The Real Reason Why Linux Isn't On The Desktop

Posted 19 Sep 2007 at 16:18 UTC by johnnyb Share This

A recent article has surfaced discussing what the author thinks are reasons Linux won't succeed on the desktop. It is his mentality, and not his reasons, which are the real culprits.

Let's look at his reasons first before we get into the real ones:

1) "Prohibitive application porting costs"

This may be true (I don't think it is), but it has nothing to do with Linux's success on the Desktop.

As for the multiple targets, it's not that much of an issue. You can avoid 99% of your troubles by statically linking troublesome libraries. You can avoid the other 1% by specifying which distributions your app runs on.

2) "The Fanboy alienation factor, or how Linux's biggest supporters drive away potential new users"

This was indeed a big problem in the 1990's. Oh look, it's 2007 now.

3) "You can't make money on the operating system"

This is an assertion unsupported by data. There are lots of people making money on the operating system. In fact, I would venture to guess there are more companies making money on Linux operating system development than on other operating system developments. Whether or not this translates to desktop usage is unknown. I'll address that below.

4) "Resistance from average users"

This is somewhat true. But the fact is, most users _want_ something different, but it isn't necessarily Linux. That's why OSX has had a huge surge.

5) "Linux is "simple"; Windows "just works""

There could have been good points made here, but the author failed to make any of them. Out-of-the-box, Linux actually works better for arbitrary hardware than Windows. Most of the time when you buy a Windows box from a vendor, IT WOULD NOT INSTALL PROPERLY WITH A STANDARD WINDOWS INSTALL DISK. That's one of several reasons why they all have those custom reinstall disks -- a plain-jane Windows reinstall disk simply wouldn't work. Try taking a machine that runs Windows 2000 just fine. Give them two install disks -- one for Vista and one for Fedora or Ubuntu -- and see which one is the smaller headache to install.

Now, there _is_ a usability problem in Linux. This will be addressed more below, but basically it is that most of the necessary applications for the desktop are NOT part of any desktop environment, or you have to have apps from each DE to work. THIS is problematic for new users.

6) "There are way too many Linux distros"

This claim is just dumb. For instance, I once had a distribution called "Serial Terminal Linux", and I believe it was in distro-watch. It was a floppy-based Linux that installed on an old laptop to turn it into a serial terminal. I seriously doubt that it was having any effect on a user trying to choose what distro to run. If a user comes in out of the blue they are going to run Fedora or Ubuntu. If they come in via a friend, they are going to run what their friend suggests. So, that limits the choice down to 2. No one is going to complain about a choice between two.

In fact, my argument below will be that there needs to be more distros.

7) "No powerful evangelist for Linux comparable to Bill Gates or Steve Jobs"

Possibly true.

Here's the Real Story

Here's the real facts: It is the unity of Linux which holds it back the most. The fact is, there are already two very powerful one-size-fits-all operating systems on the market. The whole presupposition of the article - that Linux needs to be like Windows or Mac to win - is the mentality which is preventing Linux from breaking out. Breaking their hold on generic-for-everyone operating systems is a waste of time and effort. Instead, upcoming Linux vendors need to take the following steps:

  1. Prepare to invest a lot in the code. Linux is shared and useful, but it isn't unified. It is the job of the vendor to present a unified environment. People who fail to invest in their own technology should not be surprised when the investment doesn't pan out.
  2. Pick a narrow, targeted market. For instance, say multimedia development. Then think about everything from the perspective of that user. Kernel adjustments, applications, application settings, application customizations, application integrations -- all of these should be done with your target in mind.
  3. Don't call it Linux. This is YOUR operating system. The fact that you are using a lot of shared code is irrelevant, and will get people complaining that it isn't Linux-y. That's the whole point -- DON'T do a Linux distribution, make an operating system that does something better than anything else.
  4. Make it a whole-hardware package. Let's go back to multimedia development. Sell the whole shebang -- maybe even the audio/video equipment too. This is a packaged deal.
  5. Don't bother unless you've got a truly innovative idea. Go out and use existing software to do the task you have in mind. See if you're idea will make a real improvement. If it won't, you're just wasting your time.

The real problem with many vendors who have tried Linux, is that Linux gives the false impression that vendors don't have to do much except provide an installer, package manager, and maybe some config tools. Linux does in fact ease development of new ideas, but it does not replace development.

There will always be a market for the Fedora's of the world -- a kind of kitchen-sink distro for programmers and sysadmins. But this is not a desktop distribution.

By focusing on specific market segments, Linux can in fact gain ground on the desktop.

Hah, posted 19 Sep 2007 at 17:57 UTC by bi » (Master)

Larry Cafiero scripsit:

He makes the self-serving pithy observation that, since there are so many distros, that “Linux is a forking mess.” As if we should only have one or two choices in computing, rather than the 300 choices of GNU/Linux distros (and those are the active ones) offer.

That’s because freedom — whether in computing or in the rest of everyday life — is about choice, and it’s unfortunate that Wolfe doesn’t get it. But then, most corporate lackeys beholden to the party line of their corporate masters don’t; or if they do, they wallow in hypocrisy while ignoring the truth.

And Brian Proffitt:

For all those pundits who have or are planning to raise up this inane issue, here's a list of 359 things off the top of my head of things we consume or use that we have to make a choice about. Try applying the single-choice only argument to any of these few things from our daily lives. Perhaps you will finally realize your argument isn't worth a hill of


Forking, posted 19 Sep 2007 at 18:20 UTC by johnnyb » (Journeyer)

bi --

One thing that Larry is missing is that the article is about whether or not Linux will succeed on the desktop. Note that this question is independent of whether or not choices are better in the long run than no choices. If freedom were to mean that Linux won't make it on the desktop, I think that's a discussion that we should be open to and acknowledge the problems. Of course, my own opinion as mentioned above is that what we need are for the distros to become more fragmented instead of less.

Sheep, posted 19 Sep 2007 at 19:47 UTC by ncm » (Master)

People use what is put on their desk. Put Windows, Macosix, or Linux there and they'll use it, and complain about it, but will never conceive of changing it in any way. Anything that they can't complain about, they'll forget entirely.

Worrying about that only distracts us from writing software that meets our needs, or is insanely cool if you know what it's really doing inside. I find I get most of my satisfaction from details of coding that are entirely invisible to users. Lately, that means using binary-search on sorted arrays, and contiguous "heap" structures, both in preference to binary trees; and using pipes to communicate between threads; and storing strings with '\n' termination so I don't have to append it separately when writing them out.

I would disagree with your argument ..., posted 20 Sep 2007 at 09:45 UTC by Chicago » (Journeyer)

Of the point of too many distributions. One of the problems is that there are so many distributions that are different to each other (and with different management systems on top of that) that advice on the Internet can be misleading - especially to a new user or administrator.

Sure, as the administrators get more competent, they get more competent in searching specifically for their distribution or going to the right places to look for answers, but even so, I occasionally get issues on my computers where the only answers I can find on the Internet need to be "ported" to my install - don't start the argument about "well admins should be willing/able to do this or they shouldn't be admins" because thats counter-productive and theres a big difference between an admin of a box on the Internet and an admin of a local file store behind a firewall.

I think you misunderstood the argument, posted 21 Sep 2007 at 11:39 UTC by johnnyb » (Journeyer)

"that advice on the Internet can be misleading - especially to a new user or administrator."

But note that I said that they should NOT call themselves Linux. You remove this whole issue by getting distributions to stop attaching Linux to their name. Using Linux is just an internal design decision. There's no reason a user should care what kernel you are using.

Subject/body mismatch, posted 24 Sep 2007 at 20:45 UTC by ncm » (Master)

The title of the article doesn't seem to have anything to do with the conclusion, nor the conclusion with the rest.

Suppose the mass of ignorant customers are girt about by turnkey boxes hiding Linux within -- cellphones, routers, video editors, set-top boxes, cars, refrigerators, what-have-you. Ask any of such customer what they run on their desktop, and they'll still say Windows or Mac. You can turn any general-purpose computing device into a closed system emulating a specialized device, but that doesn't do any good for the recognition of the role of Free Software in their lives. Every scrap of Linux penetration that isn't obviously Linux actually interferes with acceptance of "Linux on the Desktop".

letter to mr wolfe, posted 27 Sep 2007 at 18:12 UTC by lkcl » (Master)

dear mr wolfe,

this response can be seen on, where you might find some interesting responses to your recent article.

here's reasons why linux take-up will begin in earnest.

users are rebelling against vista.

IBM are selling 850 watt PCs whist koolu and linutop are selling 5 watt systems that do the same job (internet, multimedia, office, print, email ... er.... that's about it! want games? get a frickin console). (for completeness: zonbu do 15 watt systems; compulab in israel also do mini systems ranging from AMD Geode up to Pentium-M, all of them really low power)

discerning customers who read newspapers about the polar channel now being open in the arctic are getting increasingly concerned about the environment, and don't _want_ 500 to 1000 watt machines across their world racking up power requirements, where there are already brown-outs in some first world countries.

the recycling cost of machines - including second hand ones! - are expected to be borne by the sellers under new WEEE regulations which came into effect across the whole of europe.

the margins for EVERYBODY right from the factory through to the shop floor is a pitiful TWELVE PERCENT; the only way for stores to make money is to add a 50% markup for a 3-year-warranty and play the numbers game.

microsoft gets paid up-front UP FRONT by maufacturers for every copy of windows put on a hard drive - an insane $90 last time i heard.

so manufacturers are beginning to rebel as well.

dell have broken the ice with direct-from-manufacturer sales....

you do the math.

forget "pretty" - forget "under 2% market share so nobody must care". aside from which, 2% is in the "snowball effect" bracket.

all the indicators are that it's game over for microsoft, especially since they got into bed with disney and bond studios over the insane content protection required in HARDWARE for god's sake, at an unprecedently high level that far exceeds that used at the HIGHEST level of US military classification.

so - all the indicators - from all quarters - tell a story that's radically different from the one that you're portraying on your article.

which, as several people on advogato point out, seems incredibly out-of-date. but... that having been said, we do know where you're coming from: you're expressing a widely-held view, and in that regard it is extremely useful that you've pointed out that monopolistic-encouraging widely-held view.

i think you might find it interesting to work with the free software community to educate your readers, and explain to them what it is that they need to do, if they want to take a stand like we do.

free software people draw the line in the sand and don't really walk across it, irrespective of whatever pain or financial cost. my dad thinks i'm literally insane to not want to use microsoft products - but i Do Not Want To Support Proprietary Software Companies - it's as simple as that.

i'm writing an article outlining a bit more about that - perhaps you'd be interested in publishing it in your magazine.

hope to hear from you,


innovation, posted 27 Sep 2007 at 18:17 UTC by lkcl » (Master)

the constant "let's catch up with microsoft" game means that true innovation is dead before it's even begun.

"beyond the desktop metaphor" is a fantastic book that explores computer interaction in far-reaching and much more intuitive ways.

i don't _want_ a linux desktop.

i want a linux _innovation_ that's truly worth the time of day to use in the 21st century.

in the mean-time, i'll make do with recommending google apps to schools and businesses.

Who cares about the OS?, posted 29 Sep 2007 at 14:24 UTC by abraham » (Master)

The OS doesn't matter, it is just a foundation for running applications.

And the one application (suite) that matters is MS Office. You expect a new employee to be able to use MS Office, and your contacts expect you to be able to read MS Office files. It doesn't matter that there is "almost" compatible suites with regard to both file formats and user interface, any trouble will automatically be your fault for choosing a "weird" suite.

This leaves only two options, MS Windows for the masses, and MacOS/X for those who are willing to pay extra for being "different" in a way that doesn't matter.

"The year of Linux on the desktop" will only come after some really big player, probably only a government would be large enough, mandate a non-MS office suite on their employees desktops.

1. Prohibitive application porting costs

The listed applications matters little, as they don't represent common file exchange formats.

2. The Fanboy alienation factor

Yet another overestimation of the fanboy effect. MacOS/X is making inroads, despite even worse fanboys. It really isn't a factor.

3. You can't make money on the operating system

Plain silly, lots of people do.

4. Resistance from average users

Yes, there is always a resistance to change, especially when it means accumulated skills are becomming worthless. This is not special for Linux though.

5. Linux is "simple"; Windows "just works"

Yes, by being pre-installed. Not really a factor once the market is there.

6. There are way too many Linux distros

Doesn't matter, people will happily use whatever come pre-installed on their PC, once they can be sure their applications will run. The details of the distribution will be part of what the PC dealers use to distinguish themselves.

7. No powerful evangelist for Linux comparable to Bill Gates or Steve Jobs

Gates is hardly an embodiment of charisma, and the sales figure of MacOS/X put a hard upper limit on the number of users willing to let personal charisma determine their choice of OS. MS Windows wins by being the default choice, because people simply don't care.


And the Advogato story submitters reason:

8. Lack of specialized Linux distributions!

Specialized distributions are fine for specialized purposes, and Linux is already doing fine many such places, but for the desktop people expect to run general purpose applications.

things are simpler, posted 3 Oct 2007 at 04:14 UTC by groom » (Journeyer)

when a cutomer buy a PC, he has window$ shipped within. He does not even know it is a software, he thinks window$ is a hardware bound to the PC. So, to pu linux on the desktop, the only solution is to sell OEM linuxes. My sister has ubuntu, my father have ubuntu, friends have ubuntu or CentOS just becaus i installed it for them, and they congratulate the OS and never want to return to XP or vista. All those linux distros wer not OEM, i had to take an hour to configure it on each machine. but now they enjoy. If PC resellers could put OEM ubuntu with universe and multiverse repository, all would begin...

google apps + koolu / linutop / zonbu / fit-pc / etc., posted 13 Oct 2007 at 17:22 UTC by lkcl » (Master)

"And the one application (suite) that matters is MS Office. You expect a new employee to be able to use MS Office, and your contacts expect you to be able to read MS Office files."

that's why google apps is so damn important - it breaks the ice. google is one of the companies with the resources to pull off a coup of that magnitude - someone you can trust (or, more specifically, complain at if things go wrong or don't work).

the fact that instead of an average of 150 watt per pc, google apps is cpu-friendly enough so that you can use tiny 5 watt computers (or 10-15 in the case of zonbu) such as koolu, linutop, zonbu, fit-pc or any other industrial embedded pc such as IEI World with their wafer lx2 product, is just icing on the cake.

Re: google apps + koolu / linutop / zonbu / fit-pc / etc., posted 13 Oct 2007 at 19:00 UTC by bi » (Master)

that's why google apps is so damn important - it breaks the ice.

No it doesn't. Because when you go right down to it, Google Apps is just as closed-source as its Microsoft counterparts. Playing "meet the new boss, same as the old boss" isn't exactly my idea of promoting free software.

not "your idea"..., posted 14 Oct 2007 at 08:33 UTC by lkcl » (Master)

bi, hi,

the people who don't care about ms office being proprietary also don't care about google apps being proprietary.

it's a web service. you can't have everything - not all at once. we'll get there, don't worry.

Is the statistics still true?, posted 3 Nov 2007 at 08:45 UTC by audriusa » (Journeyer)

From that paper Linux seems not very popular on desktop, but how do they collect that statistics? Near all machines in Europe are sold with Windows pre-installed; even I was forced to buy such despite I never boot that installation. My friends put another hard drive and install Debian or Fedora there because they need them to be part of the Free world. At work we install SuSe and Red Hat at least because we use ssh -X heavily, but many machines retain an unused partition with W-XP Home which was the pre-installed OS. Not to be, by chance, that we all are still counted as Windows users?

My Take, posted 5 Nov 2007 at 21:30 UTC by nymia » (Master)

This is an article, mainly due to a class I'm attending. So I'll try adding some of my thoughts here as an exercise for the class.

First, the Linux desktop is already ambiguous because the desktop market has been segmented into several parts, each part has its own set of attributes that is addressed and targetted by your friendly neighbor--the marketing man.

Secondly, the market is way passed growth stage and in my estimate has passed maturity and is now entering decline. What this means for those who want to enter the market is that profits are going to be slim simply because highly-efficient manufacturing and delivery systems have been erected deterring newcomers from dominating. Unless the _newb_ has super-cali-fragi-efficient-systems that beat the existing ones, I'd say things will be on the side of those who own and operate these mega-structures.

Fourth, mega structures in the form of customer relations has already been established many years back, by big players like IBM, MS, Sun, Etc. These relations also act as a deterrence from new players entering from the Linux side, though.

Fifth, and this is important. Customer relationship is key to gaining market share, that's a big item for these incumbents, making sure customers stay in that relationship. The battle will be fought in this area, mining and undermining relationships is the game.

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!

Share this page