meaningful according to whom? and desktop users couldn't care less about 'hard coded nice levels' if it means their 3d games and/or X apps work better: yes, I know this is anathema to the linux developers where only super perfect code supposedly should go in, however if this supposedly super perfect code doesn't meet desktop users' needs as well as hacks, well, I'd all be for giving desktop users as many hacks as they want/need (as long as this could be changed via either a pluggable architecture or a difference in make config).
If you want hard coded nice levels, get your distro maker to hard code some nice levels (or do it yourself)! The kernel developers don't even need to know about it.
In fact, if we take everything GNU except the kernel, and port it to Solaris or *BSD, or even Mac OS X, we generally still call it Solaris or *BSD or Mac OS X with a few GNU tools sprinkled in. We don't call it GNU/Solaris, just because someone installed glibc, gcc, etc.
If glibc is the only libc, and bash is the default shell, and ld is GNU ld, etc then people generally *have* called those things GNU/Solaris or whatever. Take a look at Nexenta OS for example.
You are the one being resistant to changing your mentality to having a focus on what the end user wants.
You are massively oversimplifying and misstating the issues involved.
First, GNU+Linux distributions - as they currently exist - don't have any major specific flaws that are preventing uptake other than the fact that they aren't Microsoft Windows. There's no way to make them actually be Microsoft Windows, so that specific problem isn't solvable.
Second, in the place where it matters - distributions - there is no issue with wanting to meet user needs. Distributions like Ubuntu spend all their time trying to make the user experience better by fixing the actual problems that real users have.
I dare you to spend a few minutes playing "devils advocate" and come up with arguments that support the parent post.
These arguments - followed by applying their results to the real world - result in the creation of a distribution like Linspire. Linspire exists. Maybe Ubuntu wants to have Click & Run too. That's happening. If you think more needs to be done, do it. But don't expect that you're going to change the behavior of thousands of developers (most of which aren't even interested in your use case at all) just by complaining on Slashdot - and don't think that some single overhyped issue like DVD playback is really what's slowing uptake (and therefore worth sacrificing any other goals to fix).
What we're debating is operating system categories. "Embedded Linux" is a different OS category than "GNU+Linux". Differentiating between those is actually more functionally useful than differentiating between GNU+Linux and, say, Solaris.
Except that was redhat in 2001, closer to the present day, Debian in 2005 (which has also been studied) is almost ten times the size overall, with only one of the largest packages being GNU: OOO/Linux/NVU/Mozilla/GNU/XFS/XFree etc.
It's worth differentiating between applications and core OS components. In a general purpose GNU+Linux system, the core components unquestionably include the Linux kernel and GNU Libc. In contrast, OpenOffice.org is obviously an application rather than any kind of OS component. Exactly where you draw the line is arguable (I could argue gcc and grep as OS components with you all day) - but that minimum system definitely includes both GNU and Linux. Personally, I think the Debian "base" system or the Gentoo stage3 are a good start for a "core OS" definition.
"Linux" is a convenient shortening and covers the whole family of systems (GNU-using or otherwise) built on the Linux kernel, in comparison with eg. "Windows" as a term.
If that's really what you want, then it's correct. But... I doubt that you really want that most of the time. Frequently you're really just talking about GNU+Linux rather than including embedded Linux. Equally frequently, what you really want is "Free Unix" which includes Solaris and *BSD.
At the end of the day, language is fluid and people will call things by a name that is convenient and understood. [anything]-slash-Linux is annoying to say, so most people won't.
This is a valid argument all by itself. I fall back on saying "Linux" to be clearly understood all the time. That doesn't change the fact that what I'm usually referring to should be properly called GNU+Linux.
Most people would want Linux because of it's hardware support, why should they care about GNU to provide the user space tools in between their hardware and their desktop?
GNU isn't just "user space tools". It also is the core system libraries like libc. I think you'll find that any developer writing code for a system will be very interested in which libraries are available on the system - as will anyone who tries to install software compiled against those libraries.
However, the "GNU OS" likely includes other tools that are really applications, like grep. So what is the magic that ties these together and makes them uniquely GNU?
They were created by the GNU Project for the specific purpose of serving as components of the GNU System. This is the key thing you seem to be missing - all these pieces didn't just arbitrarily show up - they are the results of a project specifically intended to create a complete operating system.
As I've shown, X.org can exist independent of the GNU OS. However, a modern GNU OS would be hard-pressed to find a windowing system to replace X.org. Therefore, I request that, when talking about a GNU OS which includes graphical components, you refer to it by its proper name: GNU/X.org.
Sure. I'll start talking about GNU/X.org (or GNU/Linux/X.org) the minute you start talking about Solaris/X.org and FreeBSD/X.org. I made my initial comment for a reason - every modern Unix OS that includes a GUI uses X.org, so the fact that the GNU system does too isn't terribly interesting.
The whole story here is as follows:
The GNU Project has created a complete OS (as complete as Solaris or *BSD, which are generally recognized as complete operating systems).
Millions of people are running it (with one component swapped out).
The GNU Project would like to be given credit for their work. They can't demand credit, but it's not entirely unreasonable for them to act insulted when credit is withheld.
I'm tempted to start referring to GNU/Linux systems as simply GNU, and then when people ask "What about Linux?" respond "Oh, that's just some drivers - not worthy of mention". Sure - the Linux code base *is* largly "just drivers", but that doesn't mean it's all drivers or that it's not worthy of of credit.
At least if we're talking about who should be credited in the name, the logical answer really comes down to this: Everyone has an equal claim. However, you don't see X.org throwing fits and refusing talks or interviews with people who refuse to call it X.org/Linux, even though X.org is arguably as much code and as much a necessity of a modern Linux desktop as the GNU tools are.
As long as you keep thinking about the output of GNU Project as just "the GNU tools - maybe tar and some other stuff", you're going to continue to see things from a warped perspective and draw unreasonable conclusions. The GNU Project has developed a complete operating system - the GNU system. Like any other modern Unix-like OS, it uses X.org as its windowing system. The only significant difference between something like Ubuntu and the GNU System is one component - the kernel.
A request from the founder of the GNU Project that people recognize their effort when they talk about systems that incorporate and rely on the vast majority of their work should seem entirely reasonable. Responding by denigrating the GNU System as "just some tools" is insulting at best.
Not to bash you, but that's exactly the attitude we need to fight.
I disagree. The only attitude we need to fight is the belief that free software needs to change drastically to be generally palatable. Not only is it incorrect, that message is directly slowing free software adoption.
For the average user, Linux is not ready. You can't play a DVD without installing extra software. You can't play an MP3 without installing a codec.
And these things are currently illegal to provide with free software. In spite of that, codecs are installed by clicking through two dialog boxes on Ubuntu. That means that for any desktop application that doesn't require DVD playback (say... an office setting), Ubuntu works great.
Adding new hardware can require you get get a new kernel. Yes, some of those problems are due to things like stupid licensing issues which aren't the fault of Linux. But there should be no excuse for patching a kernel to get a piece of hardware to work.
This isn't a practical problem that should effect any serious deployment today. Faster driver turn-around would be nice, but that will only happen as uptake increases.
The only way to succeed is to rethink Linux -- not technically, but culturally. The "year of desktop Linux" has been every year for the past 10 years -- and Linux still doesn't have broad success. The reason is less about technology than it is about a culture of consistently trying to cover up the faults in Linux.
GNU/Linux is a wildly successful OS. What.. where you expecting Microsoft (a convicted monopolist who has spent decades on vendor lock-in) to vanish in a puff of wishful thinking?
You shouldn't need a kernel patch for Synaptics touchpads.
You don't. You just need to install one package that comes standard in Ubuntu. A package that Dell should have installed on the machine by default.
We have to actually listen to users and accept that Linux has got a lot of great things going for it but it also has a lot of major flaws that keep it marginalized.
If you have bugs to report - great. Report them and they'll get fixed (or you'll have to accept that they're legally unfixable). That's really the whole story.
Linux will achieve market success the day that someone can use Linux without ever once thinking about the kernel, or compiling software, or reading HOWTOs. When Linux "just works" then Linux can effectively compete.
Any major distribution of GNU/Linux (sans Gentoo perhaps) has "just worked" to that extent for years now. FUD about compiling software, reading HOWTOs, and editing text-based config files from internet posters like you is scaring more users away than actually ever have to even think about those things.
That day isn't going to come until we as a community stop making excuses and start realizing that we have a set of real-world problems -- problems that we can solve -- but only if we're willing to accept that there's a tradeoff between being a technology fan's OS and an OS that works for the general population. That means having a much greater willingness to find fault and being much more proactive about providing users with a quality experience.
The day has already come when distributions like Ubuntu are good enough for most normal users (who don't play Windows computer games). At this point it's just a question of uptake - and posts like yours just spread unnecessary fear.
Does it matter? All those being present doesn't make the system "largely" GNU. Unless nothing much else is also present. Including everything that comes from GNU doesn't make it largely GNU, including little that does not come from GNU would make it largely GNU.
Just because more things have been added doesn't mean that when you put all the pieces of the GNU System together you don't end up with the GNU System. Are you saying that if you bundled Oracle with Solaris it would stop being Solaris?
Well for a start I don't think you'll find Emacs in any major embedded Linux distribution, and many (most?) don't use glibc.
That's true. Many embedded systems run operating systems that really should be called "Linux" rather than "GNU/Linux". When I said "major Linux distribution", I was referring to general purpose distributions like Red Hat and Debian.
Further, the list quoted is acutally largely non-gnu (both by package count and total sloc) - at least if you define largely as "more than half".
That's also true. So in order of LOC we come out with something like GNU/Linux/Mozilla/Java/OpenOffice/Xorg/Mesa/Tetex/FreeCIV. It would be reasonable to pick any slash and leave out the projects to the right for conciseness.
Ok, we can ignore tar/grep for most desktop users. We can ignore gcc/make/etc, as they are development tools -- not relevant to anyone but developers.
By that logic we can ignore Linux too because "most desktop users" don't directly interact with it. I guess that means that we should be calling the system either "Gnome" or "KDE" (or maybe "Ubuntu"). I'm sorry... but there's no reasonable logic based on size or importance of contribution that would mention Linux but not GNU.
Is he reviewing Linux, or reviewing this specific model of laptop? If the latter, then it's perfectly fair for him to point out that it lacks a tool to adjust touchpad sensitivity.
He specifically said that he was reviewing Ubuntu (and not that specific model of laptop) in the article.
Ubuntu is a community developed, linux-based operating system that is perfect for laptops, desktops and servers.
This doesn't mean that it runs perfectly on every laptop, desktop, and server ever produced.
Are you REALLY stating that people should go look for a pc/laptop with a sticker that says "Designed for Linux/Ubuntu" in order to have a 100% working machine? M$ and Apple would love such kind of thinking from the FOSS user base.
To get a fair comparison of hardware support, yes. Windows and Mac OS X almost always come pre-installed on hardware configurations specifically designed to run them. No-one holds Mac OS X responsible for the difficulties of installing it on a "Made for Windows" machine, nor does anyone say that Windows Vista sucks if it's hard to install it on a Mac (they'd say the Mac sucks - which also isn't entirely reasonable). Expecting Ubuntu to be magically able to do better than either of the others is silly.
The fact of the matter is that 100% Ubuntu compatible hardware is widely available. That means that there's no excuse to make unfair comparisons between "Ubuntu on Random Hardware" and "Mac OS X on a Mac" or whatever.
... Linux is not the OS developed by the GNU project. It is an OS that's common feature is the Linux kernel; as usually distributed, it includes various tools from the GNU project.
I guess this is where we disagree.
Any of the common "Linux distributions" do consist largely of the GNU System. It's not just "some tools"... it's the entire set of software on top of the kernel that take the system to the level of a Unix-like OS.
We're talking about the system libraries (GNU libc), the shell (bash), the common Unix-like tools (tar/grep/etc), the development environment (gcc/make/etc), and frequently even the desktop environment (GNOME). If you wanted to build a Unix-like OS on top of Linux without GNU, your only other option would be to take vast chunks of some other Unix-like OS like Solaris or *BSD.
Sure, there's a component here or there that's from somewhere else - a good example would be init - but GNU/Linux without GNU wouldn't be a Unix-like OS anymore. On the other hand, there have been successful attempts at running the GNU system without Linux (i.e. GNU/kFreeBSD and Nexenta OS).
Mac OS X is widely considered to be good enough for non-technical computer users - even in spite of issues like not having 100% support of all peripherals. That strongly implies that 100% peripheral support is not a pre-requisite for Ubuntu to be good enough for non-technical users.
You really think that the existence of a fully functional free kernel had *no* effect on the amount of resources put into the Hurd project? After Linux was released, finishing the Hurd became an academic exercise primarily of interest to people having this particular argument - and that's not a good enough reason for software to get written.
When I hear that, I don't think they're still going to be working on that one last part for 16 years after an alternative component comes along.
Linux was good enough, so there was no technical reason to get Hurd done. Sorry, but winning a stupid argument about OS naming isn't an important enough reason to finish a non-trivial software project (especially when you can go play with research concepts on it instead). We could discuss how long Hurd would have taken if Linux hadn't been written - but we wouldn't get anywhere with that discussion either.
One thing to consider is this: a fully functional GNU can be run with different kernels (see GNU/kFreeBSD or Nexenta OS), but no complete Unix-like OS has been built on Linux without GNU.
Think about Mac OS X for a moment. Which of your comments are specific to Ubuntu, and which apply equally to Ubuntu and OS X. Now... is OS X good enough for "a dumb user who does not want to dig into the inner workings of his computer"?
As a test, I just tried to beat this pc to death by multitasking the following: excel 2003, excel 2007, word 2003, word 2007, access 2003, access 2007, publisher 2003, publisher 2007, visual studio 2003, visual studio 2005, windows media player 11 (playing an mp3 with visualization running, coldfusion studio, firefox, ie7, opera, photoshop 7, thunderbird, and last but not least, Taskman.exe.
Wait a second. You're telling me that your OS can handle 17 idle processes and one active one when they all fit in RAM? That's amazing. Or at least it would have been in 1972...
There is no damn monopoly. There's OSX, there's Linux, DELL and Lenovo offer it preinstalled even. And you can buy Mac machine with OSX.
Economically, the problem with a monopoly isn't simply that there's only one vendor. It's that there exists a single vendor who has overwhelming market power. Microsoft has such market power and has repeatedly abused it (which is illegal). That sort of behavior is economically damaging, and the sheer volume of money that it produces for the monopolist is socially damaging (they'll lobby for laws that benefit them while hurting society in general). This actually remains true even for oligopolies where the number of competitors is reasonably small.
You (and many others) are overstating the problem.
There is no significant gap in practical usability between Windows Vista, Mac OS X, and Ubuntu. There are some specific sore points (oh noes, you have to click "Ok" twice to play an MP3 the first time) and a hell of a lot of resistance to change.
Any person or company who wants to run Ubuntu on their desktop(s) can do so today - with almost exactly as much effort as it takes to switch from Windows to a Mac. Some hardware and peripherals may be unsupported and need to be replaced with supported hardware (which is readily available). The user may need to learn some new applications and/or work flows. That's basically it.
The only real problem with Ubuntu is this: People overestimate the conversion and retraining costs and underestimate the benefits because they personally don't want to have to learn anything new or change any business processes that seem to work.
Most cameras actually work perfectly out of the box on Ubuntu. Somehow these reviewers keep finding the one obscure model of camera that doesn't work - or don't even try any cameras and just repeat the (incorrect) "common wisdom" that Linux has trouble with cameras.
iPods on the other hand can be annoying, but pretty much every other MP3 player on the planet works fine.
Although I don't think the need to download new CODECs is all that severe, the poor video performance, problems with the touchpad and the crashing volume control applet are absolutely unacceptable for a pre-install.
Now - is this a problem with Ubuntu or did Dell set up a bad default configuration? It sounds to me like the latter.
Slashdot Mirror
If you want hard coded nice levels, get your distro maker to hard code some nice levels (or do it yourself)! The kernel developers don't even need to know about it.
If glibc is the only libc, and bash is the default shell, and ld is GNU ld, etc then people generally *have* called those things GNU/Solaris or whatever. Take a look at Nexenta OS for example.
You are massively oversimplifying and misstating the issues involved.
First, GNU+Linux distributions - as they currently exist - don't have any major specific flaws that are preventing uptake other than the fact that they aren't Microsoft Windows. There's no way to make them actually be Microsoft Windows, so that specific problem isn't solvable.
Second, in the place where it matters - distributions - there is no issue with wanting to meet user needs. Distributions like Ubuntu spend all their time trying to make the user experience better by fixing the actual problems that real users have.
These arguments - followed by applying their results to the real world - result in the creation of a distribution like Linspire. Linspire exists. Maybe Ubuntu wants to have Click & Run too. That's happening. If you think more needs to be done, do it. But don't expect that you're going to change the behavior of thousands of developers (most of which aren't even interested in your use case at all) just by complaining on Slashdot - and don't think that some single overhyped issue like DVD playback is really what's slowing uptake (and therefore worth sacrificing any other goals to fix).
What we're debating is operating system categories. "Embedded Linux" is a different OS category than "GNU+Linux". Differentiating between those is actually more functionally useful than differentiating between GNU+Linux and, say, Solaris.
It's worth differentiating between applications and core OS components. In a general purpose GNU+Linux system, the core components unquestionably include the Linux kernel and GNU Libc. In contrast, OpenOffice.org is obviously an application rather than any kind of OS component. Exactly where you draw the line is arguable (I could argue gcc and grep as OS components with you all day) - but that minimum system definitely includes both GNU and Linux. Personally, I think the Debian "base" system or the Gentoo stage3 are a good start for a "core OS" definition.
If that's really what you want, then it's correct. But... I doubt that you really want that most of the time. Frequently you're really just talking about GNU+Linux rather than including embedded Linux. Equally frequently, what you really want is "Free Unix" which includes Solaris and *BSD.
This is a valid argument all by itself. I fall back on saying "Linux" to be clearly understood all the time. That doesn't change the fact that what I'm usually referring to should be properly called GNU+Linux.
GNU isn't just "user space tools". It also is the core system libraries like libc. I think you'll find that any developer writing code for a system will be very interested in which libraries are available on the system - as will anyone who tries to install software compiled against those libraries.
They were created by the GNU Project for the specific purpose of serving as components of the GNU System. This is the key thing you seem to be missing - all these pieces didn't just arbitrarily show up - they are the results of a project specifically intended to create a complete operating system.
Sure. I'll start talking about GNU/X.org (or GNU/Linux/X.org) the minute you start talking about Solaris/X.org and FreeBSD/X.org. I made my initial comment for a reason - every modern Unix OS that includes a GUI uses X.org, so the fact that the GNU system does too isn't terribly interesting.
The whole story here is as follows:
I'm tempted to start referring to GNU/Linux systems as simply GNU, and then when people ask "What about Linux?" respond "Oh, that's just some drivers - not worthy of mention". Sure - the Linux code base *is* largly "just drivers", but that doesn't mean it's all drivers or that it's not worthy of of credit.
As long as you keep thinking about the output of GNU Project as just "the GNU tools - maybe tar and some other stuff", you're going to continue to see things from a warped perspective and draw unreasonable conclusions. The GNU Project has developed a complete operating system - the GNU system. Like any other modern Unix-like OS, it uses X.org as its windowing system. The only significant difference between something like Ubuntu and the GNU System is one component - the kernel.
A request from the founder of the GNU Project that people recognize their effort when they talk about systems that incorporate and rely on the vast majority of their work should seem entirely reasonable. Responding by denigrating the GNU System as "just some tools" is insulting at best.
I disagree. The only attitude we need to fight is the belief that free software needs to change drastically to be generally palatable. Not only is it incorrect, that message is directly slowing free software adoption.
And these things are currently illegal to provide with free software. In spite of that, codecs are installed by clicking through two dialog boxes on Ubuntu. That means that for any desktop application that doesn't require DVD playback (say... an office setting), Ubuntu works great.
This isn't a practical problem that should effect any serious deployment today. Faster driver turn-around would be nice, but that will only happen as uptake increases.
GNU/Linux is a wildly successful OS. What.. where you expecting Microsoft (a convicted monopolist who has spent decades on vendor lock-in) to vanish in a puff of wishful thinking?
You don't. You just need to install one package that comes standard in Ubuntu. A package that Dell should have installed on the machine by default.
If you have bugs to report - great. Report them and they'll get fixed (or you'll have to accept that they're legally unfixable). That's really the whole story.
Any major distribution of GNU/Linux (sans Gentoo perhaps) has "just worked" to that extent for years now. FUD about compiling software, reading HOWTOs, and editing text-based config files from internet posters like you is scaring more users away than actually ever have to even think about those things.
The day has already come when distributions like Ubuntu are good enough for most normal users (who don't play Windows computer games). At this point it's just a question of uptake - and posts like yours just spread unnecessary fear.
Just because more things have been added doesn't mean that when you put all the pieces of the GNU System together you don't end up with the GNU System. Are you saying that if you bundled Oracle with Solaris it would stop being Solaris?
That's true. Many embedded systems run operating systems that really should be called "Linux" rather than "GNU/Linux". When I said "major Linux distribution", I was referring to general purpose distributions like Red Hat and Debian.
That's also true. So in order of LOC we come out with something like GNU/Linux/Mozilla/Java/OpenOffice/Xorg/Mesa/Tetex/FreeCIV. It would be reasonable to pick any slash and leave out the projects to the right for conciseness.
Take a look at http://www.dwheeler.com/sloc/redhat71-v1/redhat71sloc.html, section 3.2. Are any of the GNU system components that the author mentions there absent from any major Linux distribution?
By that logic we can ignore Linux too because "most desktop users" don't directly interact with it. I guess that means that we should be calling the system either "Gnome" or "KDE" (or maybe "Ubuntu"). I'm sorry... but there's no reasonable logic based on size or importance of contribution that would mention Linux but not GNU.
He specifically said that he was reviewing Ubuntu (and not that specific model of laptop) in the article.
This doesn't mean that it runs perfectly on every laptop, desktop, and server ever produced.
To get a fair comparison of hardware support, yes. Windows and Mac OS X almost always come pre-installed on hardware configurations specifically designed to run them. No-one holds Mac OS X responsible for the difficulties of installing it on a "Made for Windows" machine, nor does anyone say that Windows Vista sucks if it's hard to install it on a Mac (they'd say the Mac sucks - which also isn't entirely reasonable). Expecting Ubuntu to be magically able to do better than either of the others is silly.
The fact of the matter is that 100% Ubuntu compatible hardware is widely available. That means that there's no excuse to make unfair comparisons between "Ubuntu on Random Hardware" and "Mac OS X on a Mac" or whatever.
I guess this is where we disagree.
Any of the common "Linux distributions" do consist largely of the GNU System. It's not just "some tools"... it's the entire set of software on top of the kernel that take the system to the level of a Unix-like OS.
We're talking about the system libraries (GNU libc), the shell (bash), the common Unix-like tools (tar/grep/etc), the development environment (gcc/make/etc), and frequently even the desktop environment (GNOME). If you wanted to build a Unix-like OS on top of Linux without GNU, your only other option would be to take vast chunks of some other Unix-like OS like Solaris or *BSD.
Sure, there's a component here or there that's from somewhere else - a good example would be init - but GNU/Linux without GNU wouldn't be a Unix-like OS anymore. On the other hand, there have been successful attempts at running the GNU system without Linux (i.e. GNU/kFreeBSD and Nexenta OS).
What I meant to say was this:
Linux can't even run the traditional C hello world without GNU System components like GNU Libc.
On the other hand, the 0% of all non-trivial user-level applications that are written entirely using syscalls will run fine on Linux alone.
You're missing my point.
Mac OS X is widely considered to be good enough for non-technical computer users - even in spite of issues like not having 100% support of all peripherals. That strongly implies that 100% peripheral support is not a pre-requisite for Ubuntu to be good enough for non-technical users.
You really think that the existence of a fully functional free kernel had *no* effect on the amount of resources put into the Hurd project? After Linux was released, finishing the Hurd became an academic exercise primarily of interest to people having this particular argument - and that's not a good enough reason for software to get written.
Linux was good enough, so there was no technical reason to get Hurd done. Sorry, but winning a stupid argument about OS naming isn't an important enough reason to finish a non-trivial software project (especially when you can go play with research concepts on it instead). We could discuss how long Hurd would have taken if Linux hadn't been written - but we wouldn't get anywhere with that discussion either.
One thing to consider is this: a fully functional GNU can be run with different kernels (see GNU/kFreeBSD or Nexenta OS), but no complete Unix-like OS has been built on Linux without GNU.
Think about Mac OS X for a moment. Which of your comments are specific to Ubuntu, and which apply equally to Ubuntu and OS X. Now... is OS X good enough for "a dumb user who does not want to dig into the inner workings of his computer"?
Wait a second. You're telling me that your OS can handle 17 idle processes and one active one when they all fit in RAM? That's amazing. Or at least it would have been in 1972...
Economically, the problem with a monopoly isn't simply that there's only one vendor. It's that there exists a single vendor who has overwhelming market power. Microsoft has such market power and has repeatedly abused it (which is illegal). That sort of behavior is economically damaging, and the sheer volume of money that it produces for the monopolist is socially damaging (they'll lobby for laws that benefit them while hurting society in general). This actually remains true even for oligopolies where the number of competitors is reasonably small.
You (and many others) are overstating the problem.
There is no significant gap in practical usability between Windows Vista, Mac OS X, and Ubuntu. There are some specific sore points (oh noes, you have to click "Ok" twice to play an MP3 the first time) and a hell of a lot of resistance to change.
Any person or company who wants to run Ubuntu on their desktop(s) can do so today - with almost exactly as much effort as it takes to switch from Windows to a Mac. Some hardware and peripherals may be unsupported and need to be replaced with supported hardware (which is readily available). The user may need to learn some new applications and/or work flows. That's basically it.
The only real problem with Ubuntu is this: People overestimate the conversion and retraining costs and underestimate the benefits because they personally don't want to have to learn anything new or change any business processes that seem to work.
Most cameras actually work perfectly out of the box on Ubuntu. Somehow these reviewers keep finding the one obscure model of camera that doesn't work - or don't even try any cameras and just repeat the (incorrect) "common wisdom" that Linux has trouble with cameras.
iPods on the other hand can be annoying, but pretty much every other MP3 player on the planet works fine.
Now - is this a problem with Ubuntu or did Dell set up a bad default configuration? It sounds to me like the latter.