> I personally think that one reason that the Linux kernel has been so successful was the fact that I didn't have a huge vision of where I wanted to force people to go

I agree. Thing is, this just doesn't work when it comes to user interfaces. You end up with a morass of principalities and wanna-be-kings who all have their own different view of what Linux on the desktop should be. This is an area where consistency really matters (to users).

What's more the primary contributors are engineers who are unrepresentative of any kind of mass user base (potential or actual). People who want to be able to customize everything when no one else cares.

Ultimately I think the pseudo-anarchic development process doomed Linux on the desktop but some external factors didn't help, namely OSX. OSX really is "good enough" for anyone wanting a nix desktop/laptop. Sure it has problems but it means the gap between it and Linux is just that much smaller. And with Linux you give up great hardware (to varying degrees) and a snappy and consistent* UI/UX.

The company that has had the most success putting Linux interfaces in people's hands is, of course, Google with Android (disclaimer: I work for Google). And even there the principality problem has created a fragmentation nightmare as different vendors seek to "differentiate" themselves.

Phones and tablets of course don't have the same expectation and technical legacy that desktops do (from X11 on up). Hell, Apple did the same thing with iOS (essentially forking OSX).

Audio is another problem. Peripherally I've had to deal with PulseAudio and it really is a solution looking for a problem. Graphics drivers are another issue. AMD/ATI and nVidia have of course not helped matters here.

I think Canonical had a chance to make a real impact here but they essentially blew it.

I keep seeing claims being made, and mostly by (former?) Linux superusers, that Canonical and Ubuntu are "doomed". Could someone point me to some numbers that show signs of this?

I know when the Unity desktop came out, a lot of people were flaming. I didn't install it on my laptop (I'm still on 10.04 LTS), but I did on a colleague's desktop. From time to time, I have to assist him with code he's writing. During those sessions I have to interact with it and to be honest I'm not pleased, but each time I ask him if he wouldn't be interested in switching to Gnome or maybe installing Mint or something, he keeps saying it's not that bad. I'm sitting next to him and frustrated because I can't find where everything is and how to bring stuff up, but when I ask, he just presses a few keys and whatever we need shows up on screen.

When Miguel de Icaza's post came out last week and I saw claims on the doom of Ubuntu, I started looking for evidence. I couldn't find anything that in fact confirmed this. I found a lot of people who complained about Unity last year when it first came out, but fast forward in 2012 and there are also more and more posts of people going I use to hate it with a passion, but I have to admit, it's now growing on me. Not everything is perfect, but with a bit of work I think they're on to something. Also interesting to me was that I had a sense that the most positive reviews came from people with no prior Linux experience. So much so that I'm getting more and more curious about it and I'm slowly mentally preparing to make a switch to it.

Is it possible that Canonical is actually moving away from Gnome and taking things in-house to avoid exactly the problems that Miguel de Icaza is underlining?

Of course my heavily customized gnome 2 were better suited for me, because every detail was tweaked by myself. Unity I can use out of the box and it is working really fine.

I sell laptop with Ubuntu 12.04 and I have a return rate of 3%, normal return rates are 20 to 30 %. It can't be that bad for normal users.

After grudgingly Googling how to use Unity, I found the steps to list applications and it's absolutely horrible. Here's how:

  Right-click the Ubuntu logo*
  Choose applications
  Click "Filter results"

For now, there's the option of installing gnome-session-fallback. It introduces a "GNOME classic (no effects)" option in LightDM which works well and has the old, functional workspace switcher. Still no F1 for help, but it sits under Applications > Accessories so it's relatively discoverable. The panels can be customised by holding down ALT+Super while right-clicking them (thank you again, Google).

Sadly, Gnome is deprecating the fallback session in favor of a software rasteriser ("llvmpipe") to run Gnome-shell where a hardware rasteriser is unavailable. The next release will also see the lock screen moved into the shell so it'll look pretty, but it better be the most stable release they've ever had since the last 2.x. Sorry to be cynical, but I doubt it.

*: This doesn't work in Unity 2D. Tough.

(Disclosure: satisfied Unity user here)

If you interact with the machine primarily through the keyboard, "search everything" is a plus, not a minus. It means you can easily launch new applications without having to take your hand off the KB and move it over to the mouse. Launching applications is much faster for me in Unity than it was in GNOME 2.x, simply because typing ALT plus the first three characters of the application name (or a generally descriptive word: ALT + "music" finds Rhythmbox, for instance) is faster than rooting around a multi-level menu.

After grudgingly Googling how to use Unity, I found the steps to list applications and it's absolutely horrible

This is because "listing applications" in Unity is a failure case. Users don't want to list applications; they want to launch the one application they were looking for. The point of Unity's launcher is to make digging through a long list of apps to find the one you wanted obsolete.

"listing applications" in Unity is a failure case

My desktop takes the logical next step, before I moved back to gnome-panel I simply dragged the applications I used out of the overlay thing onto the desktop. I keep the icons organised, so they're always in the same place. Anecdotal observations from "normal" people show that they use their Windows Vista/7 desktops much the same way, not even bothering with the start menu except to shut down. They get grumpy whenever anything is different at all in which icons are shown and in what order. Their allegations that I broke something weeks after I accidentally left, say, a copy of unzip on their desktop is of course laughable at best. But the concept of having shortcuts on the desktop, certainly not. Makes me wonder why Unity bothers with a separate desktop at all, it could benefit from the screen real estate when that overlay screen sits at the bottom of the stack as the root window. I loved the netbook-launcher it grew out of, too bad it was discontinued.

Users don't want to list applications; they want to launch the one application they were looking for

The Gnome consensus used to be that users don't care which application they're using, they just want to carry out a certain task. Which is why they all have boring names such as "document viewer" and "web browser". How is a user going to know to find evince and epiphany or firefox? In Unity, search quickly breaks because of this. If you know the name of the application, it can find it. But if you don't, it doesn't map keywords to application names. It would be impossible to maintain appropriate keywords for all applications in Debian (and by extension, Ubuntu) so it doesn't.

Fortunately, all the current desktops run a terminal by pressing CTRL+ALT+T. It's the one failure mode they all support equally well, supports tab completion and isn't limited to applications with a .desktop file. This is what I do for most tasks besides web browsing, so even after all these years of developing desktop environments the main purpose is still to run multiple terminals side by side. If only twm had niceties such as network manager and removable media.

No. I had doubts last year about the wisdom of Canonical starting from scratch with Unity rather than just building on GNOME Do (see http://jasonlefkowitz.net/2011/04/ubuntu-11-04-everything-ol...), but the last couple of releases have put my doubts to rest. Unity is everything GNOME Do is/was, but now with the possibility to extend even further into things like the new HUD (https://wiki.ubuntu.com/Unity/HUD), which does to application menu bars what Unity does to application lists.

They get grumpy whenever anything is different at all in which icons are shown and in what order. Their allegations that I broke something weeks after I accidentally left, say, a copy of unzip on their desktop is of course laughable at best

You seem to be under the impression that the Unity desktop can't have icons/files/launchers/etc. on it. It can, mine does. It's accessible via the usual "Desktop" folder in your home directory.

If you know the name of the application, it can find it. But if you don't, it doesn't map keywords to application names. It would be impossible to maintain appropriate keywords for all applications in Debian (and by extension, Ubuntu) so it doesn't.

You're incorrect here too. Unity does support searching by keywords, and all you need to do to hook into it is create a standard freedesktop.org-style .desktop file for the app (http://standards.freedesktop.org/desktop-entry-spec/latest/a...) and fill in the "Keywords" field inside it. Unity will pick up those keywords and use them when searching in the application lens. Most of the commonly used applications already have the most obvious keywords provided for them.

(Putting the keywords in the desktop entry file means that Ubuntu/Canonical don't have to maintain a master list of keywords for every application; app developers can just provide relevant keywords for their app in the DE file they ship, and users can tune them if needed just by editing the DE file.)

I get the feeling from your complaints that you checked out Unity briefly early in its lifecycle and haven't checked back in on it lately. You should try it again now, you might be pleasantly surprised.

I never got the hang of Do, either. In order to do things with do, you have to learn its vocabulary. Whenever there was something I didn't know how to do with do (or it didn't have a certain plugin), the old way of doing it was still there. This made me reluctant to ever press the button that triggers it. Maybe I'm bad for not investing time into making it work better, perhaps desktop environments are not being treated fairly in this regard. To clarify, maybe it's hard to unlearn the ways of interfaces which came before. I was taught how to browse the web with Netscape and today I'm still bitter about certain changes in Firefox (I know there's seamonkey but it's a second-class citizen nowadays). This doesn't hold true for all user interfaces, but I'm not sure if age plays a significant part (as in, learn UI X before age Y => forever stuck with the concepts of X), prolonged use or a combination of either along with not having any alternatives (thus, not building the required mental abstraction layer to differentiate the underpinnings of various UI concepts). I have only myself to sample and given that people are using user unfriendly (or even hostile) window managers such as evilwm, fluxbox and xmonad, there have to be wildly different "mental operating systems" in any representative sample group.

Gnome usually worked well despite this psychological hellhole, I could introduce people to it and it generally wasn't met with contempt. They knew their users so well that I, for one, was shocked and baffled when at the first command I entered in the preferences of the panel applet Fish, it berated me for trying to make it useful (there's only 1 command which does that).

Most of the commonly used applications already have the most obvious keywords provided for them.

I see, that's great. Sadly this hasn't worked for me, but I'll keep this in mind whenever I'm searching in Unity and something is missing. I hope seeing how it works hasn't skewed the way I use it too much.

app developers can just provide relevant keywords for their app in the DE file they ship

Yes, I hope package maintainers will do this too. This can be useful for a11y and other desktop environments, nice.

I get the feeling from your complaints that you checked out Unity briefly early in its lifecycle and haven't checked back in on it lately.

This is false. Although I have tried Unity in each Ubuntu release starting from 10.10, I've tried hard to work with it (instead of replacing it outright) in 12.04 over the course of months. I turned to Google just to figure out where my applications are for starters, instead of flipping tables. What recently drove me away is the instability, not the glaring usability quirks. At some point, compiz flat out refused to start but compiz 2d still worked (for a while). Compiz stability is very hard to fix as it can even depend on obscure GPU bugs only found in serial number x through y of model z manufactured by {.

However, I did manage to find one improvement over gnome-panel: it's easy to use the keyboard to navigate panel applets/indicators. Just by pressing e.g. Alt+F, you can move (using the arrow keys) all the way to the power button. A lack of pointer input always renders gnome-panel useless.

Still just press the Super key type in "doc" and you will see the "Document-Viewer" which is evince.

You can see most shortcuts with holding down the Super-Key for a few seconds.

I'll try the help whenever I'm in Unity again.

I switched to Xfce and haven't looked back. Everything is snappy, and everything is configurable.

Performance was a bit choppy when I had 4GB ram, and sometimes compiz would freeze my system. However, upgrading to 8GB RAM (costs about $50) totally fixed everything for me. Just upgrade.

I'm sure they could optimize things better, but that would slow down their frantic pace of innovation.

Why? Why should someone upgrade their ram just so the UI works smoothly? We've had smooth UI's for twenty years, even back when 32 MB of ram was the norm - so why should someone upgrade to 8 GB just for the UI? Hell, if I upgrade my ram (I have 8 already, but anway) I'll want it to be to benefit all the applications I like to run and I expect the UI and OS to use as little as possible and still perform smoothly.

My old eee pc was able to run windows xp just fine before I installed Arch with a tiling window manager on it and everything ran well, everything was fast, everything was smooth - for about three years - and then I installed Ubuntu and Unity brought it to a standstill despite not using any different applications than I always used. Even switching workspaces sometimes took 15 seconds or so. Yes, it only has 1 GB of ram, but like I said, I used the same applications that I always did without problems. I struggled with it for a few months and switched back to Arch and haven't had any problems since (though I now use a laptop which does have 8 gigs of ram and rather than using that ram to make unity run smoothly, I'm using it so I can run windows 7 and linux at the same time with VirtualBox).

Maybe it's a config problem. Maybe it's not. Maybe it's sunspots. Who cares? It doesn't JUST WORK and that - more than anything else - is what will doom Desktop Linux.

I run Ubuntu on what essentially amounts to a media desktop. It does work, but that's because I spent hours on researching the simplest, cheapest hardware config that would work seamlessly with Ubuntu without requiring me to compile my own drivers, etc. The fact is that I, even as a power user, really don't want to spend hours messing around with that stuff... so I can't imagine the Average Joe would want to.

Slashdot comments are full of "Works-for-me" replies to real issues with varying levels of "you are lUser" for not being able to make it work.

Actually, back then there was much more reason to doomsay, as there really wasn't a desktop to speak of (CDE on Solaris was far more sophisticated than bare X on Linux), no hardware manufacturers even bothered to provide even closed-source drivers for their hardware, hardware support was miniscule, there was no such thing as auto-detection of hardware, web browsers didn't even exist so apart from a handful of applications with pitiful GUIs by today's standards, every user was forced to interact with the system through the shell. There was no OpenOffice or LibreOffice (ie. no option to use a Word/Excel/PowerPoint alternative on Linux), no web browsers, no desktop environments, no way to configure your system through a GUI. The list goes on and on.

Since those days, Linux has improved by many orders of magnitude. And it keeps getting better every year.

I think most people complaining about Linux these days really don't appreciate how bad and comparatively unusable it was in the old days. Back then you could not even dream of making any kind of comparison between Linux and Windows or MacOS in terms of ease of use for an average non-technical user. Now from an average user's perspective, they're all quite similar, with some minor differences on the periphery.

Sure, Linux still has some problems, but it's not like the competition is without its own problems.

Yes, sometimes a user who bought some obscure peripheral (made by a manufacturer who does not care about Linux compatibility) doesn't bother to check to see if the peripheral was listed as being Linux compatible. So they may have problems getting it to work.

But how do you think the typical Windows user likes using a virus and trojan infested system? Even Mac users are starting to fall victim to this problem. Or how do you think a typical Windows user enjoys having to reinstall the operating system every couple of years because it's slowed to a crawl through Windows-bloat?

"Slashdot comments are full of "Works-for-me" replies to real issues with varying levels of "you are lUser" for not being able to make it work."

Slashdot is not representative of the Linux community as a whole. And my experience in asking questions of Linux users has been quite different from yours. Virtually all Linux users I've encountered anywhere, from usenet to web forums to irc to mailing lists, have been exceedingly helpful. Even on Slashdot, there's quite a lot of help to be found.

And what could be more helpful than volunteering to write whole applications, whole operating systems, and even the documentation to go along with them for free? Countless Linux users have dedicated years and even decades of their life to helping the community by doing this.

The mercenary, closed-source culture that dominates in the Windows and Mac worlds is shamefully lacking compared to the vibrant open-source world that thrives in the Linux community. As a Linux user you are gifted with an entire free operating system, tens of thousands of free applications, hundreds of free device drivers, your choice of free desktops, etc. In the computer world, is this not the ultimate in generosity?

Please stop the "Doom predictions". It won't happen. It does not need to be the MAINSTREAM system to exist.

And that's a shame, because of the wealth of great, free tools (whether it's development tools, photo editing tools, audio editing tools, etc. etc.) in the Linux ecosystem.

I do think this is part of it though they probably wouldn't say it way. They certainly do want a consistent user experience - and they want one that rivals OS X.

I don't understand why people complain about using software they haven't learned how to use yet. It all takes some getting used to for a new user, including OS X. Now I can definitely understand the frustration though of someone who just upgraded their OS only to find they don't know how to use it anymore.

The world is not better off with this, as it merely teaches people how to use MS-Windows and MS-Office. It doesn't go into any detail at all how a computer really works.

you shouldn't have to learn how to use the desktop! I tried out Unity, I really did. I didn't like it and my apps kept getting lost when minimized and I couldn't have lots of small windows open and it just fixed something that wasn't broken for me. Like everyone else, I went to Xfce.

desktops are a tool like any other - the mentality that they should be intuitive or easy leads down the path of eye candy as opposed to functionlity.

"one that rivals OSX" what a joke. Windows XP is a better rival to OSX than Unity.

I assume you're talking about Linux driver issues. But since we're talking about OSX vs Linux here, it's only fair to note that Linux supports a much greater variety of hardware than OSX. Linux also tends to support the very newest hardware more consistently than OSX (perhaps excluding video cards). See eg the laughable specs of Mac Pros.

Software also benchmarks faster on Linux vs OSX on the same hardware. So there is still an opening on Linux for people concerned with performance.

But yes, the Linux desktop is still light-years behind in terms of design, for the reasons you mention.

Entirely true, but I also don't have to look up what is or is not supported by OSX. I just buy an Apple computer. Of course, there's obvious tradeoffs here about having the freedom to buy whatever hardware you want.

http://www.ubuntu.com/certification/

I'll take supporting a limited set of hardware very well over supporting tons of hardware often in a mediocre at best fashion. I'm not just talking about video cards either, over the years I've had "fun" with everything from SATA controllers to sound chips that were 'supported' under linux.

The last computer I purchased I purchased at a bar on my phone while rather intoxicated. The only thing I did was sort newegg's laptop selection by "cheapest first" and bought the cheapest. I don't think I missed out on any "building" time... If what everyone says about Linux hardware support was true, there is no way in hell that should have turned out fine.

Concerning the video cards I think people miss out on the proprietary drivers from Nvidia here which have always worked brilliantly for me since mid 2007 or something. Yes, they are proprietary but so what? As far as I know ATI cards work pretty well too.

Linux sound support works fine as well, it's still mostly ALSA underneath calling the shots which works like a charm. PulseAudio, I admit, usually doesn't. You always have the option to remove it from your system though - or even better: Don't install it in the first place. The actual drivers are in ALSA though, so nothing to complain here either.

CPU and RAM support is a no-brainer with linux. Never had any issues. I actually had a ton more issues with it under Win7. Recently plugged some 32gig additional memory into a workstation (64GB now) and Win7 only accepts 48 of it. Booting into Linux everything works and I've got the full memory capacity at my fingertips.

So I wonder, what is this all about? I've been using Linux on quite a few pieces of hardware and never ran into any serious issues. Yes, I had to screw with the X config a few times but that doesn't file as "not supported", just as "stupid defaults".

So you buy Macbooks because they run OS X and I buy Thinkpads because they run Linux. Where's the difference? (Except that I would also be able to run Linux on other types of laptops if I wanted to).

Option-up/down doesn't replace an actual Page Down Key.

Btw. how are the screens these days? I've read that the resolution is quite low, but do they at least have good contrast / color spectrum / brightness / viewing angle? I'm asking because my father wants to buy one again (low res is actually a plus there).

I have the upgraded 1600x900 screen, so I can't speak to the quality of the lower resolution screen. I've heard it isn't as good, but I will say that for programming, the 1600x900 has been great. (I forgot to mention how much I hate glossy screens as opposed to matte...) From everything I've heard, the upgraded screen is much better.

Ha! Check any recent thread on Mountain Lion updates and you'll find a ton of non-"very well supported" hardware glitches.

I'll repeat myself here. Linux is the natural successor of the technical Unix workstation. It's, therefore, a niche product: built by a highly technical crowd for a highly technical crowd. Only recently efforts were made to make Linux more muggle-friendly and those efforts nearly coincide with the decline of the relevance of the desktop PC.

Many attempts were made to make Linux more user-friendly at the start of the century (and ever since).

Unfortunately, at that time, the lack of desktop applications was a huge problem for the average user. Today, many people read e-mail, collaborate on documents and engage in a large part of their social lives through a browser. That was not the case in the early 2000s. Having an easy to use (or Windows-like) file manager, application launcher or even package manager was not enough.

Perhaps it's unfair to give credit to Canonical for making Ubuntu grandma-friendly when a lot of its ease of use comes from applications like LibreOffice, Thunderbird and Firefox (and Gmail, Gdocs and Facebook).

Curious to hear you elaborate on what they blew, and how. I don't use Ubuntu myself (prefer straight Debian), but I've always respected what they're trying to do (and to some extent, what they have already accomplished).

I wouldn't want to have a machine again, where any of these things won't work:

- Set volume per source/application (More flash video for the presentation, some mp3 still in the background)

- Move applications between devices (Wife complains about the sound of my video call/video/game? Move that thing to the USB (!! No jack that directly does what you want anyway) headphones)

- Normalize output so that nothing blows up my headphones when I switch between different audio sources

Apart from that I did use the 'stream over the network' feature a couple of times in the past and while it wasn't something I'd need on a regular basis it worked fine for me.

Long story short: Pulseaudio solves problems for me. It totally might not do that for you, but please don't project.

I have begun to wonder that, given the evident superiority of Linux in mobile, given the lack of GNU within it, and given the central control of Linux mobile development by Google and Amazon, then to what degree has GNU and its development traditions held back Linux desktop adoption?

Thoughts?

The UI layers (GNOME/KDE/Canonical) may be relevant. Linux desktop doesn't have a giant company funding massive ad campaigns and retail outlets and spit-and-polish and OEM integration, and all that is needed to launch a successful product for the mass market.

Linux desktop is very successful in the developer workstation and server markets, where the users are more tech savvy.

When I go to settings in Google Hangout, it asks me which device I want to use. ALSA presents the interfaces. I choose whatever combination I want -- mic from the camera, headphone out through the mobo -- and it works.

http://xkcd.com/619/

The real future of the Linux desktop and mobile is Wayland, and Android it seems will stay apart forever, going its own path.

The real future is what users like best, not what developers think that users might like. I personally don't like the Wayland approach and I think that competition is fine. Let the "market" decide what's best.

I.e. there are practically no manufacturers which provide closed Linux drivers or open specs (so open source drivers could be created). They mostly only care to provide Android drivers. Hopefully Jolla and Plasma Active will help to break through somehow.

This situation is caused by Android's historic roots, since it started as a proprietary project, and didn't take in account any interests of the Linux community. The fact that it was open sourced later didn't really change anything - it's de facto completely separate from the Linux desktop as well as from the mobile Linux varieties which share the effort with the desktop distros. Wayland was created with collaboration in mind. Android architecture - was not.

I would like to think that if Android was available on both desktops and mobiles, we would have been happily playing Dark Souls using Steam-on-Linux by now.

Also, Android is rather narrow in its capabilities comparing to normative Linux. No need to cripple the desktop like that.

It's going to be other way around - normative Linux will start competing with Android in the mobile sphere.

These mobile devices could grow some HDMI and USB ports in a few iterations and becoming stationary.

Tada - there's your Linux desktop.

The Linux desktop has failed to become a mainstream 'consumer' OS. Meritocracy has nothing to do with it. Microsoft's monopoly is just unbreakable here. OS X is less than 10% of the market. Linux less than 1%. I believe the Linux desktop is usable in the retail space, because I installed gOS on a guest computer and several novices (including my mom) used it for email, web, music and word processing.

My linux desktop progression has been X11R6 -> KDE -> Mandrake -> Ubuntu gnome -> Mint -> latest Ubuntu HATED IT! -> Xfce crunchbang -> (plan to) Mint Xfce

While true, look around at younger companies, OS X has taken over a very large segment of the web dev crowd and I don't see any signs of that changing, especially with the almost mandatory nature of using OSX for iOS development.

OSX got its market share off spoiled 20 year old college kids getting their parents to buy them overpriced trendy hardware. I have a 15 year old and 13 year cousin who each have a macbook because its hip.

But besides the hip factor, OSX succeeds because it is the default on Apple products, where Windows succeeds because it is the default on everything else. When average Joe buys a pc, they go to best buy, and every single device has Windows on it, so they use that. They don't know any better and would never fathom installing an OS, and if you show an alternative to them it breaks their minds because it is universe breaking that Windows isn't the computer.

Sigh. Stop stating your opinion as fact. Personally, I use OSX because it's a *nix OS that actually has a coherent UI, and it's paired with the best hardware available right now. Saying that it's popular because it's trendy is wildly misguided- how do you explain all the developers that use it?

Trendy, popular, whatever... The original comment is spot on. Products becomes trendy and popular mostly because of great marketing. The reason you prefer OSX (or almost anything else in your life) is because you've been conditioned to. We make decisions based on a complex cocktail of facts and feelings, most of them external to our own thinking and, therefore, susceptible to clever manipulation.

Every BMW owner buys an "ultimate driving machine" (with an automatic transmission) and most OSX users believe they're enjoying "the best UI/UX" (and obediently squeeze their life into ~500 vertical pixels leaving the rest to the Dock and global menu).

It does not make sense to look for a technical explanation for why "Linux desktop" is not popular. Technology has nothing to do with popularity. Moreover, Linux Desktop is indeed widely popular among people who're pre-conditioned to like things it offers.

Here's an anecdote: I have introduced younger relatives to computers in mid-2000s using only Ubuntu. After years of daily Gnome 2.x use they find both OSX and Windows incredibly "retarded" and hard to use.

Don't kid yourself thinking we're far ahead of Pavlov dogs.

No, it's because I've used the other alternatives and made a rational decision. I tried Ubuntu, it didn't work with the hardware in my laptop, the trackpad was useless and it didn't run a number of programs I require (the Creative Suite, for one). I would happily use Windows, but the hardware Apple creates (in my case, the Air) is largely unparalleled in terms of build quality and weight. So, OSX it is.

These are all measurable, quantifiable things. While I can't disagree that marketing affects people, to suggest that the reason everyone likes something is because of conditioning is a gross generalisation.

I've used enough different operating systems in all kinds of configurations over the last 20+ years to decide like a grown-up what I'm going to spend a considerable amount of time using.. not because other people think it's cool, but because it fits my budget and works for me. My recommendation to others is: Try stuff and use what works for you.

If you needed the form-factor of an Air and Creative Suite there really was no decision. Rational or otherwise. There is certainly no conditioning, playing with Ubuntu was pointless.

Most of the people you are lecturing actually use 2 or 3 of Windows/OSX/Linux on a regular basis, and so have informed experience backing their preferences.

>> I have introduced younger relatives to computers in mid-2000s using only Ubuntu. After years of daily Gnome 2.x use they find both OSX and Windows incredibly "retarded" and hard to us

You are refuting your own arguments. eg How can your nephews like Linux if they have not had marketing to condition them?

I'm glad it isn't your opinion, because you'd be wrong again. Mac owners, by age, are actually older on average than PC/Windows users.

http://www.metafacts.com/pages/media/tupan06_announce_061129...

As an embedded systems developer, who works in a company filled with EEs using OSX for their personal machines, I can say that what keeps me buying macs is that

* Like Linux, OSX is posix and has a working shell and scripting environment, and

* Like Windows, it has a pixel-perfect window manager and a tastefully designed UI that doesn't make my eyes bleed.

And the meat of my complaints have nothing to do with UI, but the endless finicky regressions, typically having to do with drivers and hardware integrations.

Anecdote: my friend and co-worker has an allegedly 'good' piece of Lenovo Thinkpad hardware. He still managed to have a problem on a video driver upgrade on 32-bit (he then switched to 64-bit, where things went better -- did I mention Ubuntu's home page still suggests 32-bit Linux?) and hasn't gotten around to convincing it to talk to an external monitor. He has more patience than me -- mostly out of philosophical stubbornness -- although I suppose we both are stubborn in a way since I'm still buying desktops at every location I work frequently in 2012, and regressed to a Macintosh to fill in the gaps when moving about.

It is a disaster.

My MBP is 5 years old and still runs OK. (It could use battery #3 and an SSD replacement, though.)

I used linux for 5 years as a desktop OS, skipped in between windows occasionally. Just recently switched to a mac. It has nothing to do with being hip. OS X is an OS which runs all the *nix tools I need, has a reasonable well thought out UI and above all it __just works__. Its what I wish linux on a desktop could have been but never was and I doubt ever will be.

My laptops are just workstations to do day-to-day life stuff and a portal into my virtual computer world, where I'm working with remote displays, browser windows, command lines all over, and all that jazz. If it's anything remotely server related that needs to be on all the time, I'm likely not running it on my laptop anyway - it'll be on a linux box somewhere.

Or, if I'm on the move and it's really important to get a project done, a local VM on the little macbook.

Please just stop now. Take untog's advice and stop making arguments based on your anecdotal experience.

I did like being able to open a shell and kill -9 a hung app.

But in the end it just felt slow.

Why this step? Most people go in the reverse direction.

Why should Linux be doomed (or killed as wired puts it) just because Windows & MacOS have more idiots using it? For a desktop there are certain design choices to be made. If one wants to target users, then one should use sensible defaults and try to hide complexity. If one wants to target hackers, then one should have a lot of easily accessible APIs, preferable accessible via a CLI. That means exposing complexity. For the two groups consistency in a UI does mean two different things, users want a "intuitive" UI, whatever that means. Hackers want a UI which is based on principles, such that one can deduce the proper operation, even if it is complicated and takes some time to learn in the beginning. The list could go on, but in short: I have different needs on my desktop than the average user. And if Linux would become a sensible desktop OS for the average user, then I would likely move to BSD (or Plan9 or whatever).

[1] Perhaps Ubuntu's default install?

I am absolutely sure that Microsoft and Apple are fighting against desktop Linux any way they can as it takes away market share and any amount of market share is fought hard about these days.

In my opinion Apple and Microsoft favoured (paid off) bloggers/(ex-)developers, and opinion writers do their best to spread FUD and anti Linux propaganda all the time and everywhere. I think that a large part of the Ubuntu Unity FUD for example comes from there.

Look at any discussion remotely about Linux or Ubuntu and you will see random comments about how bad Unity is, even here in this thread. Even if the topic has nothing to do with Unity. It's not a coincidence.

I am a developer using Unity and find it the greatest thing that ever happened to Linux, and the non-developer people I support on their Ubuntu Unity machines just love it, they wouldn't want to change anything.

All I see is that the Linux desktop now is getting really competitive (yes I mean Ubuntu 12.04) and that's why we hear all this FUD now. With "competitive" I mean usability, security, ease of use and style. The market share is low, yes, but it's slowly growing and it will continue to do so.

The Linux desktop being dead is just a laughable idea when looking at the facts.

Windows is the default. Microsoft got in bed with hardware producers and computer retailors and got Windows on every product on every shelf. For 2 decades you went to buy a PC and you were buying a PC with Windows pre-installed. The extreme majority of PC users and buyers have no comprehension of how to build your own PC, what an OS is, or that they can even use alternative operating systems. At this point, Windows is the computer to a tremendous majority of desktop OS consumers.

OSX is prosperous because it is the Mac default in the same way Windows was prosperous because it was the everything default. The average consumer does not conceptualize or think about how the software is independent of the hardware, what they see and get is what they stick with 99% of the time. The other 1% is Linux / BSD / everything else.

The day of the Linux Desktop is the day a laptop running any flavor of a linux distro sits in Best Buy next to a Windows laptop and costs $50 less because it doesn't have the Windows license fees attached. Everything else is just window dressing of that issue, which is in plainest terms that Apple and Microsoft have a profit motive in getting their OS on your machine, and go out of their ways to see hardware with the OS preinstalled so Joe Shmoe uses it, whereas Linux distributions have no financial incentive or warchest of influence to push Linux as the default. And the default is what matters.

Well... It's not Best Buy, but, still, it's something:

http://imgur.com/tzv74

"None of this recent discussion is about why the average computer user doesn't use Linux or won't try it -- that topic has long since been beaten to death. It's about why Linux has the inferior desktop experience, and how OSX was able to steal away so many of the developers and users that, in a perfect world, should have been happiest on Linux."

All I know is, that I'm glad to see differences between OSes regarding not only kernels and the way things happen but also regarding UX and interfaces (where Linux is by far the most flexible, because there's more than one desktop environment - a further differentiation) and I don't want them to be the same either.

I think the bottom line is despite how much people want one, there isn't just one reason for its unpopularity, there are dozens. It isn't good enough on multiple fronts--paper cuts abound next to deep slashes. All in the shadow of two compelling choices in modern Windows and OSX.

They have their own annoyances as well, but you can expect drivers, audio to work, no major regressions--nothing you use to just disappear between releases without recourse. I'm thinking of the gnome service control manager workalike and sessions, thanks Ubuntu.

Spaces, "Save As"... :(

Changing APIs like Gnome/KDE did frequently is something else altogether, especially when you don't have the luxury of a huge user base to force commercial companies to rewrite their code. As for open source application developers, they often just flee and abandon their apps than rewrite them for the third time around some new API.

Every major release I can remember (since KDE 2.0) has kept binary compatibility for the libraries (see http://techbase.kde.org/Policies/Binary_Compatibility_Issues...). There have been exceptions for Plasma libraries, but those were announced beforehand.

Many current KDE libraries are evolutions of technology going back to KDE 2 (for better or worse).

Perhaps the biggest change over the whole lifetime has been the shell (going from kdesktop+kicker in KDE 2+3 to Plasma in KDE 4) and going from aRts in KDE 2+3 to Phonon in KDE 4.

Btw, we were highly encouraged to adopt the popular multimedia framework at the time (gstreamer 0.10) instead of writing Phonon. gstreamer will soon be releasing an incompatible 1.0, while in the interim Phonon gained native support for the sound server that become highly popular in the meantime, PulseAudio.

KDE has been doing its part to insulate developers as much as possible from mandatory API changes. It's not as good as the kernel or Microsoft has been able to do, but it's certainly not as bad as GNOME IMHO.

I can say this as an app maintainer too. Going onto the KDE 4.10 development cycle, JuK still has code that uses the "kde3support" library. :(

The thing that make to change from Gnome 2.x to KDE 4 was that Gnome was running at slug speed with VESA driver (ATI drivers not working well in a barebone that was making for the TV), when KDE/KWin was running pretty smooth. Plus in this last years and tested that when the crap ATI drivers introduce some instability to the GUI, KDE at least self-recover alone very quickly and I don't get unusable GUI, like I saw with Gnome shell + ATI privative drivers.

Finally, I love how I can customize my desktop to my needs... It's like the old times when Enlightenment was the cool and beautiful wm, but easy to change anything.

KDE 4 isn't perfect, but I think that it's much better that the crap that Gnome is actually.

There were other application-specific APIs that were used for things that were beyond the capability of aRts and Phonon (e.g. Juk could use gstreamer directly in KDE 3) but the KDE APIs themselves were pretty stable.

It's true that a Qt change in binary incompatible and involves at least some source porting effort, but honestly those have been fairly infrequent, and only one of those transitions (3->4) involved more than minor source changes. Qt 4 was released at the end of 2005 (around the time of Linux 2.6.15) and is still the "current" version, though the Qt 5 release is around the corner.

In just the KDE 3 timeframe to now, on the other hand, there has been gstreamer 0.6, 0.8, 0.10, soon 1.0, PulseAudio, ConsoleKit, HAL, HAL's replacement (DeviceKit), DeviceKit's replacement (udev and friends, upower, udisks, etc.), PolicyKit, PolicyKit's replacement ("polkit"), NetworkManager 0.8, NetworkManager 0.9.

Needless to say, trying to keep in sync with all of that has not been fun. :-/

There are also massive API compatibility issues even between those two enormous shifts -- witness the sudden breakage of all SDL games due to raw surface access in CoreGraphics moving to IOSurface, just to name one.

The difference is, I think, that Apple has conditioned developers to expect massive breakage, and the OS X app industry has structural differences from the Linux desktop app community that make it more likely for apps to get fixed quickly.

1) Both were necessary. Carbon was a stop-gap port of the old Mac way and incompatible with the OS X design, and the x86 transition had to be made because the PPC mobile roadmap was late (and bad). It was not a "let's break API compatibility because we came with a cool new idea" or "because the old maintainer got bored and the new one has different ideas", which often happens in Gnome/Linux land.

2) Not to mention that the Carbon->Cocoa took like 7+ years to be made full (Carbon apps actually still run on modern OS X), and Apple went to the effort to provide a full bundled compatibility layer for PPC for many years (Rosetta), plus "fat binaries" to make the switch easier. I.e nothing like the "let's break stuff to make them better and let them deal with it" attitude in the Gnome world.

3) Also: Apple remade the old stuff ITSELF each time, with paid developers working on the OS X, bundled userland and Apple pro apps to port them, whereas Gnome changes made lots of developers walk away from the free, OS, Gnome apps.

4) Plus: Apple had a lot of market share (compared to Linux) to force companies like Adobe, Microsoft et al, to port their stuff along eventually. Gnome did not have that benefit.

So, not at all the same, and no, it's far more than "Apple conditioning developers to expect change".

Server side stuff has a lot of support from big shops that need their iron to work well - so that can go along just fine as is.

My point is that because it exists and is nearly viable it is a success and does what we need it to, even if using it can be nightmarish.

It is a community driven effort, which means the community pushes its development and design. Who do you see using the linux desktop? The exact type of people who design and develop it. Power users and developers.

They are a stubborn bunch. Even if Unity/Gnome 3/etc were more usable desktops, the majority of users would hate it. Because something like a newbie usable desktop just doesn't fit who's designing the desktop.

Something else will have to shove it into the wider market where it can be judged by those standards, not by what set of power user keyboard short cuts it supports.

I think it will go nowhere because the current linux user base will be holding it back. Gnome3 and Unity are fighting this tide also.

I don't think the state of the linux desktop should be, "Copy XP and put a real shell behind it".

Besides, if you want a Power User version of Ubuntu, just use Debian.

The kernel did what it was supposed to do back in 1991.

What is the "desktop" supposed to do?

There will never be unanimous agreement on that - every user will have different needs and preferences - and so desktop developers live in constant denial, believing they are the only ones whose preferences are relevant - i.e. they know what's best for users - and blaming others for their own failings.

I'm not a Linux fan, but I give Torvalds' +1 for his response (and knowing the value of "not breaking stuff").

No one emerged as the "Linux Desktop Guru", taking full responsibility of a distribution completely targeted at fixing the UX, selecting a set of defaults to actually create a development platform. Fixing the problem of users supporting in a perfect way at least a subset of computer systems (targeting macs makes sense for instance).

So the Linux Desktop was the sum of a multitude of different tries, that alone sometimes were interesting, but not enough at all to define the whole OS experience from the point of view of the user.

Then the eazel guys showed up and made Nautilus, if anybody could have been the guru, it was Andy. I honestly don't know what appended there. Nautilus is still around, it's under gone some substantial changes along the way.

Seems to me it's a very difficult community problem to solve. Look at replacing sysv init, it's a very technical issue, it doesn't matter to a lot of users, there are some technical goals you can bring up to keep the discussion rooted in fact over opinion and there is tremendous disagreement about it. The community hasn't arrived at any conclusion exactly. Why should something as esoteric as a ui be easier to arrive at consensus on? Everyone has an opinion on ui.

There is another aspect to it, it's distribution fragmentation. How do you set an ip address in Ubuntu? How about Fedora? Do it via UI and command line. They are different and that's an easy one. The mechanics are so different, it's not a desktop problem, it's a distribution problem, which sort of makes KDE and GNOME toolkits rather than desktops themselves.

A distribution that tried to provide a full experience, instead of being, indeed, a Software Distribution. In that case you can have people in charge of deciding all the details. Payed people to do the best work possible.

Firstly, Corba was there from the start, and then it was removed.

Secondly, there is no .NET dependency that must be satisified to create GNOME applications.

While there are a few interesting points Tovalds made, the problem isn't really a technical one in the code, but rather a lack of focus on what really matters.

No, but many apps that are part of the "official" Gnome desktop run on Mono.

The reality is many people try many systems and settle on one they like.

There is no such story for kernel-mode drivers. The problem was so dire with network card drivers at one time that there was a project that enabled windows drivers to work on Linus. Think about it - if Linux standardized on a stable external API, then there'd be no arguments about how to implement it.

Linus kept talking about userland vs kernel, and did not address the issue Miguel talked about. Kernels need performant external interfaces too. Drawing a line around what is being innovated and changed will only help Linux the way posix did.

On another tangent, when Microsoft made IE6 the stable browser platform, it gave a clear goal for Mozilla. It became a reference implementation, allowing Mozilla to emulate all the layout behaviors and eventually carving out a nice marketshare for itself.

http://www.informationweek.com/windows/microsoft-news/micros...

http://www.groklaw.net/article.php?story=20090619161307529

http://www.pcmag.com/article2/0,2817,2331462,00.asp