Why Year of the Linux Desktop is Bullshit

The title of this alone speaks volumes greater than the exposition that’s to follow, and I’m sure that some of my peers are already bounding from the sheds with pitchforks and torches in hand, but I’ve never been one to not voice a concern even when the house is burning down.

Yet again, we in the Linux users community find ourselves at an interesting juncture. Microsoft has as of 14 January 2020 officially EOL’d Windows 7. As with XP before it, this will likely be a major issue for the immediate future considering how prevalent its use is in the desktop market (Gartner estimates still have Windows penetration at greater than 90%). As expected, most of the podcasts and reporting sources are cobbling together pieces to launch yet another slew of volleys, perhaps to rally the uninitiated to take another look at Linux if they haven’t already done so.

But, who listens to or reads these sources? Non-Linux users?

This nonsense of the Year of the Linux Desktop has been going on for as long as I can remember. And from working with those who have vastly longer tenures than myself in the Linux world, it seems as if it caught on well before my dive nearly fifteen years ago. From the gate, I too shared this sentiment. I had been a Windows user from the start, and although Linux was a monstrous beast to handle back then, I still loved it with all my silicon heart. And I, like every other witness, wanted to espouse my love to the world in the most compelling and boisterous method I could imagine. I’d rake myself through the coals of Hell to learn all I could about Linux, and while still salving my burns thought there was no way anyone else wouldn’t want to be a part of this. The computing revolution was here! Or was it?

Truth be told, the premises of either open source or free software were entirely lost to me until about seven years after I’d taken the swan dive. The fact that I was using something that wasn’t Windows, and that I could be as hardcore of a programmer as I wanted were the only things I cared about. None of this liberty crap mattered. Software wasn’t a first class citizen having attributes of sovereignty. I just wanted to be as nerdy as I could because I wanted to. The politics of software, which would come in much later, seemed to dissolve this juvenile yearning in me to a large extent. It wasn’t about being a nerd anymore. All of a sudden, it was all patents and licenses and codes of conduct and ruination abound.

So if it weren’t for being able to divorce the creations from their Gods, why would I want others to use Linux? To know and embrace it the way I had?

Two words: technological empiricism. I was a fucking God amongst men, a wolf amongst sheep when it came to sheer computing skill, and I knew it, and I was more than happy to get everyone else onboard, their readiness for the transition be damned.

Turns out, this wasn’t too dissimilar of a position for young bucks in my pool at the time. The early 2000s were ripe with young technologists who, after having done whatever it was they did to make the Internet the cultural platform it ended up being, just wanted to flex on the boys and girls as much as possible. We didn’t skip leg day at the gym, because we made the gym.

Even then, it seemed as if using Linux was a horrific experience. Most all of my peers were still using Windows 2000 and Windows XP, especially those of us who fancied ourselves programmers. Hell, the first commercial software I wrote was in VB 6 and was a team effort (which was hilarious in its own right). But I ported it to Linux on my own using the customary tools of the day: glibc, GTK, Glade, MySQL (MariaDB wasn’t even a thing then), etc. And guess what? It worked! But guess what, again? The damned customer was running Windows 2000, and that silly little abstraction layer, GTK for Windows, wasn’t going to be running this shit anytime soon. Besides, who statically links?

The horror of this for programmers was one thing, but the stage was entirely different for the pedestrian user. Games? Well, if you like spending hours playing AisleRiot, then Linux was the platform for you. I mean, why wouldn’t you want to trade playing WoW or Diablo for good old GNOME Mahjong? Total no-brainer. Office? Come on! That’s easy! OpenOffice (LibreOffice wasn’t a thing back then) was the killer app that could do EVERYTHING that MS Office did (this turned out to be a colossal lie then and still is now, despite everything that The Document Foundation does with LO; don’t believe me? Try converting a SMB to LibreOffice from MSO). Internet Browser? Oh just use Mozilla, because that was super compatible back in the day. Need to install some software? Just use the terminal!

Oh… wait. THAT thing.

We’ve already stumbled upon what is potentially the largest issue with Desktop Environments on Linux. What started as a detailed thesis in the form of a borderline anthropological analysis of why the current landscape of Desktop Environments sucks was boiled down to three quintessential matters, this being the first and perhaps the largest.

You know what’s unattractive to the pedestrian computer user? Terminals. You know what contemporary operating systems do a bang up job of getting those out of the way? Windows and virtually every OS that Apple produces for their product line. You know what operating system practically begs to be used nearly exclusively by the terminal? Linux. You know what Linux software doesn’t do at the terminal? Provide a consistent or coherent interface. You know what pedestrian users are most afraid of concerning their computers? Breaking them. You know what breaks Linux computers? Using the right command in the wrong context by accident. And without the aid of a witchdoctor who bears the scars from having mutilated themselves to possess such knowledge, those people are fucked.

Short: Every single Desktop Environment, then and now, does an absolutely piss poor job of abstracting away the need for a casual user to ever whip out the terminal.

But Greg, that’s all bullshit because there’s plenty of cases where a DE does what you’re saying it doesn’t! Yeah? Let’s through a few use cases here that the casual user takes for granted in other operating systems and environments that requires a terminal regardless of the DE:

1. Installing a group of related software. In RHEL-derivatives, this is usually handled through groups in either YUM or DNF (modules in the contemporary sense). Guess what doesn’t show up in GNOME Software? Groups or Modules. In Debian-derivatives, this isn’t even a concept that APT knows about without first getting tasksel. And even after installing it, guess what doesn’t show up in GNOME Software? tasksel groups.
2. Installing software. Despite the Windows Store being the hip place to get Windows 10 software from, it’ll never quite be the thing that Microsoft wants it to be because developers can’t distribute traditional binaries through it; they’re required to be repackaged in a fairly unintuitive manner. Ergo, with Windows having the Lion’s Share of the desktop market, pedestrian users have been habituated to installing software in the legacy fashion. And let’s face it, the Windows 10 Store is full of garbage in the same way that each mobile app store is. Guess how you can’t install software on Linux? The legacy method that every user is accustomed to. BUT WAIT, GREG! WHAT ABOUT APPIMAGE OR SNAPS OR FLATPAKS!? Bullshit, each of them. They’re great for those of us who’ve abused ourselves for years to get these programs to work otherwise, but you still run into issues with dependencies (try installing a Flatpak on CentOS 8 that requires X.264, only to find out that you can’t upgrade the base Flatpak installer because it’s about nine versions behind, or to do all of the manual hacking required to get certain Snaps permission to break their cells and access otherwise inaccessible resources on a system (virtually anything that installs with the –classic switch)). But surely, anything you install from the software store that comes with the DE you’re using is good enough, right? Yeah, it usually is, for the most part. Until one realizes that you can’t get a piece of software that you would otherwise have obtained on either Windows or Mac without incident. You want Google Chrome? Not available without downloading a DEB or RPM. Need some AV codecs because you can’t view DRM content or Flash content? Guess you’re adding some repos (you know, because EVERYONE keeps their AV in OGG or Theora, because those are solid and x-plat formats).
3. Troubleshooting DE. But wait! Didn’t know that the DE runs atop a Window Manager? Didn’t know the Window Manager runs atop a Display Server? Don’t know how to change VTYs when the whole thing goes down the shitter? Oh well.

The next point: the DE themselves miss virtually all the targets required to hit to render a casual user experience viable. I get so sick and tired of hearing people claim that the current landscape of choices in not just Linux Distributions but DE are such a good thing because, as one Matt Hartley put it, “One man’s perfect distribution isn’t another man’s perfect distribution.” Bullshit. Total fucking bullshit. I don’t even know where to start with this.

I can’t resist the urge to address the elephant in the room here that trespasses onto the larger, slightly relevant, topic of Linux Distribution Saturation. The question is this: if I download Ubuntu and make some minor tweaks to user land, does that actually mean I have a different distribution, further that it warrants creating a whole new entity for consideration and download within the Distribution Space? The crux here? Most Ubuntu-based derivatives are still using the fucking Ubuntu repositories. I can’t stress that point enough. If a distribution is supposed to be, at its core, Linux with a swath of software available to make the user experience more concrete, and you’re not offering any different software than what’s available upstream, why the fuck are you creating a distribution? Changing a handful of packages, or forking because you’re butthurt about the init system used, isn’t a new distribution. Why do we think it’s okay to do this? Why do we have over fifty Linux Distributions to choose from? How is this indicative of offering clear choice to outside users? Plainly, it isn’t. It’s offensive to not just those in the community, but mostly to those outside.

Off the top of my head, these are the DE that I can think of that are available to choose from: GNOME, KDE, MATE, Cinnamon, Unity (if you’re still using slightly older versions of Ubuntu), XFCE, LXDE, LXQt, CDE, Budgie, Enlightenment, Razor-Qt, Pantheon, Lumina, and that one that Deepin uses, which I think is just called Deepin. Each of these expresses colloquial desktop metaphors in different ways, each has their own quirks about customization, each has their own methods for enterprise considerations (actually, most don’t even consider this, and if they do, the implementation is fucking horrible and unmanagable at best), each come with their own suite of tools that functions differently than the next, etc. I can’t go any further here without wanting to throw up. The fact that only a select few of these come close to being inviting to a casual user is appalling, and even these fall fatally short of the root objective, amongst many others.

It’s worth pointing out that although most desktop metaphors aren’t codified in anyway, that doesn’t mean that casual users are malleable to the point of wanting to abuse themselves endlessly to use their computers, and you can foist whatever you want in front of them. For fuck’s sake people, we’ve had traditional metaphors being expressed since GEM, and although it might be time for some change, tell me how well that’s worked out for you on the desktop form factor? Why take something that works, something that people are accustomed to, and not just break it, but irreparably obliterate it?

Oh, and here’s the other hilarious spin sold as a positive about the Choice Paradox concerning DEs: if you don’t like the one you have, just get another one! Blech.

1. This is an unbelievably moot thing to say to a casual user. None of them view this as a benefit. Being accustomed to just using the computer, it doesn’t dawn on any of them that Explorer is a shell as much as it is a file manager, and that it can be customized or even replaced (unsure if this is true for Mac). Whatever they see first is what they’re stuck with. End of story.
2. Even if you manage to get someone beyond this point, just how do get another DE? Do you install it alongside your existing one? Do you get a fresh distribution flavor? If the latter, you better hope you partitioned your system correctly, otherwise your shit gets blown to kingdom come.
3. Having multiple DE running parallel on a single installation is a fucking nightmare.
   1. The only real safe way to install an alternative DE is to get it through a group in your default repositories. See aforementioned point about installing software groups. If you don’t see it here, you’re already about two-thirds the way jumping the shark.
   2. Once you get it installed, you’ll likely have to compete with the idea that your distribution will have a strong preference for the Display Manager. Wait, what’s a Display Manager? Oh yeah. Forgot to mention that little bit. The program that logs you into the computer? That’s the Display Manager, which is yet again an entirely different component. Anyway, for example, CentOS has a strong preference for GNOME Display Manager (GDM). Even if one installs the KDE flavor of CentOS, you’re still going to be using GDM rather than SDDM. To put the icing on the cake, let’s say you’ve got a system with both GNOME and KDE on it with GDM as your DM. You have to know enough about GDM to know that you need to change the session type to KDE Plasma instead of GNOME Shell, because if you don’t change it, you’re going to keep using GNOME Shell.
   3. If something goes wrong, or if you just decide on a DE you like and wish to exorcise the alternate beast, good fucking luck. Pulling a DE out of a running system is like getting a steak out of the throat of a lion. Sometimes it can be easy, others run you the risk of crippling your system if you’re not paying attention to how the package manager is resolving dependencies for removal. Guess what can’t pull these out safely in some cases? Graphical software managers like GNOME Software or Discover.
   4. If you decide you want to be like Lois and Clark and go deep in on running multiple DE in parallel, you’ve now got an issue where you’ll have multiple programs that do the same damn thing. There’s nothing cooler than looking for a terminal emulator and seeing Konsole, XTerm, and GNOME Terminal all at the same time. Sometimes I want to look for files using Dolphin, but other times Nautilus gets my gutchies that day.

This had to be obvious to someone at some point, because we have distribution flavors. Ubuntu has several: Ubuntu, Kubuntu, Xubuntu, Ubuntu MATE, and Ubunty Budgie (Kylin sort of doesn’t count here). The reason why these flavors exist? Ubuntu gives you GNOME Shell, Kubuntu gives you KDE, Xubuntu gives you XFCE, Ubuntu MATE gives you MATE, and Ubuntu Budgie gives you Budgie. This. Is. The. Only. Reason. These. Exist: to isolate the DE from each other for a hopefully more gooder experience than if one were to use Ubuntu and install KDE inside it.

How is this feasible? How do you attract users to this? How is it EVER going to be the Year of the Linux Desktop? Can we please stop this nonsense madness of blindly repeating ourselves about dominating the desktop space? It isn’t going to happen when things look like this. Not. Fucking. Ever. All we’re doing is circle jerking with ourselves in a fantasy where we can finally say we came out on top. If we want desktop dominance, which may never happen, we should at least attempt to start with these goals (IMHO):

• Standardize. There’s nothing more annoying to a casual user than too much choice; Choice Paralysis is a real thing, think buying toothpaste. Maybe this means consolidation. Maybe this means a new project whose focus is on these things.
• User Focus. Make a product whose core philosophy is the user and their experience rather than an experiment with cool code. Software shouldn’t abuse users or require them to abuse themselves.
• Ease of Use. This should’ve been a no-brainer, but methinks the horse died some time ago and we just agreed to leave it be and not replace it. Anything that could be done at a terminal should be able to be performed through the UI, no exceptions. Metaphors are not play things. We have established ones that work considering the form factor, so fucking use them. They work for a reason: they don’t assault the user.
• Customization. It should be EASY to customize your environment, and it should also be EASY to sell to an enterprise for adoption. It shouldn’t be a fucking sell of a nuclear power plant to get people to use this technology.