues Linus brought up, and you can track our progress here. Thankfully most of the issues are fairly minor and should be easy to fix.

This blog post is my version of Sway developer Drew DeVault’s post about the videos, regarding the question of what desktop Linux needs to go mainstream. Drew emphasizes accessibility, and I agree, but with a slightly different conclusion:

Desktop Linux needs to be pre-installed on retail hardware to succeed in the mainstream.

That’s it.

Allow me to explain.


People get hung up a lot on features and usability, and these are important. But they’re means to an end and not good enough ends by themselves. Quality means nothing if people can’t get it. And people can’t get it without accessible distribution. High quality Linux distros aren’t enough; they need to be pre-installed on hardware products you can buy in mainstream retail stores! “The mainstream” buys products they can touch and hold; if you can’t find it in a mainstream store, it doesn’t exist.

Think about it: why do normal people use Windows or macOS? Because the physical computer they bought included it. iOS or Android? Because it was shipped by default on their physical smartphone. The notion of replacing a device’s operating system with a new one doesn’t exist to “the mainstream”. Only the “three-dot” users ever do that, and they’re about 5% of the market. If the only way to get your OS is to install it yourself, you have no chance of succeeding in the mainstream.

As for features, people generally use only a very small fraction of what’s available to them. When it comes to usability, most users memorize their software rather than understanding it–and you can memorize anything if you really have to. A better user interface helps, but it isn’t needed for the memorizers and mostly benefits power users (the 30% of the market “two-dot and up” crowd) who recognize patterns and appreciate logic, consistency, and good design. So these are not good enough on their own.

This doesn’t mean we should forget about features and usability! Not at all! But if the goal is to “go mainstream,”we have to understand the true audience: hardware vendors, not end users. The goal is to have a software product appealing enough to get picked up by vendors when they go shopping for one, because that’s mostly how it works. Companies like Apple that do their own custom top-to-bottom hardware and software for big-name products are rare. Most build on top of 3rd-party software that requires the least integration and custom work from their in-house software team. If your software isn’t up to the task, they move onto the next option. So when some hardware vendor has a need, your software better be ready!

And what do hardware vendors need?

  • Flexibility. Your software has to be easily adaptable to whatever kind of device they have without tons of custom engineering they’ll be on the hook for supporting over the product’s lifecycle.
  • Features that make their devices look good. Support for its physical hardware characteristics, good performance, a pleasant-looking user interface… reasons for people to buy it, basically.
  • Stability. Can’t crash and dump users at a command line terminal prompt. Has to actually work. Can’t feel like a hobbyist science fair project.
  • Usability that’s to be good enough to minimize support costs. When something goes wrong, “the mainstream” contacts their hardware vendor. Usability needs to be good enough so that this happens as infrequently as possible.

It doesn’t need to be perfect. It just needs to do that stuff. This is how Windows conquered the PC market in the 90s despite being terrible! And our stuff is much better!


I see evidence that this is already working for KDE. Pine ships Manjaro with Plasma Mobile and Plasma Desktop on the PinePhone and PineBook Pro, respectively. Valve also picked Plasma Desktop for the Steam Deck, replacing GNOME for their new version of SteamOS. I see KDE software as well-positioned here and getting better all the time. So let’s keep doubling down on delivering what hardware vendors need to sell their awesome products.

57 thoughts on “What desktop Linux needs to succeed in the mainstream

  1. > The notion of replacing a device’s operating system with a new one doesn’t exist to “the mainstream”. Only the “three-dot” users ever do that, and they’re about 5% of the market

    True, but that 5% is still a lot of people. If we can manage to reel those in, hardware vendors will be more likely to want to support Linux. We need to give the hardware vendors a reason to support Linux

    Liked by 1 person

    1. It’s true! And I would be thrilled if we could get all of them. Advanced users are generally influencers. Their friends and family members turn to them for help and ask them for recommendations. If they all use KDE Plasma, that’s probably going to double or triple KDE’s market share simply due to their passion and recommendations getting listened to and acted on.

      That’s why I’m very serious about never going back on “powerful when needed”. Plasma always has to remain useful for real work, complicated workflows, serious professional use cases, and extensive customization so that it fits like a glove. I feel quite strongly that it’s possible to keep and expand these while also making the default UX a bit simpler and friendlier. Hence, “simple by default, powerful when needed.” 🙂

      Liked by 1 person

  2. I couldn’t agree more! I’ve made a hobby out of taking old Windows 7-8 era laptops, fixing them up, putting Linux Mint on them, and giving them away to people who need them. I’ve given away well over 100 computers at this point.

    I followed up with all the people after the fact to see how they got along with it. Literally everyone loved it. The ONLY machine that came back, was because the users brother, decided to put windows on it, and she hated it. She wanted “Windows Mint” back on there (lol)

    Some of these have been out there for over a year too and I still have people dropping me a message saying, “still have this computer and runs great, thanks again”

    I can’t even imagine what would happen if i gave away 100 windows machines.. seriously

    Liked by 1 person

    1. Linux Mint is great for this, yeah. Before I was a KDE guy, I used Mint myself for a few years and put it on the computers of local friends and neighbors who turned to me for help and were fed up with (at the time) Windows 8. It was the same, I never got a callback about it. It just worked–until they tried to upgrade, but that’s another story.

      I’m still jealous of that level of success and I think we’re not there yet in KDE. We don’t have a distro that’s as good, and our default UX is still a lot buggier than theirs was and is. But ultimately I think we can get there. Our tech is better, we aren’t dependent on the software of a direct competitor, we have more contributors, and we can appeal to a wider range of users.

      Like

  3. I agree desktop Linux needs to be pre-installed on retail hardware to succeed in the mainstream. Problem is, even with pre-installed on retail hardware plus marketing, windows phone is not succeed.

    Like

  4. Hardware vendors and usability come hand in hand. Vendors have money and incentive to improve all the paper cuts that three dot users tend to overlook as they seem minor.
    From Your knowledge, do Valve engineers work on upstream Plasma and KDE stack?

    Liked by 1 person

    1. Yes, a bit, and Valve is also paying for others to work on KDE software–myself included, indirectly, through my employer.

      This is another reason why I think hardware vendors are important: they have money and are willing to pay for development that benefits them! And if they’re using your software, the development work they pay for benefits you too, and anyone else in the future who wants to use it.

      Like

  5. I also believe that Linux needs to keep up with providing alternatives to the newer features available on MS Windows.

    For instance, Win11 now provides a way to run Android apps. On Linux, this is still in its early days and usually just doesn’t work. Until recently, there was one I couldn’t convince my friends to make the switch: the absence of Adobe and MS Office. Now there is going to be two: my friends are starting to rely on their Android apps on desktop.
    The support for apks could also become a game changer for the future of Linux phones, which, I am convinced, are going to get popular with people looking for privacy.

    Another example is to cast your screen over Wi-Fi. My enterprise removed the HDMI cables and now I cannot show my screen any more so I have to ask for someone’s laptop each time I have a presentation. So I get pressured to use Windows again.

    Like

      1. I advise to try android-x86.org. It’s not perfect. but i can use WhatsApp with it.

        https://www.android-x86.org/releases/releasenote-7-1-r5.html

        The 7.1-r5 is the optimum. the newer 8.x gets a bad interface.

        Here is my QEMU invocation.

        #!/bin/bash

        qemu-kvm -enable-kvm \
        -m 2048 -smp 2 -cpu host \
        -device virtio-vga,virgl=on -display gtk,gl=on,zoom-to-fit=off \
        -device usb-ehci \
        -device usb-kbd \
        -device usb-mouse \
        -device usb-tablet \
        -device ich9-intel-hda \
        -device hda-duplex,audiodev=snd0 \
        -audiodev pa,id=snd0 \
        -device usb-host,vendorid=0x046d,productid=0x081b \
        -boot menu=on \
        -nic bridge \
        ~/QEMU_VM/android_x86_7.1-r5.img \

        and a help

        https://forums.opensuse.org/showthread.php/551824-qemu-kvm-user-management-create-launch-create-a-bridge-use-a-bridge

        You must use a bridge to connect your OS to internet.

        Like

      2. That’s a sort of irony. Android runs on linux kernel. So if on any platform android apps can run without virtualization (I’m talking about kernel not userland) it is linux. On windows subsystem it is still virtualized just like wsl2.

        Again wsl2 is seen as something that linux can’t do. Wrong, xen is there for longer and is far more stable and mature and big industries rely on it. But there is little way to modify and integrate proprietary os’s with desktop so linux subsystem of windows is not possible without modifying substantial amount of windows.

        Liked by 1 person

      3. @Nate could you think/explain why oem vendors install windows or chromeos by default? Is it the profit Microsoft/Google gives them? Is it the legacy? Is it that there are companies behind so it seems more reliable and there is someone to blame in case of issues? Like linux dustros usually come without warranty and support contract and the project might get discontinued anytime.

        Notice that devices that come with linux usually come with a distro like Ubuntu, pop_os or manjaro. Each have a company behind.

        PS: by linux usually I mean here GNU/Linux free-desktop os. Not something else like android.

        Like

        1. Mostly because it’s the simple easy conservative choice. “Nobody ever got fired for buying IBM” etc. Microsoft and Google are known quantities. Moving to anything different is always a risk.

          This is why I think you’re seeing Plasma being used on new devices more so than replacing the OS on old ones. When you make new hardware and your OS choices are wide open, then every choice represents a risk, so you might as well choose the best one. 😎

          Like

  6. I agree with you but
    Driver, driver !
    openSUSE 15.3
    Even with the last driver from Epson repo i can’t use the scanner of my Epson XP-7100.

    Like

  7. At this moment I feel that only Zorin OS 16 Core and KDE Neon could have a small chance of making it to pre-installed level. And Neon is a MAYBE, it needs to not f**k up. Zorin is just a rock and it murders Ubuntu, Mint etc etc EASY from a new potential lInux user point of view. Easy and functional. KDE Plasma is not easy, but it can be..if many defaults would be “new user” friendly.. but yeah, Nate speaks the truth here. Please don’t start listing all your favorite distros that you like better than Zorin or Neon, this is not about you. It’s about the supernoob that is totally clueless about what goes on inside of computers, and how we reel those people in.

    Like

      1. Single click to open, for starters =) Small things for developers and heavy users, but huge dealbreakers for absolute beginners. If people don’t like it from the start, they WILL hate it and that newfound hatred gets etched in their brains and spreads like wildfire in every computer-based conversation they have from then on. Some kid overhears dad slamming linux and tells their friends and in school then nobody will use linux, to put it simply. Humans are simple, but powerful when needed haha.. First there needs to be a good distro to spread KDE. That distros maintainers need to work with KDE developers, it affects everything. You can have the greatest most wonderful and stable KDE Plasma and then the distro maintainers ship it with some flamboyant off-putting horror wallpaper (yes Picasso is art but not a wallpaper) along with a default colorscheme straight from the murkiest discos from 1979 and tops it off by slapping on 258 programs THEY think everyone needs. The brave new user that doesn’t run away screaming upon first sight of this marvel WILL break their system removing 90% pre-installed apps that are bloatware but the hardcore distromaintainer says otherwise.. I could go on and on, but this is all just coming from a new “clean, empty minded” possible linux user. It’s very hard to picture it when you’ve used linux for years, and the motives for using it are different than yours. I don’t remember which distro it was but it came with a small app/script with a window where the user could choose from a bunch of browsers, and that window was also an uninstaller for the browser should they want to switch. I thought that approach was brilliant, could be useful in some other areas as well, office suite, mediaplayer, musicplayer.. and those nice introduction tours that many distros now come with are a very good place to start. “Reel them in, don’t scare them off” should be a mantra. Seduce them with your awesomeness! haha oh man.. But yeah, small annoyances they do gather up to a big ball of nope pretty fast when you do not have the interest or time or any desire to tinker around like crazy just to get a nice OS for your computer. A computer is a tool, not the user. I installed Zorin OS 16 for a friend to try out, added Google Chrome (yes, because that’s what they use) and a couple other programs I know he needs so it was a painless starting point. Told him the password for updating, oh and I set the updates to not harass him all the time, once every two weeks instead of every day..and I never heard from him after that.. maybe two months later I sent him a message “so..how did that linux experiment go..?” He answered: everything works great here, no problems. His wife uses that computer too now. So yes, I would say that’s a SUCCESS! I’d like to thank the academy..nah, the Zorin brothers are the ones to thank here, they’re doing Gods work.. Now, I wish I had a KDE distro to spread like that, but as of now that’s not gonna happen. I’m not putting a “stable” ancient distro with old plasma on anything, that’s just asking for trouble..and then there’s that dad going “KDE Plasma, NOTHING works. This is the dumbest garbage I’ve ever used!!” ..little Johnny overheard that and it’s gonna be a fun day at school tomorrow..

        Like

  8. Totally agree, it needs to be pre-installed on the hardware that people buy, but there’s an elephant in the room here! What happens when the release version reaches end of life? Now we would be asking non-technical people (Nate’s sub-4dot users) to reinstall a new system if they want the latest features. Pardon me if I’ve missed something here, I haven’t tried many other distros except for Fedora, which I have used for almost twenty years (and exclusively for over 10 years), but only in the last few releases has it been possible to install the latest release without having to download/create boot disc/reinstall/cross fingers. This process needs to become much smoother and reliable before mainstream folk will be comfortable with it.

    Like

    1. Discover has the ability to upgrade from one major release to another in several distros. I know Kubuntu and openSUSE Leap are supported. Fedora is not, which is personally my last blocker fo recommending Fedora to everyone.

      Like

    2. Two solutions. One: what Nate wrote. Discover handling major system upgrades.
      Two: a rolling release distro like Manjaro or openSUSE Tumbleweed.

      Btw, people need to upgrade Windows too. In 2020 I would have used past tense since Windows 10 is “teh last Windows ever”, but we have Windows 11 now.

      Like

    3. That’s the reason why I use OpenSUSE Tumbleweed – a stable(!) rolling release. Using OPI most of the stuff I need just works, but it would be great if OPI’s functionality was included in zypper and the graphical package manager. Some stuff would need some clean up, some GUIs need some love to make make it viable for newbies or people that just want to use their computer and not tinker with the OS.

      Like

  9. Hi Nate Graham.

    My name is Joshua (Josué) and I am the developer of the Brazilian Linux distribution Regata OS, which comes with KDE Plasma by default. I was very happy to hear that you have been adopting feedback from Linus Sebastian of Linus Tech Tips.

    We have been working to convince local hardware manufacturers to adopt the Regata OS and offer laptops and desktops with our operating system in Brazilian hardware retail.

    I am a huge fan of your work. Great article!

    Liked by 1 person

    1. That is so cool! You’re doing it!

      What are the sorts of things your vendors have been asking for? What are their big concerns and blockers, if you don’t mind my asking?

      Liked by 1 person

      1. Luckily, local hardware manufacturers have been offering some Linux laptops and desktops for some time now with the aim of reducing costs, as the consumer doesn’t have to pay for a Windows license as well. But, unfortunately, the Linux offered by these hardware manufacturers doesn’t offer a good experience and isn’t focused on the home user (it’s nothing we know like, say, Ubuntu, KDE neon or Fedora). So, we are trying to improve this scenario.

        To make Regata OS more attractive, we made our own app store, to try to offer an experience similar to what we found on Android. In addition, we also made a settings app, where the user can more easily choose which apps can run with the laptop’s dGPU.

        On the KDE Plasma side, we make it easy for users to restore the “factory defaults” of the graphical environment, for when something goes wrong when the user tries to customize the graphical environment, removing the panel and changing the themes, for example. And more recently we decided to lock widgets by default, but with an option that is easily accessible to unlock widgets.

        There are many other changes we have made and we are now trying to convince local manufacturers that the Regata OS can be a good choice for home users. 😀

        Like

  10. If you consider people in general, especially in more ‘liberal’ countries like us, most Europe and south asia, people do not care about privacy. They dump their daily life in all sorts of product that doesn’t value privacy at all.

    If your selling point is just privacy you’re out of luck.

    Like

  11. Agreed again! I did a lot of tech support a few years ago, and it was obvious that users, no matter if they used Windows or MacOS, encountered loads of usability problems and papercuts, big and small – but they pretty much just considered that a part of using a computer. In other words, while a polished UI is certainly important, the biggest issue is making it easy for people to get their hands on a computer with Linux on. Promoting, spreading and suggesting such devices is important for us as a community in order to spread the use of Linux, KDE and other free and open source software.

    Liked by 1 person

  12. Linux will never succeed in the mainstream and there is nothing wrong with that.
    Businesses who outnumber individual users by 3 to 1 margin will never accept Linux because there is not support for it and it is too technical for average user.

    Like

  13. Yes, all this ‘no can do anything with any file outside home folder’ Dolphin bullshit is flat out embarrassing. It has been like that for years now and there is no excuse for that.

    Like

  14. But let’s talk about the normal people who use it:

    Joe inserts a USB device . No sound .( Is the device ok?). Where can I see that is working?

    Joe has a new USB storage device. But, he needs to format it, compatible with Windows, to give his friend music.. right click on the USB and … Oh wait… Never mind.

    Surely, Joe, after downloading a game without Steam, can easily uninstall it from the GUI, right?….

    And may other problems.

    Sure, Linux is not Windows, but doing something easy would attract users….

    Like

    1. A word of advice: Those seem like valid things to fix, but the mocking tone probably makes people less likely to want to fix them for you, not more. You catch more flies with honey than vinegar, as the saying goes. 🙂

      Like

      1. One Linux problem is Linux community attitude. Loren is 100% right and you are Linux guy with attitude. Average guy wants simple thing, mouse and GUI not command line. Linux community mostly just won’t accept things that are simple, logical and accepted but support chaos in the name of choice. My example is here how applications are numbered . There are alpha beta and final version of the application where 1.0 and above means application is ready for commercial use and that is everywhere except in Linux where programmer just dumps any number he wants so you just don’t have no idea in what stage of development application is. And about lot of other things it is more important to support chaos than to go one established and accepted way all other OSs use.

        Like

        1. There are alpha beta and final version of the application where 1.0 and above means application is ready for commercial use and that is everywhere except in Linux where programmer just dumps any number he wants so you just don’t have no idea in what stage of development application is.

          Hmm, what were the version numbers of MS Windows over time?

          1.0
          2.0
          3.1
          95
          98
          2000
          ME
          XP
          Vista
          7
          8
          10
          11

          Clearly this sort of nonsense isn’t a real issue for normal people. 🙂

          Like

          1. There is not any nonsense – normal people know MS follows either consecutive order or year number – exceptions were me, xp and vista and that was to respond to Apple version names. Vista was just enhanced xp – they got it out as separate version only to charge more. They skipped 9 to get to Mac OS X (10) so windows version does not look like they are behind Apple.
            Wish Linux developers use as much logic as Apple and MS.
            I am Debian user but Linux developers have to start listening to users as much as Apple and Microsoft do.

            Like

          2. @Nate
            You just don’t get it. Linux guys like you are reason why Linux will never go over single digit market share

            Like

          3. @Nate
            Are you really that stupid that you can’t understand what I wanted to say ?
            Are you really that stupid that you can’t understand what Loren and Marvin want to tell you ?
            No wonder Linux can’t go above 7% market share when such ignorant guys like you are main developers.
            I run my own business and I have to listen to my customers and keep them happy or I won’t be in business for long. I run it on Windows because ignorant Linux developers like you tell : Why do you complain when I give you my product for free and instead you should ask : Why people don’t want to use our product even when we give it for free?
            Microsoft and Apple who listen to their customers are among richest companies on this planet and their employees are in higher middle class and rich and you stay at home with your mom and be happy when you are such an ignorant asshole.
            I run Debian at home for some of my hobbies so I do know what I am talking about.

            Like

          4. This blog is my space, and you need to be polite to post here. You just lost that privilege. Dissenting views are fine, but insults aren’t.

            Liked by 1 person

      2. This is a major reason why Linux will never go mainstream.

        Linux developers, building for Linux community and both holding an elitist attitude.

        Mainstream users aren’t technical. They don’t want the technology to be endlessly customisable. They don’t want to have to tinker to get it to work.

        They just want to work and consume.

        If you want more distribution to the masses, the interface should perform the same, familiar functions as Windows or Mac – THEN add customisations that add to the experience.

        Simple examples-
        Zorin is getting there with their layouts, but failed on simple things like only having a dozen backgrounds and no way to set a plain, solid color background.

        Mouse wheel scrolling seems broken. Needed some user-created script to fix it.

        I have tested a dozen distros recently.
        I wanted to switch once again to a Linux distro as my daily driver.. but all have hurdles and or failings that are annoying deterrents or break my workflow/s.

        The other thing is my windows install on ~5yr old HW isn’t slow. It does not crash. It is the same as my corporate build for work. I just wanted to see if Linux has progressed.

        I feel the distro and desktop environment teams need to respect the competition and not dismiss it so easily.

        The 1990’s windows/Mac experiences that the Linux community quotes and bashes are long gone. Well-honed to solid, usable work tools with decent looks.

        Like

          1. Yeah, it’s especially useful to complain about the linux community’s attitude by insulting them and being condescending…

            I gues Nate is famous now: this blog’s attracting trolls!

            Like

  15. My 2 cents: I think Telegram has something in common with this and it has success.

    Telegram entered the market having to compete with Whatsapp, Facebook Messanger and other big names but it has a mission and values, just like Linux community.
    Telegram reached mass adoption thanks to
    1) ease of use but with advanced features (for me it realized the motto by KDE better than KDE)
    2) reliability and a bug-free experience (at least better than competitors)
    3) a bot platform that extend a lot its capabilities (the equivalent for Linux would be a real app platform for third party developers, compare Flatpak & other Freedesktop stuff to Android app platform to see what’s missing)

    You could argue that Telegram is easier to install because it’s an app available on major app stores but I think it’s clear that Telegram would not have been successful if it was at the same level of Linux desktop regarding these three points.

    Liked by 1 person

  16. Hey Nate,

    Please understand that I’ll be talking in the tone of a consumer – i.e. if I buy a device with your software pre-shipped (as you are proposing), I see it as a commercial product that I *am* entitled to expect to work right. This may make my tone sound demanding, but that is for a currently hypothetical scenario of e.g. Plasma shipping broadly on mainstream devices, I do not intend to insult or demand anything from you or anyone else here.

    In my experience, breaking bugs in mainstream Linux distros are handled very poorly. I’ve had all sorts of issues which were solved relatively quickly in the git repos, but for them to be shipped, it took whatever time was left on the release schedule. I think you folks try hard to QA the software enough that it doesn’t happen in the first place, but it is also obvious to me this just does not work out, and there is no way around accelerating bug fix distribution on a big scale. I had several small crash and reliability issues throughout my time with Plasma, but the biggest issue, which I unfortunately never tracked because it made me switch to GNOME entirely, was that the “present windows” effect had a chance of just vanishing all my windows. They were gone, there was no window management trick to make them re-appear, I had to quit them all and start from scratch. For software I want to rely on, this is absolutely unacceptable, and because I had no time or will to spend the effort of debugging the issue, as no consumer would, I switched to GNOME instead. Yes, totally my fault as a today’s Linux user. No, totally not my fault as a hypothetical future customer for pre-shipped Linux software. One cannot conquer the mainstream *and* resort to “but we are doing all this for free”. And yes, Windows also has bugs, but with no version of Windows did I have this many bugs in such a short time that lasted as long as they did on Ubuntu or Fedora with Plasma and GNOME.

    As that is a common narrative of the rest of the Linux community, I hope nobody feels like starting a distro war now as in “if you used rolling release, your issues would be fixed much faster!”. It is a very awkward thing in the first place that the OS family with the, by a gigantic margin, least amount of consumer users, has the, again by a gigantic margin, biggest amount of flavours, all with their unique issues of why they might be a bad idea for the regular user. Ubuntu LTS has old packages and may fail to work on bleeding-edge hardware. Fedora has new packages, but the first-party software support is much worse than for Ubuntu. Most Arch-based distros are a pain to maintain, have abhorrent stability, or both, and certainly the least broad first-party support. So, in comparison to what we could have, we have very many but very tiny groups of people working on strongly related but marginally different things, duplicating a large amount of work, yielding dozens of distros that are so-so than (preferably much less than) a handful that are just very good.

    I’m not too technically incompetent to maintain a Linux install. I’m not too incompetent to read technical documentation. I’m not even too incompetent to debug a bunch of the software I faced issues with. I just wish to use Linux as an OS that just works, and it just doesn’t. You need to research, you need to read technical documents, and you need to tinker with some of the technical details. With me having literally any issue with its usability, I would not for the life of me think of recommending it to someone else. My father needs help locating elements in an already open context menu – no, I cannot explain to him how to e.g. reset a cache. 🙂

    It does not really help most Linux distros are based on embedded system or server software concepts, which do not work for consumer machines at all, but most people still praise nevertheless. Package management is a nice thing when you are a system administrator or a platform designer. But for the normal consumer, all its deeply-rooted nature does is dramatically increase the risk of things like the Linus Pop!_OS debacle. And no, that was not just “unlucky timing”, and no, that was not Linus’ fault for not reading ten paragraphs of text. Something like this must simply not even be possible on an architectural level – and it is not on Windows or macOS. This should not be “guarded more” with even scarier warnings or awkward text files the user must create to overwrite the behaviour, this must be gotten rid of. Things like Flatpak need a *lot* more attention and a lot more first-party support to iron all issues out, especially regarding flaky sandboxes.

    I could go on for hours, but I think this is long enough. 🙂

    As a last very vague point, maybe compare a Linux laptop experience with a Mac one – out of the box (yes, yes, people also have Mac bugs…). If I get a modern MacBook and I set it up, I know it basically has maximum security (Secure Boot, SW measuring, FDE, Gatekeeper, …) right after logging in. I know that it runs the software that a Mac runs, and I know I can get it from the App Store or from first party pages. I don’t need to worry how to enroll a MOK for my NVIDIA driver to support Secure Boot, I don’t need to figure out what LUKS is and whether an attacker cannot just replace my unauthenticated initrd to keylog my password despite Secure Boot being on, I don’t need to worry about the package format my distro is using or how Discord expects me to install a .tar.gz-packaged application, I don’t need to worry what a PPA, rpm-fusion or the AUR is, I don’t need to worry whether my apps support Wayland pipewire capturing, and I (usually!) don’t need to worry about Linux suddenly introducing a bug where my user session would retain a black image no matter what when waking the monitor from sleep (but any other TTY still working fine) out of the blue. It’s like buying peace of mind.

    Thanks for your work and efforts!

    Best regards,
    Marvin

    Like

    1. That was a bad bug, yes. It’s fixed now thank goodness, but I understand that something like that can be so traumatic as to tarnish a product in your eyes forever. It’s understandable. This is why I focus so much on QA, but clearly it’s not enough and we need to do more.

      Like

  17. From operating system kernel point of view and the device drivers for it. The situation could improve but is good enough. From user interface point of view. The situation could improve but is good enough. From past and future development point of view the situation is currently rather bad. That is technologies like Wayland, sanboxing and overall mobile influence on computing proved too be too big of an obstacle for Linux on desktop. Overall this development just isn’t moving ahead with a pace that would make it meaningful. Nobody takes you seriously in real life if you need 20 years to reach some goal. But this is not what is holding Linux on desktop back. This is something more technical users tend to be aware of. But an average user doesn’t care all that much about it. More hardware vendors are not the solution for this problem. Hardware vendors will adapt quickly if it will make sense for them to adapt. The real reason on why we don’t see more Linux on desktop is applications. It really just comes down to this. When an average person will be able to install the latest version of their favorite application on their Linux desktop without much hassles involved. Then the rest will follow. To get applications you need developers. This is by far the number one reason on why. You can persuade just about anybody to give Linux on desktop a try. Most of them will like it. The reason they don’t stick to it is they can’t install an application they use. And for most people the story ends there.

    Like

  18. As long as Microsoft bullies schools to install Windows, new users never have a chance to use Linux. When they are grown up, they only accept Windows.

    Like

    1. There is usually a person involved that other persons ask for advice. On what operating system to use. Recommending Linux on desktop in 2021 makes much sense as it’s fast and easy to install, free as in speech and usually free as in beer, respects privacy, the overall experience is pleasant and on top of that customizable … BUT there is one mayor drawback. There is that one application or game or service … that currently doesn’t support Linux on desktop. And there is where the story usually ends for most people. But each year this does improve. More and more applications, games, services … do support Linux on desktop. At this rate i would say that in around five years time Linux on desktop will reach the point where for most people that application, game, service … will work for them. And that should reflect in Linux on desktop numbers to start growing substantially. Hardware vendors will for sure follow and adapt to the demand.

      Like

  19. I think KDE will eventually have to produce or be heavily involved with a shippable distro of its own eventually. KDE Neon is fine, I think, but the fact that it’s based on a super old LTS distro means that a ton of stuff will be outdated at any given time, when distributions already struggle with outdated and missing packages. Kernel supporting the newest unobtainable GPUs? Inkscape getting updated to its latest release? OpenConnect needing updates to handle the company VPN, in addition to the accompanying network manager GUI? Steam battling with incompatible system harfbuzz?

    It’s bad enough that Discover shows Flatpak backend and PackageKit backend in the “Missing Backends” section, despite Flathub having a checkbox and pacman fetching packagekit data on updating. I figure this is my fault for using Arch, otherwise known as one of the few KDE-friendly distros that aren’t hopelessly outdated most of the time. And this is fixable.

    But on top of that, the basic use case of installing a non-distro app still doesn’t have a clear-cut answer. If I go to the Balena Etcher website, why does it make me download a .zip file with an .AppImage that loads as hex garbage when opened with Ark? Why does Sublime Text “only” list a bunch of terminal commands (which we now know shouldn’t be required) and no hint at a Flatpak source? How can I help this GitHub person with packaging their app for consumption by others without becoming a distro maintainer? And how am I supposed to install optional library dependencies when Discover doesn’t list my system packages? How do I configure whether my leftover Windows app installer .exe will run under Wine or Proton when I open it from Dolphin? I read that it runs, but now some dialog is complaining about some C# mono thing being required on my system and it doesn’t automatically install that.

    But the worst part of all of this is that the multitude of answers to these questions depends not on KDE, but on the particular distro. Which puts KDE in a position of not being able to solve the issues to begin with. Which makes it hard for anyone to put KDE on devices, unless they have the resources to run an entire distro by themselves like Valve or Jingpad. But good luck getting me-too vendors like Dell and Lenovo to put KDE on machines when there’s an all-in-one solution from Fedora (GNOME) and Ubuntu (GNOME) out there.

    I know this is not a fair comparison because GNOME doesn’t run its own distro either. But actually in a way it does, because for the most part it’s just the desktop arm of Red Hat at this point, and boy do they make sure that stuff integrates well.

    I think KDE will need to have a long and hard look at providing an authoritative answer to software installation and integration, or get steamrolled by even new distros with attached desktop environment that look at these issues holistically. One might argue that this has already happened, not just with GNOME-based distros but with Linux Mint, elementaryOS, Zorin.

    I also know that Plasma needs to be flexible enough to fit any random base platform, so that Valve can run it as a secondary environment on SteamOS instead of KDE dictating the exact OS dependencies. So clearly there is a balance to be struck between integration and flexibility.

    But right now, if I go to the KDE website, I can’t even find a frickin’ Download button. For anything, OS or apps. So users interested in a KDE OS aren’t getting guidance. Developers have no standard platform and leave users to figure out the deployment mess. Hardware vendors get to choose between outdated KDE Neon, various distros that don’t default to KDE, and rolling their own OS.

    Okay, perhaps some of this was overly dramatic. Not by much, though. I believe this is a major pain point currently with no clear plan or task force to attack it.

    Like

    1. Correction: While the website definitely doesn’t provide a prominent OS download button, there are indeed links from the “See all applications” subpages. Not from the Kontact feature page on KDE.org, though.

      Like

      1. Correction #2: The Kontact feature page does also have a Download navigation item, which I stupidly missed when quickly checking on this smartphone screen because it’s not a button, just an expandable menu entry. On desktop this shouldn’t be an issue. And yeah, links to various distributions with optional KDE can be found two clicks away from the frontpage. I’d like to retract my mention of KDE products download discoverability, in the hope that the points about distro-specific installation difficulties still stand.

        Like

    2. I agree that we probably need our own distro to really up our game, and that neon isn’t it–at least not in its current form.

      Like

Leave a comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s