Why didn't Linux win on the desktop?

Why didn't Linux win on the desktop?

Summary: Heady forecasts showing Linux would become a major player on the desktop, on the server, and in embedded systems were published back in 1994. While the prognostications proved correct for server and embedded operating systems, what happened on the desktop?

TOPICS: Mobility

While waiting for the snow plows to clear the streets, I spent some time thinking about all of the heady forecasts we in the industry research community made back in 1994 about how Linux would grow and take a large share of the shipments of operating systems.

Although the forecasts were based upon the observed very rapid growth that the operating system experienced in the server, client, and embedded operating system markets, things didn't turn out as projected.

Those early forecasts

Based upon the rapid growth seen back in 1994 (Linux server operating system shipments grew almost 200 percent and Linux client operating system shipments grew almost 100 percent), research firms published forecasts that in five years — that is, by 1999 — Linux would hold a major share of all three markets (server, client, and embedded). Usually, the forecast methodology took into account the following points:

  • The architecture of Linux was very similar to the well-established Unix operating system

  • The cost — as an open-source software project, Linux was available at no cost if someone was willing to download a community-supported version of the software

  • How many Linux distributions (over 400) and OEM partners (over 20) were available at the time.

Industry researchers felt that it was very supportable to declare that Linux was on the path to be an industry winner.

What actually happened?

As projected, Linux went on to take a large share of the server and embedded operating environments. I won't steal the thunder of the industry research firms by publishing their forecasts. Let's just say that in those two markets, Linux more than established itself as a leading player.

Although much ballyhoo was made about Linux becoming a major player on the desktop, that prognostication didn't prove true.

While there are quite a number of developers, academicians, researchers and content creators doing their work on Linux desktops every day, the overall share of the market never matched the early forecasts.

Let's examine the desktop Linux inhibitors

Desktop systems are deployed as application or workload delivery tools. For the most part, they aren't purchased merely to run an operating system. This means that a customer first seeks out the applications and tools needed and the selects a system that supports them.

For example, if a customer wants to use Microsoft Office, the selection of operating systems would be limited to either Microsoft's Windows or Apple's OSX. Microsoft never made a version of Office available for Linux. If specific features of Excel are needed, say a robust implementation of pivot tables, then the choice degrades to Microsoft Windows or Microsoft Windows.

Even though there were credible alternatives to software products such as Microsoft Office, there are enough issues to deter any but the most enterprising customers. (Remind me to tell you about an obscure presentation formatting issue that drove me to abandon OpenOffice.Org and come back to Office some time.)

Furthermore, enterprises have a strong tendency to pick a path and stay on that path unless there is a very strong reason to change. They consider the costs of replacing systems, software, established procedures and training their staff and usually decide to continue to do what they were already doing.

The fact that companies are still holding on to Windows XP and Windows Server 2003 is evidence that supports that view.

Another point is that there isn't a single version of Linux for the desktop nor is there a single competitor. Desktop Linux is like a smorgasbord not a product. There are many different user interfaces, file systems, and ways to install and update a system. While proponents would say that this is a strength of the operating system, enterprises and many individuals saw this as a confusing mess.

Furthermore, there isn't a single proponent of Linux or a single set of messages. This makes it difficult for decision-makers to get a single, simple view of what's happening in the world of Linux.

In the end, Windows was and is still the leading operating system for desktop and laptop computers.

Is that the end of the story?

Even though there was a point back in 1994 in which Linux passed Mac OS to take the number two spot in the client operating system market, Linux didn't hold onto that spot for long. In the long view, Linux didn't win on the desktop; but is that really the end of the story? Other forces came into play that have reduced the influence Microsoft's Windows has in the lives of end users.

As we know, the world has been changing rapidly and this certainly is an important factor to consider.  Desktop and laptop systems are no longer considered the only way to use applications. Smartphones and tablets are increasingly seen as the platforms for applications customers are using.

If we stop and look at the winners in those categories we see that Linux (in the form of Android) and Unix (in the form of IOS) are leading in market share.

Desktop hosted applications are no longer the only choice. Customers increasingly were happy with the applications and network access available from handheld devices and didn't feel the need to also use a Windows-powered device.  After all, Web-based tools, such as Web applications, email, collaborative software, and search, can be easily done from a lower cost device. An expensive laptop or desktop may not be needed at all.

Is this how Linux and Unix will win over the mighty Windows? If the current trends are considered, the answer appears to be yes.

Topic: Mobility


Daniel Kusnetzky, a reformed software engineer and product manager, founded Kusnetzky Group LLC in 2006. He's literally written the book on virtualization and often comments on cloud computing, mobility and systems software. In his spare time, he's also the managing partner of Lux Sonus LLC, an investment firm.

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • it's a pretty easy question to answer

    50% no motivation on developers, 50% stuck on command line
    • That's not it at all...

      What happened on the desktop was that coders were not thinking
      Like users and that hurt them big time!

      All it would take was one player to enter the market that knew what they were doing and one player actually did! Crud, it took a bribe to get Corel to drop their Desktop Linux Ambitions before 2.0 was released because they were surely moving in the right direction.

      Another note, Server Apps were plentiful but, Desktop apps weren't as fruitful in terms of quality.

      Now, we have a whole lot going on with Linux and as much as it might not be the current target, the Desktop is certainly in play here!

      Hell, you've got...

      Luminance HDR and Tone Mapping Software
      Rawtherapee Excellent RAW image converter.
      Darktable and Lightzone development workflow tools (Think Lightroom)
      GIMP has improved
      Monitor Calibration Software to match standard colorimeters.
      Libre and Open Office (Not the best but, pretty good)
      Chrome and Firefox Web Browsers
      VLC for BD and DVD playback (Including encrypted)
      Ripping and Burning software
      Project Management Software
      Blender 3D modeling
      IDEs for multiple languages.
      Steam now has more than a 100 games and some of them are former A list titles...

      Seriously, there's more than enough reason to try Linux on the desktop and a sudden market shift could end up bringing about an interest in the products.

      Oh and ASUS is offering one of their 10.1" laptops with Ubuntu preinstalled for $269 and I'm seriously debating it. If it were baytrail, I would have already jumped.
      • I have to take exception to the notion that the GIMP has much improved

        If it had, Photoshop would not retain the traction that it has. The GIMP has had years to get its act together, and it is ironic that in the decade or so since people realized its UI really sucks, the freeware alternative that came by and surpassed it - Paint.NET - is a Windows only program!

        Even on the Mac, people forked the Gimp's impressive engine and replaced its UI (with SeaShell.)
        • No, Gimp really HAS much improved.

          I run the windows version of Gimp, which has really tamed the mess of the original Gimp with it's single window mode. version 2.8.10 is really quite useable for the casual crowd. The high-bits-per-pixel is still lagging behind photoshop, but this is also true of Paint.NET.
          D. W. Bierbaum
          • GIMP is greatly improved and it's usefulness is also about synergy

            Winduz apps have to be "all-in-one" because they cost a lot and running two apps at one time is still a pain after something like 28 years of Windows and pre-Windows OSes.

            In Linux, selecting an image in gThumb or some other viewing app, then moving it to RawThereapee for raw conversion, color tweaking and first stage noise reduction and then moving it to GIMP for some more sophisticated layer-based processing is seamless. I'd never try doing that in Winduz because it was designed to be a big monolith with big monolithic apps that don't play well with each other.
            D Soup
        • Why?

          Seriously, I never said it was a Photoshop replacement, I said it has improved quite a bit, assuming it tops Photoshop would mean that Adobe was standing still.
          • Photoshop is too ponderous for real life

            Pros may find the fine details of CMYK color space and 16-bit processing a big deal. I use 16-bit processing in Rawtherapee and then I don't worry about it when I go to the final stages of GIMP processing. I've had multiple versions of PS and never found them quick and intuitive to use and never ended up using them.
            D Soup
    • WTF

      Thats a bit rich, i spend probably %80 of my time in the command prompt and powershell, very rarely do i ever need a .msc or .cpl to access what i need. I Administer Windows servers only (Accept for Linux proxy servers) and live in the command line so maybe a rethink should be in order, or are you one of them people who replies on buttons and check boxes?
      • Lol

        Ok, YOU may spend 80% of your time in the command line. I do as well. Guess what? We're server admins. Things like cmd, PowerShell, and the terminal don't bother us at all. But the average end-user does NOT spend ANY time in a CLI. I've said it for years: Linux distros could take over if 3 conditions were met: 1.) make it possible to do everything without ever having to touch the terminal; 2.) The big-wig application developers need Linux-native builds (I'm talking about you, Microsoft and Adobe). Hell, if the Office and Adobe Creative suites came to Linux natively (without things like WINE to re-interpret), I bet Windows would lose at least 30% of it's market share to either Ubuntu, Fedora, or Mint. And 3.) Native Steam client and games.

        One of those 3 things have already happened / is in the process of happening now. And I can attest, of the native games that Steam has on Linux when compared to their Windows counterparts on the exact same hardware, Linux performs better in every category.
    • Don't forget

      The "me, too" aspect. Linux copies. If you want to see what Linux will look like next year, use Windows or OS X today.
      • Yeah right cause umm...

        It isn't like OS X had any roots in Open Step or Open Source right? I mean, we all know that Jobs invented BSD and Every GUI known to man right?

        I got news for you, there are plenty of open desktops out there and only a few resemble other platforms. Sure, all of them have similarities but, claiming one is copying the others without having any influence of its own is just outright ignorance.
        • ?

          Jobs played a huge role in the design of NextStep, when he started NeXT computer after leaving Apple. When Apple re-hired Jobs, they bought NextStep (with the OpenStep API, which Next had developed for separate sale) for $400m. OpenStep was developed in collaboration with Oracle.

          Xerox and Apple/Jobs did more to develop the modern GUI than any other company that I know of and the Xerox Star was a commercial flop.

          As for the various UNIX kernels, BSD is one popular, open-source one used on Macs. NeXT used a microkernel (IIRC). Linux is not a UNIX kernel, although GNU/Linux is often described as being "UNIX-like."

          No doubt your heavy-handed sarcasm serves your aim of discrediting what Jobs achieved, but not even Gates changed the the computer world in so many ways.
          • Alan Kay surely

            Surely you're thinking of Alan Kay who came up with Object-oriented programming, the modern GUI (Windows etc). The Dynabook concept (still not matched by tablets/phones) etc.

            All Jobs was good at was commercializing those innovations,
          • Partly agree, but think you've changed the context

            However, there is a big difference between inventing something in an academic environment and making a commercial success of it - hence my comparison with Gates rather than Kay.

            I do not mean to denigrate the work of academics in any way, but since ZDnet is all about rather old (academic) concepts and the computing business race, I thought my meaning would be clear.

            (Kay's Dynabook still hasn't be built. Kay's work at Xerox and Apple in OO languages, including Objective C and Squeak had a big impact. Of course, the original Mac used Object Pascal, designed by Nicholas Worth, but that had a very different form of messaging, quite unlike the SmallTalk/Objective C form.)
          • apple rehired jobs? did not know that... when?

            i thought that apple was literally dead in the water when jobs (with enough mercy, energy, connections and capital) decided to save "his kid" and returned, turning a belly up ghost into a miracle (before Steve passed on apple used to have more market capitalization than msft + gogl)
            so, when did apple rehire Jobs?
            Dan Marinescu
          • December 20, 1996

            ... is when Apple made the public announcement that they'd be hiring Jobs back as a consultant reporting directly to the Chairmen (Gilbert Amelio). He later (and not much later, mind you) through fickle maneuvering and downright cold-hearted tactics ousted the CEO and had the board instate him as CEO in his place.

            Jobs was fired from Apple, plain and simple. During that time, he came to peace with himself, and then they begged him to come back. He didn't ride in like a knight in shining armor, Apple came to him and said "We need you, please please please come back".

    • "50% stuck on command line" ?

      Are you LoverRock? Linux has had arguably better GUI than Windows for longtime.
      • Linux has a better GUI than Windows?

        In what universe is that true? The command line IS the main hold-up for wider Linux adoption.

        You're a techie - you don't mind or probably even notice how often you shell out under Linux. If you were to present a Linux command line to the average Windows user, for any reason whatsoever, they would run screaming for the exit.
        • Screaming for the exit?

          I am a primarily windows user, I use the command prompts fairly regularly, in general you can do more in the command prompt by stringing things together. Most programs that run in a GUI will not do that. The GUI also takes up a decent amount of resources and bloats the software. Sure some programs such as GIMP need a GUI, but for general workloads like automated scripts that are strung together to execute almost simultaneously, a GUI is unnecessary and slows things down. GUI's have their uses, but they are limited. command prompts also have their uses, and in general are more powerful. It really does depend on the situation.
          Antony Clements
          • ...You Guys are WAY too techy to understand.......

            The *typical* Desktop user is a Data entry person of extremely nontechnical capability.
            We, who manage these resources, do not need or want an ecosystem that pushes learning curves into the equation. If you would dumb-down a bunch, you might comprehend the true needs of our environments.