Resignations Signal Generational Change at Apache Foundation
Why Gnome 2 Continues to Win the Desktop Popularity Contest
Where Open Hardware Is Today
Writing in Style With LibreOffice
Huawei and Other Mobile 'Tech Giants': You Should (Really) Break Free from Google/Android
July 8th, 2019

How User Revolts Shaped the Linux Desktop

The user revolts against KDE 4, Gnome 3, and Unity have left desktop Linux developers with a fear of innovation, exactly when that’s what’s needed.

mob in revolt

Source: Pixabay

Between 2008 and 2011, KDE, Ubuntu, and Gnome all released radically new interfaces. The mediocre reception received by all three has left developers so cautious that innovation on Linux desktop environments has been curtailed, except for minor changes. Yet innovation on the desktop is long overdue.

Before this brief era, Linux DEs were focused on equaling the functionality of their proprietary equivalents. However, by 2005 this goal had been realized and developers had begun discussing what should come next. Particularly among KDE developers, the feeling was that Linux DEs had the chance to become pacesetters in interface design.

This feeling was expressed by Mark Shuttleworth of Canonical in a speech a few months after KDE’s radically different 4.0 version was released. Speaking at OSCON in July 2008, Shuttleworth asked, “Can we go right past Apple in the user experience we deliver?” The question was a challenge, and developers who were not already considering how to innovate soon were.

Unfortunately, one by one the major attempts at innovation faltered, not always because of the programming. KDE 4.0, for example, was a major departure from the KDE 3 release series but might have been better received had KDE explained more clearly that it was a developer’s release, and distributions had not rushed to include it in their repositories. The result was a users’ revolt that was only cut short by a rush of new features in the next two releases that answered most of the complaints.

Other programmers, though, were slow to learn the lessons of KDE 4.0. At the time, Linux’s popularity was still new, and many programmers were not used to criticisms from end users. Not long before, programmers and users had been close to synonymous. Consequently, many responded to criticism with sarcasm and suggested that the complainers should code the changes they demanded for themselves.

This attitude perhaps explains why two more user revolts followed.

In the case of Ubuntu’s Unity, the design was so uninspired that the DE never managed to top 5% on any user polls. Users resented the fact that interface designers newly hired by Canonical suddenly had control over Ubuntu’s DE. More importantly, many of the changes were not so much innovative as arbitrary, such as the placement of application menus in the top panel rather than in separate windows. Other changes, such as the so called Heads-Up Display that was supposed to replace menus, were never finished.

Even after the initial outrage receded, Unity was regarded with indifference. From my observation, the majority of users replaced it immediately after completing installation of the distro. Later, built-in Amazon searches lead to privacy concerns. By the time Unity was abandoned in favor of a return to Gnome in 2016, few could have been surprised.

The Gnome 3.0 release in April 2011, less than a year after Ubuntu introduced Unity, was in many ways a replay of the KDE 4.0 release. With little or no consultation with users, Gnome developers decided the best way to improve the popular Gnome 3 was to remove clutter from the desktop (“minimizing distractions such as notifications, extra workspaces, and background windows,” was how Gnome’s press release put it).

No one, apparently, had counted on the fact that many users wanted those distractions for their work. Gnome resisted complaints for months before abruptly capitulating with the release of Gnome Extensions that allowed more customization. By then, though, Gnome’s user base had fragmented, much of it gone for good.

The Fallout

The short term result of these user revolts was increased choice. Xfce, for years the third most popular DE, surged in popularity, and today often polls better than Gnome. To a lesser extent, LXDE revived. New DEs, like Budgie and Linux Mint’s Cinnamon and MATE, were also created. All these DEs had two features in common: their developers listened to users, and changes were mostly small improvements rather than radical new directions.

If that second feature seems too much of a generalization, consider the upcoming Xfce 4.14 release. Developers have labored over it for several years, and I am sure it will be gratefully received by users. But although I have no wish to denigrate Xfce’s developers — who are only giving users what they want– consider the key features of the upcoming release: a migration to GTK3, bug fixes for the panel and other parts of the DE, and the optimization of the file manager. All are no doubt welcome, and the migration is no small matter, but have more to do with housekeeping than innovation.

Nor is Xfce an exception in this regard. The same can be said of any of the leading DEs today, although Cinnamon has shown flashes of innovation in recent years such as the addition of hot spots on the desktop. Even KDE Plasma, the most innovative DE today, emphasizes what is safe and de-emphasizes innovations like Activities. In fact, code-cleaning actually removed some features of Activities.

Moreover, when it came time to produce a mobile UI, Plasma developers chose to build a conventional interface from scratch rather than dust off the more advanced Plasma Active with its spinner racks and Activities. The collective instinct seems to be to shy away from anything too different.

The Need for Innovation

Modern developers cannot be blamed for their reluctance to change DEs too greatly. No one needs to excuse themselves for preferring not to have the work of months dismissed by often ignorant outsiders.

Yet, at the same time, we are still using interfaces based on the same principles of those used in the days of 20 MB hard drives. The menu in particular has become a click-driven dive through a maze of menu levels. We endure the inconvenience because it has settled in slowly, but a better alternative is overdue. The same goes for the crowded panels and notification areas. Yet we continue to trundle along with obsolete makeshifts.

Perhaps, like monolithic corporations, DE projects have become too large for us to expect innovation from them. Perhaps a small new DE is necessary to provide the innovation that is needed. In the meantime, the lesson of the user revolts encourages DEs to keep safe.

Bruce Byfield has been involved in FOSS since 1999. He has published over 2000 articles, and is the writer of "Designing with LibreOffice," which is available as a free download at http://designingwithlibreoffice.com/download-buy/

32 comments to How User Revolts Shaped the Linux Desktop

  • Mike

    I don’t think you really make the case that a better alternative is overdue. After all, change for change’s sake is the approach that led to the aforementioned massive user revolts. You can get away with unilateral change if you are Apple because you have massive marketing and public relations machines, as well as captive users who actually have no choice. In the FOSS world, users will go where their needs are being met. Unarguably, Gnome and KDE’s radical changes did not meet their users’ needs, leading to the rise in popularity of alternative desktops. Trying to blame those uprisings on user resistance to change would be missing the forest for the trees.

  • @Mike I would tend to agree. I use Xfce precisely because it stays out of my way. One of my favorite desktop interfaces was the default screen for Windows 3x, which was basically a file manager.

  • Eddie G. O'Connor

    I’m gonna have to disagree with this article. But hear me out first:

    Take a look at Windows any version (even the dreaded 8!) the one thing that remained constant (in the case of “8” you had to find the settings I believe that would take you there!) but you had the same “Taskbar”….the same “Start Button” the same “File Explorer” etc. Take a look at Apple and you’ll notice that although they change colors or themes?…their setup is basically the same as well. Now, I ask you exactly what’s wrong with these scenarios?

    Answer: NOTHING.

    Understand; I’ve been using Linux since 2002 / 2003, and I have seen all manner of things within the Open Source universe, and while I can appreciate this author’s desire for innovation? To scramble things up and remove features people use and move things to different parts of the system all in the name of “innovation”? Is a recipe for disaster. Just take a look at the examples given in the article. I used GNome 2 for the longest while, and when they made the light-year jump to 3?…at first I revolted, but eventually it became my favorite desktop environment. And while a lot of folks might feel the same? we’re the minority, because a LOT of people used Gnome 2 and saw NOTHING wrong with it! So when the “cataclysm” took place, they logically jumped ship and found other desktops that did WHAT THEY WANTED. There is never a time when changing things just for the sake of changing things makes sense. ESPECIALLY if people are relying on your product / package / application to get “work” done! (and by “work” I don’t mean a home hobby, I’m talking about income producing work!) It’s hard to be productive, or to get work done when every other month some developers decide to switch things up leaving their users hard pressed to try and figure out how to go about working in the new interface. I feel that with the plethora of desktops that exist today, there really isn’t a need for “new” just yet, this isn’t to say that the devs should rest on their laurels and not push to make their DE the best it can be, but there’s other things that need to be worked on besides how something LOOKS, sometimes there’s the nagging error users get everytime they update, regarding some glibc file, or its the way something seems to “revert” back to standard mode when the user has already customized it. There are a ton of things that need “fixing” regarding the Linux desktop as a whole, but tinkering with looks, and trying to become the next iOS or Android/tablet interface UI is the LEAST of our problems! At least…this is how I view things…YOMD (Y-our O-pinions M-ay D-iffer) LoL!

  • Bruce Byfield

    A DE that keeps out of your way and innovates is not necessarily a contradiction in terms.

    Plasma is a good example. It installs with a basic desktop, but you can go on to use Activities if you choose. If you don’t choose, the DE is still perfectly functional. But it’s not a case of either/or.

  • Bruce Byfield

    “tinkering with looks, and trying to become the next iOS or Android/tablet interface UI is the LEAST of our problems”

    A strange assumption. Believe me, imitating Android or iOS is the last thing I’d suggest. Both offer mediocre interfaces.

  • James

    I don’t know what is ‘getting out of my way’ really means? It’s NOT that DE is an overlay that hangs over the screen. XFCE, KDE, LXDE or dozens of other DES pretty much do the same thing. No one is more obsessed with the desktop Linux communities. Everyone is happy with their 10,000 users. Desktop Linux is a cess-pool. This is what you get when you don’t get corporate sponsorship. Linux is successful virtually EVERYWHERE. Even Microsoft is now using Linux. It has failed in one area and that is the desktop. The reason is quite clear. It’s the desktop communities who are digging its grave.

  • Mike

    @James

    Many people make the same mistake of conflating linux’s failure to dominate the Notebook/Desktop PC market with some kind of failure of the linux ecosystem. Fragmentation of developer effort and of user community is often held up as a sort of bogeyman to explain all of FOSS’ woes. It’s a strawman.

    The two things are completely unconnected.

    First, the quality issue: Several of the linux desktop environments are “good enough” to be used in a commercial environment, and acting like Windows or OSX environments are any where near perfect is laughable. As an example: OSX has a crappy and pretty much universally despised file manager, yet somehow Apple manages to limp onward despite this. Meanwhile Microsoft shoots itself in the foot with broken updates every over week it seems.

    Linux hasn’t succeeded in the Desktop space due to one single reason: Microsoft’s monopolistic behavior. No other reason. Microsoft does not have a better product, but they can direct market forces to keep the status quo using a variety of tactics, from payoffs to outright threats and bullying. Even Apple with its veritable mountain of cash can barely make a dent in the PC market. What hope is there for a set of loosely related open source operating systems which are definitely not a monolithic commercial product? Google uses these same tactics to maintain Android’s dominance in the world of mobile devices.

    No: Corporate sponsorship isn’t the answer. That way just leads to the Googlization (it’s now a word because I say so) of everything.

  • Doctor X

    @James

    FYI… just to correct you on something…. Gnome is owned/coded by Red hat. The same ones that owns/coded systemd… KDE is still independent, so is xfce. Cinnamon and Mate are products of Linux Mint.

    By owns, i mean they do the majority of the direction and coding. RH are so full of themselves, they do not take constructive criticism. That is what is the problem. They think they know what everyone else needs and wants. Ripping out functionality is pervasive. KDE at least gives you choice.

  • Rob. S.

    As Mike already put it, it’s change for change’s sake that offends people; “innovation” just for “innovation”‘s sake, just to get something ‘new’, or to make things look flashier instead of actually making things better, more functional, more usable.

    A concrete UI change for change’s sake, even if it doesn’t make things worse, does, by nature, reduce usability and efficiency, if only for some time until people get used to it.

    Change for change’s sake is what we see with every new Windows version, and every new Windows version since Windows 2000 brought more deterioration than improvement, with Vista and 8 probably being the most prominent examples, and 10 still being a catastrophe from a usability-oriented UI design perspective.

    Just because of change for change’s sake.

    It’s what finally made me migrate to Linux at home. And while I’m an IT professional, I know several non-tech people who went to Linux for the same reason. They didn’t want each new Windows getting more in their way. And they chose Mate or Cinnamon, if not Xfce. And they did so for a reason. Not because they’re neophobes or something. Just because they don’t want things changed to the worse which are working perfectly well as they are.

    If Linux DE designers/developers really want to innovate, they should perhaps first try to achieve the functionality and usability of OS/2’s Workplace Shell, from the nineteen-nineties, none of the existing desktops does. Just changing the looks of things doesn’t help anybody.

  • James

    Linux hasn’t succeeded in the Desktop space due to one single reason: Microsoft’s monopolistic behavior.
    – Putting blame on someone else for your own failure is desktop Linux’s worst argument. Chrome, ChromeBook, iPhone, Android,PlayStation, they have ALL proved that despite its deepest pockets Microsoft can’t own a market if there are real usable products. Desktop Linux is barely usable. It’s great for developers who might want a workable GUI, but it’s pretty much uselss for mass market.

  • Dutchkind

    Back then, when there was KDE3, everything was working well. Then came KDE4, and it started from scratch, with many important features from KDE3 missing. If I remember correctly it was a also the time pulseaudio’s incorporation in the system took place, as well as networkmanager. They both had there problems, adding to the DE’s problems. Other things that complicated KDE4’s reception were Akonadi and those tries for desktop search, as well as networkmanager’s need for kwallet and it’s overly complicated usage having the choice of either no password or having to type the password every time you login just to connect to the wifi!

    After years we had finally a stable DE back with a beautiful KDE4, and then it started al over again with Plasma, starting from scratch! So here we went again with a basic DE without any features, like a basic thing as automounting removable drives. Took again a long time to get to a usable DE. And I admit, I like Plasma. When Plasma started being developed pulseaudio was stable, as well as networkmanager and systemd, they just work, so it was easier to endure the DE with it’s missing features.
    So for me there is no problem to switch in the future to Plasma 6, but don’t start by removing all features, take Plasma 5 as the base of features and start adding from there on.

  • connie

    One desktop not mentioned here is Unity…that was essentially removed because of negativity from people who could just easily swapped to another desktop, but chose to also voice a hatred with such persistence that the desktop has been for all intents and purposes killed. It is a pity that this irrational hatred is always going to be noisier and more noticeable than quiet contentment.

  • Steve Chow

    Gnome’s user revolt was caused by themselves and in my opinion from their hubris.

    When they decided to go from Gnome 2 to Shell they decided to throw out the baby with the bath water.

    The popularity of Gnome 2 at the time I would say was as much due to Ubuntu as to do with Gnome 2 itself. Their whole they know better than their users approach really irked me at the time, especially when their overall design in my eyes was a mess. The workflow seemed to be based on i3 which is heavily reliant on keyboard interaction, but their graphical elements seemed to be based on a tablets.

    Then there is the whole choice of JavaScript as it lowers the barrier of entry to develop for Gnome Shell… Have they not heard of idiot filters???

    At least with Unity they had a decent strategy (the whole convergence thing), but when Ubuntu Phone failed and finally they decided to pull the plug on it, instead of looking at where Gnome 2 went (basically the Linux Mint team with Cinnamon and MATE), they picked Gnome Shell…

    Gnome Shell which only recently admitted to a memory leak and fixing it with a grenade…

  • Johnny

    Well, personally, I don’t think there is a need at the moment to design a whole different desktop. IMO, it’s a little bit like cars, there are new things being added, changed, but in general the interface stays the same, because it just works. What I see is needed in most linux desktops, is theming, easy theming. KDE is probably the best one there

  • Alan Robertson

    @Doctor X

    “Cinnamon and Mate are products of Linux Mint.”

    Actually the Mate desktop was developed by Martin Wimpress from Ubuntu:

    https://wimpress.com/

    Linux Mint does adopt the Mate desktop but they only created Cinnamon.

  • Mike

    @James

    > “Desktop Linux is barely usable. It’s great for developers who might want a workable GUI, but it’s pretty much uselss for mass market.”

    I know lots of people who aren’t developers who manage just fine (daresay even enjoy) using linux desktops. Everyday PC users don’t reject linux for technical or usability reasons. They have simply never heard of it. There is decades of evidence for Microsoft’s monopolistic behavior shutting out competitors, yet you have provided nothing to back your claims.

    There has been (and continues to be) significant market pressures that work in Microsoft’s favor in the desktop market that do not in others. Software market forces tend to reinforce the leader of a market. This is not always true in hardware, but Microsoft has wrangled control of the PC OEM market for decades. Your examples like Playstation, iPhone, etc. are radically different markets, where Microsoft never had the dominance necessary to exert the same level of control. Chromebooks have enjoyed a limited success against PC’s in education contexts because they are cheaper and easier to maintain, yet have failed to gain traction outside that bubble (one once dominated by Apple).

    Are there problems with Linux on the desktop? Sure, nothings perfect, but nothing really worse than the failings in Microsoft’s, Apple’s, and Google’s own offerings.

  • Glenn Thigpen

    When Microsoft hit the world with its Windows, buggy as the thing was, it was a step in the right direction for everyday users that wanted something that just worked.That graphical UI was an innovation that made working with computers easy for a lot of people that did not care to learn esoteric commands etc. to get work done.

    I was a KDE3 user and liked what it offered. When KDE4 hit the road, to me it was a step backwards and I switched to XFCE and have been with it ever since. It offers all I need in a user interface and the changes the XFCE4 team makes are incremental and usually not regressive. And I can make it as pretty as I wish.

    It seems to me that the developers are looking for the next big thing for the desktop but are forgetting what people want for their next big thing. Actually, I don’t think most people are looking for the next big thing. Most are pretty happy with what they they have. But, if some person or a group of persons come up with something that makes things significantly better and easier, users will be clamoring for it rather than revolting. Innovation that solves problems or produces a product that grips the imagination rather than innovation for the sake of innovation.

  • Mike

    @Glenn

    You brought back some memories I’d forgotten:

    Namely, people utterly failing to navigate the Win3.x UI. They didn’t understand the difference between minimizing a program group and a running program. Even on Windows 10, I see the same things with users who don’t understand a pinned application from a running one. Ah, good times!

    In DOS all you actually needed to know was DIR and CD so it wasn’t really very esoteric. Find the right directory and type the name of the program you want.

    Honestly I think interfaces can be made somewhat intuitive for beginners, but it’s a mistake to pursue that exclusively or to the point of compromising functionality.

    Make it simple, but not too simple. If you let them, people will demand something that requires no learning and does everything for them. Great for them, but a nightmare for anyone wanting anything ever SLIGHTLY different.

  • Jesse

    Basically, this article is only describing the mechanism of “convergent evolution”.

    DE’s and UI’s are stuck in the same ruts because users don’t want to be jarred by the transition from one machine to the next. Because, (also) gone are the days of only using one computer per day. 25-30 years ago, it was rare for the average person to use a computer at all. Now, children are being raised on a large variety of available devices. But, the experience is one of two mindsets: the workplace desktop of menu-driven clicks, or the touchable GUI icons of our handheld devices.

    In my experience, “brand loyalty” (where DE’s is concerned) is often largely governed by the comfort of familiarity. People feel smarter when they don’t have to re-learn basic tasks with every release.

    And, while the typical Linux user may be more adventurous and given to exploration than the typical device user; if a larger market share is truly sought after, there will be much pandering to those who don’t wish to waste time tackling ever-changing learning curves.

  • Yes indeed the Linux desktop has self-destructed … by the choice fantasy; usrs and devs together! Many options within each optional GUI … fools. For people who need to get work done … this work … at this time … in this way … choice is bad. All choice is all bad, or do you prefer a screwdriver that works like a ratchet .. unless it works like a pliers ? No screw would ever get turned ! I happily use MATE 1.2X desktop — GNOME-2 was the Mona Lisa of GUIs. Why should I suffer with Motherwell (KDE etcetc) , unless I like sticking nails thru my nose ?

  • Jim

    I laugh at complaining about the traditional desktop. To many clicks. How is it I can use an icon of my favorite programs on my desktop for one click, but that is to many? Or I can put it on the task bar, one click but that is to many? I have a variety of menu configurations I can use for the most efficient for me, but that is inefficient! I can edit my menu for what I use most, but that destroys my work flow? I can have 1 – 4 desktops, but that is not good for my workflow? Yet all these other “Modern” desktops hide my icons, prevent me from putting them on the desktop, or a dock that continually disappears and reappears, is more efficient? Or they arrange my icons as the developer finds efficient, not myself for my needs. According to whom? Not me. Maybe the traditional desktop survives, because it was already the best! Quit trying to fix what isn’t broke. I agree Gnome 2 was the best, and now I use Mate.

  • rclark

    I still say Win7 was one of the best UIs in the Windows world. With W10, I just tolerate it because I have too…. Ironicly, I find myself using the search (typing) to find what I want! Back to the DOS days! With Linux KDE was my favorite GUI up until 4.0. Then I started looking around and found lxde and xfce. Finally settled on LXDE. Point is, I can find in the Linux world a GUI that fits ‘me’. I don’t care about ‘pretty affects’. Big whoopie do. I want an easy way to get to and run the applications I use. Period. So bottom line, I do have to disagree a bit with the article.

  • Rick

    We don’t “need” innovation in the Linux desktop. What we do “need” is developers writing good, error-free code! The Linux world did fine with GNOME 2 and KDE for many years. GNOME 2/MATE is still the best and most efficient DE today. Except for a handful, virtually all distros are buggy to some degree today, some much more than others. That is exactly why the Linux on the desktop has still not caught on and probably never will.

  • KPPL

    With the current hardware (keyboard and mouse) and graphics hardware and software, the GUI is a problem that was solved already twenty years ago. One can tweak things here and there, but that’s all. Coming up with different paradigms here (Gnome 3.0, Unity and, to a lesser extent, KDE) is a recipe for disaster – they do things differently, not better; in fact, in the majority of cases (Gnome 3.0 being particularly obnoxious here) they do things worse. You might just as well be talking about innovating the wheel.

    Kudos to the Xfce developers, who seem to understand this. The Gnome 3.* developers may burn in hell.

  • epii

    James,

    “Desktop Linux is barely usable.”

    I’ll take Linux Mint, running the Openbox window manager and Tint2 panel, any day over that horrid mess that is Windows 10. FWIW, I use Linux Mint personally and professionally.

  • Please stop it with the innovation. Give us a back a plain usable twm-like desktop system with no stupidity, no file manager gui, no fancy crap that eats CPU like there is no tomorrow.

    I am sure there are people who like heavyweight GUIs and because Linux is modular it should be no problem to make that heavyweight GUI available for them.

    But it should not be the default and it should not be so tightly integrated into the system that it is nearly impossible to remove and replace with a simple X11 server.

  • Jeff Rollin

    Kludge beat me to it – and that’s just after skimming the responses, so maybe others did too. People have been using round wheels, living in houses and wearing clothes for thousands of years. We don’t need to start living in wheels, wearing houses and using clothes, still less eating clothes, inventing square or triangular wheels or living in potatoes in the name of “innovation,” particularly if said innovation is (a) barely functional, (b) buggy as hell, and (c) butt-ugly (here’s not looking directly at you, GNOME3. Yes, plenty people like GNOME3. Plenty others don’t.) Sometimes there’s no innovation in a given area because what we have works just fine.

  • Peter

    The day I found dwm and Arch, my search ended. The strength of Linux are the many DE and distributions, so where’s the problem? If users revolt, let them go to the competition.

  • Jim

    “The day I found dwm and Arch, my search ended. The strength of Linux are the many DE and distributions, so where’s the problem? If users revolt, let them go to the competition.”

    That is exactly what users did, and when developers lost users, they were forced to change again. Other developers decided to fork and go their own way, and they took many of those satisfied users. That is a revolt.

  • cmdr

    Most of the interesting and progressive things happening on the desktop are happening around tiling window managers. I would say most power users have converted to one by now (i3, i3-gaps, dwm, awesomewm, bspwm, etc). Check out reddit.com/r/unixporn for examples. People are building really impressive keyboard driven work flows using very simple tools, each performing one tasks, in a very unix-like way.

  • Jim

    @James

    > “Desktop Linux is barely usable. It’s great for developers who might want a workable GUI, but it’s pretty much uselss for mass market.”

    A Microsoft troll telling me my desktop that I am using, and that he does not, is barely usable. LOL What amaze me is why these Microsoft trolls spend so much time on Linux Sites, trying to convince Linux users Linux won’t work. Why the fascination with Linux for these people? I spend zero time on Microsoft site trashing Windows.

  • Bo Ek

    Sway, Plasma, Gnome .Enlightenment, I hope Lxqt, Mate and Budgie will work with Wayland in the future. My revolt is using Wayland.

Leave a Reply