In the commercial software world, user interfaces are generally designed by one group. Like Microsoft for Windows or Apple for Mac OS. Those desktop environments were designed by one company who did things like user testing and statistical analysis to try and make the desktop they thought would work best. Linux is different. Large groups definitely DO perform user testing and statistical analysis, but one group can also say “Here’s what we want” and, if they have the ability to code it, their idea comes into being. It’s pretty amazing, when you think about it. Linux lets people create what they want. If you don’t like what’s out there, fork it! Or start from scratch! You’re in control!
What’s really interesting to me is the theme that has come out as of late in these debates about usability, and it really comes down to two things: do you want simplicity, or customizability?
Keep it Simple Stupid
Another way to characterize desktop environments (or DEs) that are striving for simplicity is this: get out the way and let the user do their thing.
It’s a pretty compelling argument, really. Why should the user spend time interacting with the OS at all? Why should they have to customize things? Just make it as minimal and intuitive as possible, then let the user actually USE the programs they want.
We have a perfect example of this view taken to its logical conclusion in the form of Chromium OS. For all intents and purposes there IS no desktop, since the desktop consists of a single, full screen program. There’s a few other goodies tacked on, but for the most part that is what you get. You can’t change your desktop background because there is none. You can’t add a panel widget because there is no panel. It’s just … the Chromium browser. Nothing else.
Other DEs have gone in this direction, but not nearly so far as Google did. Gnome released Gnome 3 to mixed reviews, largely because they tried to reinvent the desktop. And by reinvent I mean get rid of most of the desktop. Customization is at a minimum because EVERYTHING is at a minimum. No widgets or applets or any kind of -ets. When 3.0 came out there wasn’t even a plugin framework (still isn’t really, though it is in the pipeline) because, well, there was very little to plug in to. Ubuntu’s Unity struck a very similar chord, trying to keep the OS to a minimum. Reviews for Unity were about as enthusiastic.
The big question is: is this desirable? Is it OK for a certain group to make design decisions that influence a huge number of users? Luckily in Linux, if you don’t like it you don’t have to stick with it, which leads us to the other side of the coin.
I want it that way
Was that a Backstreet Boys reference in the title? I say no, it was a simple statement of fact. Some people want things their way. They don’t want a dumbed-down interface, designed to appeal to the least common denominator. Give me the tools, they shout, and I shall make a desktop that works perfectly for me!
Customizability really is more in line with the Linux mindset. Linux has always been about giving you tools and letting you do what you want. But in an age where Linux is attracting more attention from a less tech-savvy crowd, is that necessarily a good thing?
Take KDE (I’m sure most everyone knew the comparison was coming). In KDE you have a desktop, you have panels (of which you can make as many as you desire), you have widgets, you have … plasmids? I don’t know if that’s a thing in KDE, but plasma has something to do with it.
The result of so many features is a huge amount of customizability. However you use your computer, there’s a means of getting KDE to act that way, and it’ll look good to boot!
The problem with customizability is, as always, complexity. For each option there are settings. For each axis of user interaction there are methods of controlling said axis. To review for this article today I went into KDE and really went wild, and by the end I was amazed at how much I could do with it.
KDE, and other customizeable DEs allow the user to create the perfect working environment for them. Have a little ADD and want everything hidden but your main window? You can do that. Want a panel on each edge of the screen and widget all across the desktop? Yup, you can do that too.
So who is right?
That’s the big question, and, unfortunately, it is one that doesn’t have an answer. But not having an answer does have consequences for the Linux community.
Inconsistency in DEs leads to difficulty for new users. No matter how internally consistent a DE may be, it’s still wildly inconsistent with the dozens of others out there (which only makes sense–they’re built for different reasons). This means that, if I’m a Linux user, I may use Gnome 3 at home, Unity at work and KDE at school, though all are called Linux.
If I’m moving to the Mac world I can go to a Genius bar, or even a friend who owns a Mac, and ask them how to do something and they will have an answer, because it is the same on every Mac. Likewise if I’m new to Windows 7 odds are a co-worker or friend can show me how to do what I want. But if I’m starting to use Gnome 3, and all my Linux friends are die-hard XFCE users, there’s nowhere I can go to get that personal touch.
…And thanks to readers like you
It’s a problem without a true solution, at least in the short term. Sure, we Linux users may have flame wars over which DE is better (even Linus does that), but in the end it’s what we want. We want to have choice.
So is the choice a problem? I truly believe that the vast array of options, especially in terms of DE, is one of the biggest things keeping Linux from mainstream success (on the desktop, at least). But should we even care about mainstream success?
If we were to try for mainstream succes, and make one desktop to rule them all, what would it be like? Where do you fall? In the camp of customizability or simplicity? Or somewhere in between?
What I’d love to see in the comments (besides answers to the above questions) are real life examples. When you started using Linux, where did you start and how did you react to it? What finally felt comfortable? When you introduce family or friends, what is their usual reaction? Which DE do you usually make your pitch with? What are you using now, and why?
About Allen
Allen was born and raised in Oregon, studied Sociology in college and currently works as an operational engineer for a major non-profit organization. In his free time he loves to write novels (if you’re interested, you can read some samples at allenwritesthis.com), play video games, and hang out with his wife (not necessarily in that order).
When it comes to simplicity vs customization, you really don’t have to choose between the two. It’s like the debate on whether hardware devices should be locked down to prevent users from hurting themselves, or wide open so that users can do anything they want with it. The trick is to ship with the device locked down, and add somewhere an option to unlock it, but make it JUST hard enough to find so that anyone who wasn’t looking for it would not unintentionally unlock on accident.
I think of customizations the same way – set up the defaults to be ‘idiot proof’ to make the most users happy out of the box, and set customization options just enough out of reach so that nobody who wasn’t looking for them would ever find them. This way, tech tards don’t get confused and power users don’t feel hindered by somebody else’s design choices. Granted, maybe this won’t work well in EVERY SINGLE CASE, but I think is is a good, general guideline to follow. Personally, I’m tired of using apps that were designed with my grandma in mind, that give me no control whatsoever over the user experience.
Oh, and if you’re going to put the configurations in an ini file or ‘about:config’ screen, PLEASE take the time to document the f**king things, ok? There’s nothing more frustrating than having 5,000 different options and barely an idea about what each one of them does beyond a brief description that may or may not provide sufficient information.
I absolutely agree with you, and even wrote a section in the article originally stating that this is how I thought it should be, but took it out because, even though that’s how I think it should be, I really don’t see many DEs like that. XFCE is probably closest, in my opinion, but still not quite there. You make a really good point, and I’m glad someone brought it up.
I completely agree with everything you said here, but that isn’t the big picture…
It isn’t _just_ a lack of customization that is rubbing some people the wrong way. It is that the nature of many of the abstractions are changing. It is happening slowly to be sure, but it is happening. Menus are being systematically eliminated – hierarchical organization is being eliminated – these are being replaced by more dynamic methods of organizing things that are not exposed by the UI in the same ways – and the most efficient methods of doing these things under the hood will eventually change along with it.
If you look at how things have changed in Gnome 3 for example, it is obvious (at least to me) that there will be a point (that isn’t far off) where recreating the look and feel of a Gnome 2 desktop will simply not be possible – or if it is it will be dramatically less efficient because the information to do so in a familiar way will become harder and harder to squeeze out of the system.
Abstractions change. Some people don’t like it. You can’t always customize your way out of it. I’m not taking a side here, I’m just saying that pervasive customization is not the solution to the problem – the problem is really about the abstractions themselves.
The biggest problem I have with that is that every added option is added complexity. For the user, for an admin, and for the developer. Every small option requires lots of testing in any non-trivial application. I actually prefer the theory of super simple, to the point of elegant, by default, with a basic (or even advanced) plugin API. That seems to be the best way to get around all of the little things everyone finds issue with in anything a developer does.
The problem with ‘elegant’ is that there really is no such thing, because your idea of elegant is somebody else’s idea of ass-backwards thinking. The best you can hope for is a set of defaults that will piss off the least amount of people.
As for built-in features vs plugins, I’m not really sure how offering plugin support makes things any easier, for either the developer OR the user. For the developer, instead of having to do a bunch of testing on built-in features every time you change something, now you’ve got to test a bunch of plugin APIs. And it’s even worse for users, because now instead of having to bring up a dialog box or a config file to change option, they must go searching for a plugin to do their bidding, who’s code may or may not be thoroughly tested, and the plugin may or may not work when a new version of the app comes out.
The choice between “customizable” and “not customizable” is not a binary one, though. Even Gnome 3 and iOS have a settings panel, and even e17 and fvwm put limits on the amount of tweaking that they allow.
IMO, the solution is that when developers want to introduce options in their software, like for any other feature, they must answer the question : “Will a significant portion of my user base use this ?”. Then they should perform telemetry on installed software to see if users actually use the feature after all.
It is sometimes straightforward to answer the question. Everything which is highly dependent on taste, like theming and notification sounds, should be customizable. At other times, it’s more difficult to find an answer. But I think it’s part of the software design homework.
Edited 2011-11-08 07:31 UTC
Even that is not always straightforward. For example, Is it really necessary to add theme support for every single app on the system, instead of leaving themes at the OS level where they belong, so that all apps have a consistent look and feel?
Oh, I was talking about OS-level functionality*, sorry for not pointing it out ! On their side, applications should certainly do their best in order to avoid duplicating it.
* I include GUI toolkits and DEs in my definition of a (complete) desktop OS
Edited 2011-11-09 06:25 UTC
I have been using Linux since the early days where twm was the only option. Since then I have used:
– FVWM (the original one, not the new one with the same name)
– FVWM 2
– AfterStep
– WindowMaker
– Enlightment
– GNOME
– KDE
I remember spending endless hours customizing my environment to feel just right.
Nowadays I use mostly Windows (XP and 7) as my main desktop with Linux mostly on the serve side, and I hardly care about desktop configuration as long as it works and some minimal configuration is available.
And I belong to the ones that think GNOME 3 is actually nice to use.
EDIT: copy-paste error
Edited 2011-11-07 10:09 UTC
I like Gnome 3, too. And I use KDE4 pretty much with no customisations apart from turning off the maximise-by-accidentally-moving-a-window-to-top-of-the-screen malfeature. I think spending hours configuring a desktop shell is something you do when you’re learning a new system and obsess over the new shiny and, i.e. something you do when you’re young — or if you have highly specialised needs and use either huge or many or very small screens.
Being able to do something doesn’t mean you automatically bother doing it.
And for me, that “malfeature” is the most used feature of the new Kwin. With a dual-monitor setup, it’s the most handy way of moving a maximised app between screens (for me). Grab the title bar on screen 1, move app to top of screen 2, release. Also use the “move app to side of screen to “maximise vertically, and use half-screen horizontally” a lot.
Same here. Desktop customization and tinkering is a phase. I did it with Windows (Litestep!), then every window manager under the sun in Linux. Now I’m using a Mac and Windows 7.
It’s a great experience to go through that phase and learn all sorts of things about how software works when you’re young. Eventually you grow out of it and you just want the computer to get out of your way.
Edited 2011-11-07 17:13 UTC
I still use Windowmaker almost exclusively, even though it is old. That’s because it is both simple and customizable. I don’t take endless hours. I have a small set of customizations I do to a new Windowmaker install which take about 10 minutes using the graphical tool (a few preferences, a few keyboard shortcuts), and then I get the fastest, simplest and most usable desktop I have been able to find.
http://www.redhat.com/archives/rhl-devel-list/2008-January/msg00861…
http://www.islinuxaboutchoice.com/
Readers of tech websites have a real bad case of internet disease lately. Don’t like it? Change it. Complaining doesn’t help anyone.
I am so sick of hearing everyone moan about this.
I hate kde.
I loved gnome 2.
I like gnome 3 better.
Whatever you like is fine by me. I just don’t care why you hate gnome 3. You’re wasting everyone’s time. At least mint is using the technology to extend gnome 3 into what they want instead of bitching about it.
I used the crap out of KDE Console from KDE v1.1. It was Perfect. Then things changed.
Moved to GNOME v1.1 and then 1.2 then 1.3 and finally pinnacle of customization 1.4. Downhill from there.
Started using XFCE vs CDE, since I was able to get it to work on OSF/1 v4.01. Also on HP-UX v8.04.
Have used XFCE for my fallback when ever GNOME failed in the 2.x.x variants in Debian SID.
SID wne to GNOME3 recently… guess what? Back to XFCE. some say that XFCE is a MS clone from the 90s.
Considering XFCE had compositing well before most other DEs (Enlightenment also did) and still IMO has better handling of it even now…
I’m using XFCE right at the moment and its able to made like I like it, period.
I ask for a few simple things:
1) Customize-able panels, (width, height, font, transparency, shortcuts, window lists, swallowing, actions… etc)
2) Lots of space with nothing in my way
3) Good “terminal” application
4) Ability to run everything I need/want, I use mainly GTK and some GNOME apps. Only a couple KDE apps. (surprise, I can use both in XFCE without effort)
5) Good support form the mailing list.
The XFCE mailing list sees an awful lot of “Hey your stuff is broken you need to fix it” posts… which go un-responded to, since the list needs to know, what version of XFCE, what app and how it was launched.
To address the menu editing issue everyone brings up… oh my, it was for one stable version where a lot of changes were made. Get over it, its been fixed in newer versions. It was an add-on that wasn’t seen as a stopper.
Anyway… XFCE for the foreseeable future, again.
It looks like Gnome is heading towards a tablet world.
There have been voices in the Linux world for a more consistent design approach. Not only in terms of UI, but also on the API level. While I use a lot of GPL software myself, More than often, I look at it as a means, not as an end. GNU/Linux fits the description “enabler of product” more than it resembles an actual product.
Maybe thats why Linux as such never got a real foothold in the desktop space, I don’t know.
I think it was around Ubuntu 9.10 that GNOME really got to be usable enough. I am sure it was usable before this, but Ubuntu customizations helped in a bit.
I personally think that GNOME should have taken the evolution path, ported to GTK3, sporting new theme, new icons and keeping the same aesthetics. It was good enough for the too stupid, and good enough for the power users. It was not overwhelming like KDE, nor unpolished like XFCE.
All there is now is MATE – and all it needs is a call out support from the big ones – Linus Torvalds, Eric Raymond, etc. That is, those unhappy with the state of GNOME 3, Unity or KDE. Simply landing in XFCE does not makes things better, because many people don’t like it. Why not embrace MATE and make the new real “GNOME 3” out of it?
I can’t code on that level, but I am sure these big fishes can call out, support it and exercise their influence on the community. A new GNOME 2.
Now with GNOME Shell, I can’t make use of it since I open too many applications and windows, and usually I can’t see what’s going on in the background. Perhaps one may like it since all he does is browse the web, check the email, and enjoy the smooth windows transitions done by clutter. But my brain will have an kind of apnea as soon as I get many windows open. Suddendly I don’t know where I am, or what I am doing – this is the feel of GNOME Shell.
We all knew that sh1t was going to be thrown against the spinning fan. By the early screenshots of GNOME 3, I could tell – this is going to be really fuckep up.
Come on, big fishes, you thou excels coding, help MATE become the new GNOME 3.
Long live, MATE.
PS: I don’t think Linux Mint will be much better with its extensions, although it’s a better alternative for GNOME Shell.
There still is a fallback mode i GNOME3 which closely mimics GNOME2. If MATTE people were less ignorant they would better go improve it rather develop a clone for those who are unable to locate “Forced Fallback mode” button.
The truth is that the fallback mode has nothing to do with GNOME 2 because it’s not GNOME 2. There are no applets and things don’t work the way they used to, not even closely. Besides, with news that GNOME 3 can now be run in software mode and doesn’t require any kind of special hardware, the fallback mode will simply disappear.
What’s really interesting is the work that the Linux Mint people are doing on top of GNOME 3. That’s the way forward.
I am sick and tired of people saying “just switch to fallback mode”. Why don’t you add to your comment that fallback mode is a transitional hack? It is not going to be supported. It’s just there to help transition. It will vanish, faster than GNOME 2 itself. You are forgetting this. Talk like fallback mode like it was a saviour, but it’s not. It’s just there to help the thing run on low-end video chip.
The experience as far as I am concerned, using fallback mode, is terrible. It’s not configurable at all. It’s far from a GNOME 2 experience. You can’t possibly have been using this, can you? I doubt.
But next time you mention fallback mode, please, attach the line [which is not going to be supported for very long, and it is a transitional hack, apart from the fact that is going to be removed in the future of GNOME 3 releases.]
You’re nothing but a Gnome 3 troll who’s never actually use the utterly worthless fallback mode. It actually acts nothing like Gnome2.
it’s a useless piece of crap just like the rest of Gnome 3
Where does all of this “GNOME provides minimum configurability” come from?
GNOME has GSettings framework containing a lot of configure options, that can be tuned in dconf or dconf-editor. For those, tight on time or on brain power there is a GNOME Tweak Tool.
AFAIR, KDE provides the same amount of configurability. The difference is n presentation:[list=1][*]KDE gives You all the configuration options applicable in context (don’t know, is it still true for KDE4, but in KDE 3 proxy settings were included into Konqueror configuration dialog though were accessible separately in KCC).
[*]GNOME tries to minimize the amount of configuration options in context to leave just what may normally be reconfigured several times. All the rest goes to System Settings UI or even with no separate UI.[/list]
Note, that the same approach as in GNOME is taken for Windows with its infamous registry. But Microsoft has lowere the usability bar in extra configuration – You don’t even have the description of options in Registry Editor.
AFAIR, the truly limited configurability is on Mac. And there it is delivered as a feature. Failing to see the difference between OSX and GNOME is a reason of failure in GNOME3 adoption.
Ok so its now dconf and dconf-editor. What about/happened to gconftool/gconftool-2, gconf-editor, gconf-merge-tree or gconf-schemas? gtweakui (or is that dtweakui)?
Wouldn’t it be easier to have the config tools built to be more friendly? Come on… this is an alternative to Fallback-session? Which is going away in successive iterations of GNOME 3 and really hasn’t *ANY* way to be fixable to the way it worked in GNOME 2.
If GNOME want to get rid of customization, they do why use GTK+ at all? Lets go back to developing display drivers especially for GNOME and remove X and Wayland all the way. GNOME seem to be re-doing everything else with the sever case of “Not Invented Here” syndrome.
In fact, lets just make a GNOME Kernel and GNOME system libs… Heck lets just make everything from the bootstrap all the way.
Let make the OS GNOME specific and all the libs and compilers for GNOME only… lets go back and refactor everything to rely on GNOME since that is the only way going forward, it appears.
Simplicity, “Sane Defaults” and removal of choice. Perfect, now lets get rid of all the Car Manufacturers except GM and Ford (since there has to be choice) and give you three choices from them: Sedan, Light Truck and Mini Van. No need for anything else… why let the consumer have any choice when they don’t need it.
Oh, lets extend that to McDonald’s and Burger King ONLY abolish all the other distracting restaurants. hmm… what abut clothing manufacturers… homes designs?
Wouldn’t it just be easier if we just had one cell phone to pick from. Just one kind of soft drink. Oh, why not just one kind of bed… or carpet.
Yes, let us brand them all GNOME!
Just an FYI, I’m of the opinion the Gnome v1.4 was probably the best in customization. They allowed you, without having to resort to arcane registry editing tools, to set what you wanted, how you wanted it to work and to not have to use third party tools like devilspie to get the way you wanted to work.
Sorry, but the whole GNOME Development Team is going after the wrong set of targets. No not the wrong set of people… just the wrong set of targets for these people. Nobody likes being forced to eat the same dogfood everyone else is eating. Hence there are about 25,000 choices in automobiles, about as many choices in Motorcycles and well more than that in Food choices and clothing and… well mostly you name it.
GNOME3 is by far a good attempt at making things more easily usable, but they continue to remove changeability, for the sake of “SANE DEFAULTS” which will drive and has driven people away from it. Sure “good riddance; we wanted you not”… well when its whole swaths of people (Ubuntu, Linux Mint and others) its not a good thing to just ignore. Though they will continue to.
In all, GNOME is a great product and is well done. They do have some very “divine” ways of pissing on the userbase.
Is that it still uses X. Despite improvements, using Linux is much the same today as it was in 1998. It’s not a good desktop.
Please give me one single example that makes you say that the problem is X and why that wouldn’t happen if we dropped X. Just one.
dropping X would get people away from the alphabet soup of toolkits and libraries and get us out of the quiltwork mess that is the X desktop.
We’ve had this argument before though. I know what the arguments for sticking with X or moving to a proprietary windowing system are.
I’ll stand by the claim that, to me, the Linux desktop doesn’t look a whole lot different than it did 15 years ago.
The Linux desktop doesn’t look that different than Windows 7, or OS/2 2.x or Mac OS classic or AmigaOS.
It is called the ‘Desktop Metaphor’ for a reason. Besides that – xorg isn’t the real culprit behind the gazillion toolkits found in software today (proprietary and FLOSS alike).
I’ll go as far as claiming the X-server is a Good Thing. The large number of toolkits have nothing to do with the X-server or any particular implementation of it.
That’s simply untrue.
The Linux desktop has changed radically, even with X still being at the core. Heck, let’s take X as an example: you now have things such as automatic hardware configuration. Or the window manager: most GUI applications include a desktop launcher. Or UI consistency: most users will get away with using gtk+ or Qt applications. And those examples ignore an abundance of under-the-hood changes, from the kernel up, that affect usability.
Saying that the Linux desktop hasn’t really changed because X is still there is kinda like saying that Windows hasn’t changed since the registry is still there. Both examples are a critical parts of an overall system that do have their failings. Yet those failings are not critical and they also have benefits. For the most part, the people who spend an inordinate amount of time criticizing them are doing so because they simply need something to criticize.
Edited 2011-11-07 15:30 UTC
Exactly so.
Here is a video of someone going completely overboard using a recent version of KDE4 (KDE SC 4.7.2, running under X)
http://www.youtube.com/watch?v=5AIgyGmHH50
You couldn’t do that 15 years ago.
Now you don’t have to go overboard and do things as complex as this video, but that doesn’t mean that there should be no ability to do it for people who do want, and appreciate, this kind of power and flexibility.
Just because Windows can’t do it doesn’t mean that no-one wants to do it.
Edited 2011-11-08 03:52 UTC
Exactly. The opposite is true too: just because you can do it under Windows doesn’t mean that you want to do it under Linux.
Linux gives you the choice of running anything from a console to the glitzy KDE environment demoed in the video. X itself will rarely be a deciding factor. Indeed, the DE may not even factor since non-technical users may simply choose a distro and leave it at that. (Granted, non-technical Linux users are rare. But they do exist!)
Circa 1996
http://www.kde.org/screenshots/images/medium/matthiase1.jpg
http://en.wikipedia.org/wiki/File:KDE_1.0.jpg
Latest
http://en.wikipedia.org/wiki/File:KDE_4.png
http://kde.org/announcements/4.7/plasma.php
http://kde.org/announcements/4.7/applications.php
The latest KDE4 desktop can run under Wayland as well as X.
http://en.wikipedia.org/wiki/Wayland_%28display_server_protocol…
“KWin, the KDE window manager, added support for OpenGL ES output. It shipped with KDE SC 4.7. So far KWin has received its initial port to Wayland. In January 2012 KDE hopes to support Wayland under X with release KDE SC 4.8 and to run directly on Wayland in summer 2012 in the KDE SC 4.9 release.”
Wayland isn’t proprietary.
Desktop users don’t see, or care about, the list of or thw exact spellings of the names of the desktop rendering libraries.
Edited 2011-11-08 03:09 UTC
The lack of X hasn’t prevented a patchwork quilt of GUI toolkits on Windows. Even MS’ own applications use multiple GUI toolkits. Just compare the looks of all the apps that come with Windows XP, Windows Vista, and Windows 7, with the looks of the various releases of Office.
15 years ago, Motif was the dominant (only?) X GUI toolkit. KDE 1.0 was barely there. GNOME 1.0 was barely there. CDE was probably the only real DE around. And you think that looks the same as a GNOME 3/KDE4/XFce4 desktop?
I don’t see why things are that different in Linux. Canonical, Red Hat and SuSE, among others, do user testing and statistical analysis of various DEs, for example GNOME and Unity.
Edited 2011-11-07 12:58 UTC
That’s the question, and this is one of the first text that I read and really states my opinion of Linux versus Windows/MacOS. To me Linux is the freedom of choice, at least when it comes to the desktop. But talking to many end-users who are on Windows/MacOS, their main response is that they do not want to re-learn their graphic environment. I was always using the car as an argument: What would people say if they only had brand names of car to choose ? And once their car is chosen, there is only one gas they can put in their car ? And their engine is locked, so that only authorized servicemen can work in it ? Would they accept this ? This is what happens with Windows/MacOS. Some people accept to lock themselves in a system, others want freedom. Unfortunately, in the world of desktops, most users prefer to be locked in. I prefer freedom.
There seems to be lots of discussions about simple desktops suited to new users vs configurable desktops. Personally I think its a bit of a red herring, Gnome and KDE are both usable by newcomers. My wife and kids picked up KDE, often cited as a desktop too complicated due to its configurability, without any help. In fact they enjoy the ability to change the way things behaved not just the way they looked.
For myself I’ve been using Linux/BSD since there was only twm; I usually use KDE and find the constraints of Windows or OSX with their one way of doing things restrictive and frustrating.
Edited 2011-11-07 13:24 UTC
You could be right there. I recently went through a re-install on our home computer and installed Gnome, Unity and KDE on it and gave my wife (someone who isn’t particularly linux savvy) the choice of which DE she liked most after giving demonstrations of each. She chose KDE.
But you still need some configurability, not a lot but enough to please most users.
Most of the discussions and articles i see on this topic are like this, very black or white regarding the issue. In reality you need both.
What’s wrong with having both?
KDE it pretty close to that. I feel like I have full control over my desktop and at the same time the defaults work pretty well for everyone. I set different accounts for my friends if they ever come home and need to check anything and all of them feel comfortable with the default configuration of KDE.
And for those(like me) who need control? You can alter practically everything: you don’t want taskbar? just get rid of it. You want the panel on the right monitor on the right side? Set it! You don’t like maximizing on double click? change the behaviour… and that’s a never ending story of things you can change
Exactly! I want simplicity, but I want simplicity my way. I’m a great fan of fluxbox but I think KDE looks a bit better, so I’ve removed all functionality I don’t need and made it look like my fluxbox desktop (but with prettier eye-candy). All I have is a taskbar (with tray and clock) and a few keyboard shortcuts. I did have to replace the ALT-F2 launcher with fluxbox’s though, KDE’s launcher is completely useless to me without tab completion. The result is simplicity for me.
If you do customisation well, it’s perfectly possible to build a sane default configuration for your target group (i.e. let the distros do the job). KDE does it right, although I think Gnome3 took an interesting path as well with their javascript customisation (see http://blog.linuxmint.com/?p=1851). Let’s give them some time to get things right, Rome wasn’t built in one day, nor were Gnome2 and KDE3/4.
Edited 2011-11-07 17:05 UTC
In light of Gnome 3 coming to Debian testing in some foreseeable future, I switched to KDE, and pretty happy with it. It’s very customizable and has nice design.
Edited 2011-11-07 16:41 UTC
First of all, options are necessary, people saying otherwise are simply arrogant jerks who only care about themselves. Users are different, their purposes and scale of problems they are working on are also different – nothing a programmer or a UI designer can do about. Decisions whether to switch between applications or windows, remove empty workspaces or not, have a global menu or not, etc. belong to users. If the options are gone there is always a nuclear solution – switch the whole desktop environment.
OTOH, options are just what they are – options. KDE4 went wild with asking users to actually design the desktop from pieces. It’s like building e.g. a table lamp from Lego. Yes, it can be done, it can be a lot of fun. But the final result is kind of clunky and most people don’t really care – the just want the result.
Still the worst in the UI design are poor abstractions. The abstractions that are convoluted, difficult to use, or don’t match the user intuition or expectations. That’s the problem with Gnome Shell. Personally I think the worst decision was making its UI modal (like in Vi, except it is totally unnecessary, and it is animated to make the mode switching slower). No one ever has asked for it, it didn’t solve any problems and only managed to alienate a large fraction of Gnome users. There is nothing wrong in having a special mode for complex mouse-oriented Expose-like effects or a dashboard but putting most of the basic desktop tools in it (taskbar, workspace switcher) is like forcing Vim users yet again to switch modes and use “hjkl” keys to move the cursor around. It looks like Mint guys understand the problem and have resources to change it so I’m looking forward to seeing their fork.
After years of trying out the linux desktop I finally made the complete switch 2 weeks ago,there are some applications that I was using already on windows so there where no problem using them in linux and there are others that didn’t existed but there was an alternative application for it, here is the list of those applications and its linux alternatives I’m now using:
Windows – Linux
Windows 7 – Ubuntu with GNOME-Shell
Virtual box – Virtual box
IE9 – Firefox
VNC Viewer – XVnc4Viewer
Terminal Server Client – rdesktop
PeaZip – File Archiver
Windows Media Player – Totem / RhythBox
Windows Live Messenger – Empathy
Firebird SQL – Firebird SQL
FlameRobin – FlameRobin
Skype – Skype
What is crucial to me is hardware detection, if the distro detects all your hardware then 80% of the migration work is done.
Edited 2011-11-07 18:01 UTC
openbox or xfce if you really need all the other functionality.personally apart from a few widgets to launch the most used apps id rather not think about the desktop at all.cuts through all the time that gets sucked into playing around with settings screenshot http://min.us/mA6TPtTzx
The more I read the arguments – pros and cons – of the direction that GNOME (with its GNOME Shell) and Unity both have taken, trying to shove a tablet/smartphone UI on to desktop users, the more I think that Aaron Seigo and friends were right when designing Plasma and that ultimately KDE shall reap the benefits further down the road.
Unlike GS and Unity, Plasma was designed from the ground up to be flexible and it shows; it can be easily tailored for different use cases: Desktop (Workspace type Desktop), Netbook (Workspace type Netbook) and Tablet/Smartphone (Plasma Active) all sharing the same codebase and hence the same extensions (plasmoids such as weather applets, twitter/identi.ca utilities, etc.) can be used on all cases. KDE not only does not try to shove an alien metaphor on to desktop users but from all the major DEs is the only one that still allows one to remain on his/her comfort zone, with taskbars, applets, etc. if that is what floats his/her boat.
Want a desktop panel on the left? Done. Want a global menu bar on top that works somewhat like OSX’s? Done (in fact, this has been an option since the late versions of the KDE 2.X series, if memory serves me right). Do not want a panel at all and just a right-click on the desktop to bring the application launcher? Done. Focus follow mouse? Done and done.
Ironically, KDE is also ahead of the pack with the ongoing work to adapt most of its mainstream applications to be “touch-friendly” to be used on a tablet while both GS and Unity, as touch-friendly as the DEs themselves might turn out to be, expect the user to cope with what is basically a WIMP interface (with menus and what not) which kind of defeats the point of having a touch-friendly DE in the first place.
It also strikes me that both GS and Unity take an aggressive stance towards multitasking, making it so hard to multitask that it actually frustrates the hell out of intermediate to power users. Don’t want to take my word for it? Carla Schroder and several other high profile Linux writers have gone on record reporting pretty much the same complaint.
Of course such flexibility comes with some complexity but as correctly pointed out by a previous poster, it does not automatically means that the user *has* to fiddle with obscure settings as long as the default settings are sane. I have yet to set up different activities on my desktop for instance but I am glad that the option is there for those that want it and even for myself in case a see the need for it in the future.
As a matter of a fact, changing the desktop to Desktop View and setting up a rotating wallpaper every hour or so is about the extent of the changes that I made to make it comfortable for me and it took about 10 minutes to set up.
And the fact that the entire DE is flexible allows distros to customize it enough to fit their userbase needs: MEPIS presents by default a somewhat conservative desktop whereas OpenSUSE’s offering is highly polished and corporative-y and Mandriva’s looks all shiny and gorgeous.
I’ll concede that KDE can be somewhat confusing at times and that some design decisions still do not make sense EVEN to die-hard fans like me (such as the separate UIs for Plasma and desktop themes, the very small buttons, etc) and that both Plasma and KWin are in a dire need of fixes for some glitches and optimizations here and there but I’d still argue that it is highly usable as it is (have two young kids on the house to prove it) and that all this flexibility, despite its price, is better for the end user.
DISCLAIMER: I’m a power-user, developer and user of environmental models. Most of the apps I use are not found in any distro.
If I don’t have all of the following
* Multiple workspaces
* Multi-tasking
* Easily customized application menu
* Focus follows mouse
* Virtual desktops larger than the physical screen
then it is a showstopper, and if you refuse me those capabilities, then to Hell with you!
I think this is key. A user doesn’t have to use the full power and flexibility of the desktop software, but if that power and flexibility isn’t provided at all then it can’t be used by anybody.
IMO, one has to have that power and flexibility in order to set up a desktop that gets out of your way. If you can’t set it up how you like it, because you haven’t been given enough configurability, then you are left with someone else’s preferences that you cannot change to your liking. This is bound to get in your way.
awesomewm got both:)
I have used the same setup/.conf file the last couple of years on about 10 linux installs, laptops as well as desktops. Perfect each time, took a little to set i up, but much less than the time I wasted looking for fancy functionality in gnome/kde. Its tiling also, and stable as heck.
awesomewm <3 Pure awesomeness
Awesome until a new version comes out and your old configs don’t work anymore thanks to changes in syntax, that is. I gave up on AwesomeDE (it’s not a WM!!) for that very reason.
A good thought-providing article.
Most commercial GUIs are driven more by the need to produce something new for the next sales cycle than by actual improvements. If Microsoft gives the next version of Windows the same interface as the last, they have no claim to get you to upgrade… nor would many users, who only see the GUI, see any reason to upgrade.
Linux world has its own problems but at least it’s not driven by upgrades for the sake of sales. It’s unfortunate though, that the de facto leader of the Linux camp, Ubuntu, tries to compete with MS using the same philosophy of worthless GUI upgrades.
At least Unity behaves a bit better in Ubuntu 11.10 in 11.04 I switched back to gnome 2.3.x
Gnome 3 doesn’t do it for me at the moment, but I do like the improved file manager Nautilus.
Looking at xubuntu atm.
I also run Windows 7 if I have to but there are those tiny issues which annoy me.
The minimac came with OSX Lion, reasonable but the finder and the dock can be pretty stupid and it doesn’t behave nicely with network drives, ntfs drives and I hate the Apple fanboys.
Useless article and it contributes nothing to a rich and complicated topic. How are people taking seriously a computer science article written by a sociology major?
“That’s the big question, and, unfortunately, it is one that doesn’t have an answer.”
Valuable insight.
Design of user interfaces is not the sole domain of computer scientists. In fact, I would argue that computer scientists are the worst people to design interfaces. I would rather have a sociology major designing user interfaces.
You’re right that you don’t want mathematicians designing human interfaces. You want artsy nerds or nerdy artists. But sociology has nothing to do with it!
No!, That would be far worse as all the text would be in Klingon!
Thanks for pointing that out! I wrote the article and figured that, in a field often dominated by college dropouts, my studies wouldn’t count against me, and I’m pleased that for most people they don’t. I never thought as a sociologist I’d sympathize with “Haterz gotta hate” but I guess that’s the internet for you …
Just to be clear, the article sucks regardless of your education, age, race, creed, gender, sexual identity, profession, height, limb girth or length.
But I do assume that relevant education or experience is better than not.
An interface is designed so it can be used by a novice. Likewise a QWERTY keyboard is designed for “hunt and peck” usage.
The smart thing would be to design an optimum interface and teach users how to use it properly. This would include replacing the normal QWERTY keyboard with a modified stenography or chorded keyboard. A trained stenographer can easily type at up to 300wpm with very high accuracy ~4-5x as high as an expert QWERTY typist.
Chorded, etc. keyboards were tried (for example http://en.wikipedia.org/wiki/Microwriter ); not even self-professed “power users” seem to really adopt them.
Ultimately, it probably boils down to how a staggering majority of people doesn’t type that much (and, similarly, vast majority doesn’t work with the OS but with applications; so they want an OS which mostly doesn’t get in the way …alas, some recent OSS UI refreshes do tend to do that)
This group seems to be a bit missing the point regarding mainstream success:
1) Linux is over 60% of the server market
2) Linux is around 92-97% of the super computing market
3) Linux is around 40% of the embedded OS space something like 3x the size of the next largest player, and the Linux kernel doesn’t even support RTOS
4) Linux is the 2nd most popular virtual operating system on mainframes
5) LInux is doing fantastic in mobile phones and is around 40% of the smart phone market, expanding into every area and is likely to be the dominant OS
______
That is huge mainstream success. That’s it. That’s what it looks like.
In terms of the desktop market, Linux has replaced all the workstation OSes except OSX and OSX is frankly better in too many different ways that Linux can’t complete with.
However… there is a lot of open source in OSX. If you consider what happened to the propriety Unixes servers I think the example is instructive. Closed systems was the initial state of commercial Unixes. The way OpenSource won on the server was a progression:
1) Commercial OSes running commercial applications
2) Commercial OSes running primarily commercial applications with some open source
3) Commercial OSes running primarily open source applications with some commercial.
4) Open source OSes running primarily open source applications with some/no commercial
5) Open sources OSes running open source applications.
In the last decade the windows platform moved from (1) to (2). In this decade it may be moving from (2) to (3).
I’d argue that Android can’t be counted as success of Linux on mobile, because Android doesn’t share that success with the rest of the Linux world at all.
Real mobile Linux systems like Meego, Mer/Nemo and so on are still way behind in maturity and adoption. They however can be counted as Linux on mobile.
Edited 2011-11-08 00:07 UTC
What is the difference between Meego and Android, as far as you are concerned? What is Android not giving back?
Can those differences really be used to categorize one a Linux system and the other not ?
Keep in mind, Linux is a kernel. It is not anything more or less.
This was already discussed here at length:
http://www.osnews.com/thread?493825
Yeah, thanks for that long thread to read which only remotely talked about it seven replies down… You could have just linked to the one semi-relevant comment http://www.osnews.com/thread?493913.
In any case its not true. There is no “linux system” definition that anyone has defined. A more correct terminology would be “unix like system”. Which is what I was getting at. Android is Open but not completely Free software ( as defined by FSF).
I’m not talking just about definition, I’m talking about being selfish or sharing sucess, that’s it. For me success of Linux means sharing it in the first place. Android doesn’t share success, while using the kernel. That’s not “success of Linux” for me.
Edited 2011-11-08 17:11 UTC
“Success of Linux” means sharing what exactly? Other things that are not Linux? What would sharing success look like?
I think the only valid criticism is that the Android branch of the linux kernel, isn’t in in Linus’s kernel.
I think what you’re really getting at is, that Android is not GNU in color or hue. Which is a fine complaint, but there’s no need to throw Linux under the bus. Android is a great Linux success, but a GNU Failure.
I’m not sure what for example Wayland, Xorg or Plasma Active have to do with GNU. May be you know?
I think it’s clearly explained there in the linked thread what can be considered a deficiency in Android from community perspective. GPU drivers and graphical stack being one of the worst issues. It has nothing to do with GNU, but has to do with collaboration. The fact that Android doesn’t share success (and was never even intended to in the first place) is something rather obvious, and strangely you don’t understand it.
Edited 2011-11-09 00:52 UTC
Gnu’s not unix. Its a unix like system and a philosophy. It still embraces the Unix Legacy of a system broken up into different loosely connected layers, but marries that to a philosophy of software freedom and collaboration.
That’s what I would call what you’re calling ” Spirit of Linux”.
I would agree that Android does violate the older UNIX philosophy, it combines too many layers into the Dalvick VM. That kind limits the sharing and collaboration to other members of the Android community and the Linux kernel community. But they’re doing a decent job of sharing, with the notable and mystifying exception of Honeycomb’s missing source.
In that sense I agree, but I wouldn’t call it GNU, it’s too specific, but yes, it’s an example of the project that values and promotes collaboration.
I didn’t say that Android isn’t successful – it is. But it was done disregarding community interests, and that’s the bad side of it.
[q]This group seems to be a bit missing the point regarding mainstream success:
1) Linux is over 60% of the server market
2) Linux is around 92-97% of the super computing market
3) Linux is around 40% of the embedded OS space
4) Linux is the 2nd most popular virtual operating system on mainframLinux LInux is doing fantastic in mobile phones and is around 40% of the smart phone market, expanding into every area and is likely to be the dominant OS[/q[
1, 2, and 4 are the exclusive domain of the IT pro.
Even there your employer may insist on standard configurations of UI and apps because others may have to take your place on very short notice.
3 and 5 are the exclusive domain of the manufacturer and distributer.
The entertainment and navigation system in your car may run Linux, but the UI will be designed by GM or Toyota.
3 and 4 are defined by the device manufacturer.
Simplicity and Customizability?
Do yourself a favour and try Enlightenment, E17, Bodhi Linux.
I was taught to use Gnome. (I later discovered that my instructor is a Gnome fanatic.) I changed to KDE because a KDE app I used did not work well on Gnome. I tried several desktops subsequently and found I was unable to customise them the way I wanted or that they used strange effects I didn’t want.
I later changed to XFCE for several years because KDE kept opening Amarok for several file types for which it was obviously not suited. Changing file associations didn’t help.
I subsequently abandoned XFCE because the developers made it more difficult to customise the apps menu. Their wiki page made no sense; it refers to paths like ~/etc. I got the silent treatment on their mailing list.
I’m back with KDE and I’ll stay with it unless something drastic happens. I’d rather do something with my computer instead of messing about trying to configure the desktop, learn the intricacies of the command line and all the other time-wasting things you can do with Linux.
That is the worst part of XFCE. They’ve cloned a Microsoft design that is more than 10 years old, and still it is difficult to use. The head start wasn’t enough for them.
I’ve said it before and I’ll say it again. Consumers want an exceptional default and the option for customization.
This entire article is premised on a false dilemma. Simplicity and customisability aren’t mutually exclusive properties of UI design.
I have used Ubuntu since the version 7.04, but now with the latest versions I have chifted to KDE.
Unity have problems with multitasking and more advanced stuff. And Gnome 3 have issues to.
(Linux Mint write about thoose in their website for the new Mint 12)
So I only using Pardus 2011.2 at the moment. A very solid and stabile KDE distribution. Not so known but works very good for myself.
http://www.pardus.org.tr/en/
If it was not for the KDE desktop I would have gone back to Windows 7 and stop with Linux permanently.
That reminds me. Simplifying the desktop is a bad idea because it kills a viable niche (power uses) and competes with an enemy that everyone has in their pocket and which cannot be defeated (smart phones).
These desktop boys breaking from the crowd into the wild blue yonder might find themselves hitting the ground hard after the giant real world smashes them with his spiked club