Prediction: Linux will Eat Microsoft on Most Desktops

Sometimes, several unrelated changes come to a head at the same time, with a result no one could have predicted. The PC market is at such a tipping point right now and the result will be millions of Linux-powered PCs in users' hands.

Or so opines Steven J. Vaughan-Nichols of DesktopLinux.com.

Vaughan-Nichols sees the following changes as basic to and causal of a sea-change in the desktop world.

1) The continued maturation of dektop Linux, to the point where "no one can argue with a straight face that people can't get their work done on Linux-powered PCs

2) The development of the "$100" laptop (available at stors for a little as $199.95).

3) The subsequent development by others of nearly-as-cheap PC's allowing one to build a desktop for about $200 (presumably without a monitor), as well as mid-range PC's costing just a few hundred dollars more.

4) The ease and ubiquity of broadband internet connectivity.

"Four trends ... Put them together and you have a revolution."

The previously entrenched pattern was to spec-out 1-2 thousand dollar desktops and run Windows on them. So let's say you need 100 computers, each will coast 1,500, and you are going to put Windows Vista, Microsoft Office, and a few other applications on it. No, wait, you will need a 2,000 computer to do that. Let's assume the software costs a total of $1,000. So we are up to about $3,000 per desktop, give or take. What did we need? A hundred of these? That would be about more than $200,000 conservatively estimated.

Now, compare that to a $500.00 desktop running Linux and OpenSource applicaitons. That would be about $50,000. The difference? Two or three employees, a huge bonus, a couple of vehicles, whatever. Not small.

Vaughan-Nichols uses different calculations and makes different comparisons than I do here, and you can have a look at his post, here. He concludes:

So, by my calculations, all those trends have joined together to make a Linux-based small business using Google applications instead of Exchange and SharePoint cost less than half its Microsoft-based twin.

Worse still, if you're Microsoft, you can't really defend yourself. Linux desktops run just dandy on low-end, under-$500 PCs. Vista Basic, which comes the closest to being able to run on these systems, is unacceptable since it doesn't support business networking. Office 2007 also won't run worth a darn on these systems. And somehow, I can't see Microsoft optimizing its applications to work with Google Apps instead of Exchange and SharePoint.

Tags

More like this

If the difference between success and failure in your business, as the economy comes crashing down around you, is money, and you have ID demands, consider this: With both people and companies having to squeeze a nickel's worth of good out of every penny, how long do you think people will be paying…
It could be called the Vista Bounce. Here is how it works: First, you have to understand that a good desktop computer costs about $3,000 and always will. A good laptop costs the same. If you believe differently either you have been believing the ads or you do not know what a "good desktop" or a…
This is one of four related posts: Should You Install Ubuntu Linux? Installing Ubuntu 16.04 LTS How to use Ubuntu Unity Things To Do After Installing Ubuntu 16.04 LTS Some Linux/Ubuntu related books: Ubuntu Unleashed 2016 Edition: Covering 15.10 and 16.04 (11th Edition) Ubuntu 16.04 LTS Desktop:…
Wine is the software that runs in Linux which, in turn runs Windows programs. I have experimented and I could not believe how incredibly well it worked. At the time, I was able to achieve greater stability and performance, and an easier install, for Microsoft Office and Endnote on my Linux…

I doubt it. Eventually MS will go the way of all big companies, but I don't think Linux is going to be the turning point anytime soon. What is ignored in these calculations is the fact that most businesses have to consider not only their own systems but the systems of their customers and those for whom they are customers. For example, when we made viewgraphs in my business 20 years ago, we did them as transparencies made on Apples and delivered hardcopies. It didn't matter which system we used. Now we deliver the Powerpoint files to our customers, who read them on their PCs running a MS system and Office. So we use PCs running a MS system. The same thing applies to many, many companies. Despite claims of compatibility of other applications, we find that even within the MS family there are enough incompatibilities that no one is likely to risk using an application from a different source.

I also doubt it. No matter how capable and mature these flavors of Linux and Unix may be, none of them have anywhere near the graphical end-user experience that Windows or, especially, Mac OS X does. Until there is a Linux that is as graphically oriented and easy to use as Mac and Windows, there is no chance that you will see a wide adaptation of Linux in the office or the home.

Corporations and governments do not select operating systems on the basis of cost, reliability, or any other practical criterion.

For corporations, the winning system is the one that offers the best 'free gifts' package to the few people who have the authority to make the decision for everyone else.

For governments, the reward for making the right choice is a cushy job to take at retirement, and then to soon retire from with all the perqs.

Ruprecht: I would love to tell you that you're wrong. But I can't.

People have been telling me this for ten years. I don't see it as really being any more true today than it was ten years ago. Ten years ago Linux ran on cheaper hardware with cheaper support software, and Microsoft had looming problematic product upgrades on the horizon that threatened to screw up networks and compatibility. What's different now from then? Ubuntu's nice, sure, but Ubuntu-like levels of niceness have been available in certain places for years.

Really, much as I'd like all this to be true, I find it harder to believe now than ten years ago. Ten years ago there seemed to be momentum-- there was, at least, a bunch of new products and companies starting up in the Linux space and an unruly horde of slashdotters screaming for change. Now all the Linux software companies are happily calcifying in the server space and the slashdot horde is too enraptured by their XBox 360s to care about free software.

As far as Google Office Apps go, is anybody actually using those in an actual commercial business? Anybody at all?

Again, I'm not trying to say Linux shouldn't be taking over the desktop space, I'm just skeptical it will. I think you have to do more to create change than just hope for it.

Coin: I agree with you about Google apps, but I think you have it totally wrong about Linux, with all due respect.

Microsoft is doing the same thing it was doing ten years ago, but an order of magnitude worse. Linux is not the same as it was ten years ago at all. It is much, much more regular-person friendly, and the productivity apps are very mature.

The current desktop market share of Linux ten years ago was a fraction (percentage wise) of what it is today. Today, Linux is approaching numbers that Apple has had in the past when it was very viable but also very much not Windows.

I don't know about a total turnover to linux. Maybe, maybe not. But its market share is growing now like it has not in the past.

There was a time when the idea of Linux taking over the server market from Sun and "professional" Unix systems seemed very unlikely....

The document-format lock-in mentioned above by Mark P is, I believe, going to be the very largest stumbling block. To too many people, a presentation is a PowerPoint document, and a word processing text document is a Word document. Inroads are being made, but not very fast.

Re: Google Aps, some folks here at Linden Lab use 'em. Some folks also use OpenOffice.org files to sling things around.

Of course, up-front costs are only a tiny (almost completely insignificant) part of the TCO... If you could make a really convincing argument about large-scale manageability (say, 10,000-100,000 desktops spread over 50 sites in at least half-a-dozen different locales) then you might be getting somewhere. (Note: I'm not saying this argument can't be made. I don't know. But neither Greg nor Steven even mentions TCO...)

Also bear in mind that many large enterprises are still running NT4 as their standard desktop, and have very substantial investments in bespoke applications written for the Windows platform. Your $200,000 desktop replacement bill is roughly equivalent to the development cost of a single medium-scope enterprise application (of course, 100 desktops is a pitifully tiny roll-out). Most decent sized enterprises will be spending millions of dollars per annum just to support their existing business-critical applications.

If you think that the up-front cost of the desktop platform is a significant factor in corporate IT decision-making (at least beyond the small end of the SME market), then you haven't got a clue what you're talking about. It's peanuts.

Oh, and your $50,000 saving? That's (barely) a single month's on-site cost for a half decent IT contractor. Heck, I've met guys who would charge that in a week. I've seen corporates hire dozens of contractors at that level for six months at a time, and then have 'em all sitting around with nothing to do.

Dune:

Nope. I work for a very very large entity. We have a zillion destop computers and some very, very expensive enterprise software. We could change each and every desktop to a Linux box tomorrow, and this would have no effect on the enterprise software. This stuff is all delivered to the desktops via web clients. Indeed, although the office I'm in now has a windows machine, in the last office I had I wiped out the Windows installation and replaced it with a Linux installation and for two years inter-operated with everyone, every thing, at every level, and not one person or unit knew I was using Linux on that desktop unless I chose to tell them.

We're huge, but we are probably pretty typical in this regard.

To give one little piece of the puzzle - I'm a statistician. The biggies of the stat world are SAS, SPSS, all others (including R). I can't run either SAS or SPSS on Linux. That means that I can't use a Liinux box, until the day when the stat world switches over to R, which is probably 20 years from now.

Note: "This stuff is all delivered to the desktops via web clients. " is the way that statistics software *should*, especially for SAS, which is a huge set of programs. However, it's not.

This stuff is all delivered to the desktops via web clients.

[...]

We're huge, but we are probably pretty typical in this regard.

No, you are not typical in that regard. Not even close. I've contracted for everything from fairly minor government departments to major multinational FSOs, and that's a scenario I've never encountered. Most enterprises I've worked with have a largely unplanned hotch-potch of client-server apps, distributed n-tier apps, web-based apps, rubbish little VB apps (often developed at the departmental level, with no oversight from IT), absolutely horrible MS Access "apps", and a huge amount of business critical, legally mandated data buried in unstructured documents (mainly Word and Excel) stacked up on file servers and in Exchange. Also, most won't even let you install a new text editor, never mind a new operating system. Heck, many places won't even let you change your desktop colour scheme.

I also have to ask: do you actually know about every piece of software used by every single person in your organisation? While all the enterprise software you use may be web-based, is the same true of Accounts, or HR? Is all your business data (from end-of-year financial statements to sick returns) available in open formats? Really?

And none of this changes the point that I was actually making, and that you seem to have completely ignored: that up-front desktop installation costs are such a small fraction of TCO as to be insignificant, therefore any argument based solely on those costs is irrelevant. Re-training alone will typically cost you ten times as much. (Remember, most places have a significant number of staff who needed a five-day course to learn how to use Outlook, and who go into a blind panic if you change the layout of their toolbars.)

Seconding Dunc here - I have experience at fewer places (some corporate, some academic, medical), but it's the same story. And Microsoft Office is predominant to an extent which would make old-time IBM or GM execs die from sheer jealousy. And there's usually 20 years of specialized software which is in daily use - the sort of stuff which gives enterprise system installers/programmers steady jobs and ulcers.

Dunc:

I'm intimately aware of the details of my organization, having served on all the major committees and been involved with the process. It is likely, as you say, that a giant university is not typical.

But listen to yourself, man! You are describing a nightmare and using your description of the nightmare as evidence that the nightmare will never end! Does that not depress you?

I'm not saying it will never end. I'm just saying that the end is not imminent, nor will it be entirely straightforward. And above all, I'm saying that the end will not be driven by lower up-front desktop installation costs, which I believe is the principle subject of this post.

There are perfectly good arguments to be made in favour of both Linux and F/OSS in general. Lower installation cost is not one of them, at least in the corporate world.

Oh, and as for whether it depresses me - it keeps me in a job. A very comfortable, well paid job that can occasionally be quite rewarding, but one that I certainly wouldn't be doing if it weren't for the money. Whether that's depressing or not depends on what sort of a day I've had. But the real problems have nothing to do with the underlying software platform, and everything to do with people and organisational politics. Changing OS will not fix them.

Dune: OK, OK, no low end desktops...

But, what about thin clients??? :)

One thing I'm sure of: There have rarely been accurate predictions about how change happens, and what direction it goes in, when it does occur in IT.

Thin clients? I've been hearing that one for most of my career. Actually, it's been most of my career... Sure, there's a big niche there, but client/server (and just plain desktop) ain't going away - thin client is a fundamentally flawed approach for certain kinds of task. And there's no getting away from that massive installed base, at least not in any kind of a hurry.

Look, I'm perfectly happy to agree that ultimately Microsoft is doomed. Just like IBM was. But IBM didn't lose their position because somebody came up with a cheaper mainframe - somebody made the whole mainframe concept obsolete. Whatever wipes out Microsoft will probably make both Windows and Linux look like electric typewriters. Until then, their massive pre-existing dominance more-or-less guarantees their continued success, at least in the short-to-medium term. (Remember, many people were still running their old IBM mainframes well into the late 90s - some probably still are. I believe you can still make big bucks as a COBOL consultant.)

Oh, and it's "Dunc" (short for "Duncan"), not "Dune". No offence taken. ;) Sorry if I'm losing coherence too... :) hic

If I may...

Linux, on the whole, does not have a prayer of taking over the desktop. As long as IT departments have a decent amount of control over the software they run, Linux may in fact retain and expand its dominant position in the server room, but should any version of Linux become significant on the desktop, it will become THE Linux. Outside the power user community, GNOME vs. KDE vs. whatever else, version numbers, etc, in other words the very things that make Linux the superior solution for cutting edge software development, hobble it for end user systems. "Linux" is not an end-to-end solution; something like Knoppix, or Fedora, or Ubuntu is, and has more weight as a whole than a single component of the system.

The differences can be readily evened out of course; all these disparate APIs are more or less interchangeable. GTK apps can run in a KDE environment; ALSA apps can run happily alongside Linux-native apps. But the flamewars that currently rule the world of Linux preclude this.

I like Linux, but we do have to be realistic.

OK, so we'll just say ... "Prediction: Ubuntu will Eat Microsoft on Most Desktops"

I'm fine with that.

There's a point at which Linux could conceivably come from nowhere and become the dominant OS. However, IMO we're years away from that point for reasons that Duncan and others have outlined.

OTOH, SJVN and Greg both are right about how the capabilities and the market conditions have changed, far more than many have realized. For example, take a look at the sales figures for the eee and the gPC. The eee is literally flying off the shelves as fast as Asus can build them. Asus hope to sell between 3 and 4 million of them in 2008. Everex, the company that built the gPC for Walmart, has announced plans to release a small laptop for roughly the same price as the eee. They clearly mean to steal some of Asus' thunder if they can.

Meanwhile, the gPC sold out its initial run within 2 weeks! There's clearly a huge, pent up demand for low end solid, reliable PCs. Linux based boxes can certainly fit that bill far more easily than Vista or OS/X boxes can.

OT(gripping)H, both SJVN and GL have made a logical leap that I think is premature. Just because people desperately want something like this at home does not automatically translate to immediate demand for corporate IT.

In my view, I expect that we'll begin seeing people sneak Linux based boxes in the same way that they sneaked in Microsoft DOS PCs 20 some years ago and for the same reason: They want something that Just Works, and that works with what they already have at home.

However, building up a consumer base of these Linux boxes will definitely take time and introducing them to the U.S. corporate world will take even longer. (Europe already has several fairly large implementations of Linux desktop to use as examples, and parts of Asia are also headed down that path.) If I were a betting man, I'd say that we're still 3 to 5 years away from a time when suggesting a Linux rollout for anything other than a tightly controlled, limited desktop. Think POS terminals, call centers, etc.

Sorry, I meant to mention South America as well as Asia as starting down the trail that Europe has already blazed. In fact, it's not too much of a stretch to say that the adoption of Linux desktops (FLOSS in general to a lesser degree) is one area where the U.S. is definitely lagging most of the rest of the world.

As for the $500 budget computer thing: Yes I agree, the Linux OS works much better than Vista or XP on low end or budget PC's. It wouldn't surprise me if Linux completely took over all 2nd-hand or dated machines within the next 20 years. However, the crux of the issue is not really about the OS, but about the physical requirements of the software. Take this for example:

Sony Vegas is a high end video editing tool, requiring lots of RAM and processing power. Linux can obviously not run Sony Vegas because it's a Windows program. However, if Sony suddenly this very instant released a version of Vegas that was compatible with Linux, it would become obvious WHY the average price of a computer today is over $1000.

On a budget machine, Linux (or any OS) would be completely incapable of running such a program in timely or useful manner. This is not because of the OS, but because of the hardware. Sure, Linux uses, say, 128mb less RAM to run the OS than XP or Vista does (I made this number up), but in the scope of things that doesn't mean anything. For a $500 machine with only 512mb RAM, freeing up 128mb will have AMAZING results. However, if you want to run a program like Vegas you'll need 2-4 Gigs of RAM, and the 128mb that Linux would free up would no longer be noticeable and the experience would be exactly the same as a somebody using a Windows OS. While Linux runs great on low-budget PC's, you will need something bigger and better to use the more complex programs around today REGARDLESS of what operating system you have. If Vegas was available for Linux, do you thing movie editors and studios would rush out and purchase a bunch of $500 PC's? No, they'd go for a souped-up $1,500-$3,000 PC so they could run Vegas and other high end programs well. Once again, it's not about the OS, but the software. No matter how efficient and stable Linux is, if the program says it needs 2 gigs of RAM, than it needs 2 Gigs of RAM.

However, at the moment Linux cannot run Vegas or many other high-end, CPU/RAM intensive programs. This is why it can get away with the "budget PC" headline. As of now Linux doesn't really have a use for a 3.6Ghz Quad-Core processor, so there's no point in getting one since a 800mhz or so will do everything Linux can do just fine.

If major software titles like FL Studio and brand-new games like Bioshock or Crysis had dedicated Linux versions available, a $500 computer is no where close to what would be required to run them and you would be looking at prices close to the ones you listed for a Vista desktop in your article. Remember, Microsoft doesn't make computers: It makes software. If you want to run certain programs, the price of the computer is going to be the same regardless of what OS you choose.

Andrew, those are good points, but there are two other things that should be mentioned. Because of the way *nix handles memory and various other factors, if you need a real application, like the app that draws the box around the strike zone in MLB and computes where the ball goes, etc., you can't run that on windows, on any machine, because you'd be insane to develop such a demanding application for a system that traditionally will not use more than a certain amount of system ram, and taht won't use 32 or 64 bit architecture (all depending on where in computer history we make the comparison). In other words, it has always been true that the really demanding apps do not run on and are not developed for baby computers like Windows machines.

The other thing is actually two parts: If you need lots of ram and other system resources, no matter what the amount you have on your hardware is, you will always get more on a Linux box than a Windows box because Linux does the same job as Windows (only better) with much much less demand on resources for the system. The second part of this is that software that runs on Linux typically uses less code. A given program may or may not use less code (but usually does) but if you run two or three programs interacting with each other there is a huge savings if they are using common libraries on Linux, and making use of the system. That just does not happen too much in Windows.

This is something that Microsoft actually touts as part of their design strategy. Bill Gates has been known to say that it does not matter if the applications and the system are sluggish now, the hardware will catch up.

The two systems are fundamentally different in philosophy in a way that will always result in much, much better software on a Linux machine than on a Windows machine given the same development effort.

I understand the "less lines of code" thing and big, bloaty programs are definitely a big flaw for Windows.

I'm kind of confused about the MLB strike zone thing though... I think you are saying that they use Linux computers to calculate the batters strike-zone in real time during games, and they use Linux because Windows just can't do the job. Is that what you're saying? Anyway, while thats very cool I was specifically talking about the $500 Linux PCs that you talked about. While I'm sure there are many amazing super-Linux-computers out there, I don't think the $500 ones will be able to do that (and if they can I'm going to rush out and buy one right now, haha)

My point wasn't about the OS. I know the Linux OS is more stable, smoother, contain less code and is more efficient but my point was that I don't think its going to matter. Even if the Linux was some genius, unbelievably fast and efficient marvel of engineering, if a program requires 1 Gig of Ram and the computer only has 512mb, Linux (or any other OS) will not be able to run it.

That was what I was pointing out. I'm sure the Linux OS will run nice and smooth on the $500 PC you were talking about, but what about hardware limitations? Programs and games are using more and more RAM and CPU power every year and they aren't going to stop anytime soon. Linux will run well on a $500 PC, will new software be able to do the same? Once again this isn't an issue with the OS, but the hardware. Linux can't just "try reeeeeaaaallly hard" and push its processor to go a little faster: It's locked in. Same with RAM. Once they are reached the OS can't really do anything about it.

Andrew, you do know that all those Hollywood movies (at least as far back as Shreck and Titanic) do their rendering and CGI on Linux these days (and SGI before that), right? It's NOT a question of whether Linux can do it as well as Windows. Whatever hardware you have, Linux can use it more efficiently. So the really "high-end" video is done on Linux already, partly for performance, partly for cost.

The real stumbling block is that certain corporations -- which probably should have known better -- chose to develop their applications (eg. PhotoShop or AutoCAD) as OS-specific, OS-dependent software. Worse, they have developed a sort of tunnel-vision, trashing Linux and clinging desperately to the old familiar ways, rather than learning to apply their specialty knowledge in a new environment.