Harvard Concludes Linux Will Remain Second Best

watzinaneihm writes "A Harvard Study which uses formal economic modelling to determine "Will OSS ever displace traditional software from its market leadership position?" came to a (not so?) surprising result. Linux is likely to remain second best as long as Microsoft has a first mover advantage."

Comments

OSX

[Transcendent]

Now that Macs are developing/supporting a BSD based OS, I think Linux will also lose some desktop share here as well.

In fact, I know of a few friends who chose to get a MacBook and keep OSX on it because they described it as "Linux with more hardware support" (or at least better support directly for the Mac). Not saying this is true, but that it is another well supported Unix alternative.

Re:OSX

[GigsVT]

While Linux advocates are a fiery lot they will probably agree that users switching to osx is better than users staying with windows.

Meet the new boss, same as the old boss.

Apple is worse in a lot of ways. While competition has driven them to use more open software, I don't view that as likely to remain the case if they were to become dominant.

Apple, in the past, always worked on the strategy of telling the user what's good for them and not giving the user any choices. They've only discovered fairly recently that open standards in various ways can be leveraged to give them a good competitive advantage.

Just remember, their OS comes with the ultimate copy protection. Palladium and trusted computing is only MS's attempt at doing what Apple has done since the beginning of time, making the hardware into a giant dongle for the OS.

Second Best Where?

[nbannerman]

Surprisingly enough, I'm finding the exact opposite to be true.

I've talked at length about how I deploy an entirely Microsoft-enabled enviroment for my college. 600+ machines, all running XP and Office 2003. 24 servers, all 2000/2003. A pretty typical Microsoft-enabled environment really.

However, I've personally just gone down the Linux route for my work laptop, and I'm giving projects like Edubuntu serious consideration for older, non-Vista compliant hardware.

I have no doubt that companies with ££££s to throw around will buy new machines that are pre-loaded with Vista, and they'll inevitably begin the Vista rollout come SP1. But big business is not everything; I know many of my fellow network managers in education are giving serious consideration to OSS solutions.

We're educating the business people of tomorrow, and if they are introduced to OSS at a younger age, I think we'll see some interesting changes somewhere down the line.

Well, I hope so... ;)

What's in a word?

[Anonymous Coward]

"Will OSS ever displace traditional software from its market leadership position?" came to a (not so?) surprising result. Linux is likely to remain second best as long as Microsoft has a first mover advantage."

The only problem here is that OSS isn't the same thing as Linux. Apache is OSS, but it's not Linux.

problem right now is that linux is unknown.

[Albert Sandberg]

When we (and by we, I mean the linux community) hit a larger portion of user base, say 10% of desktop market (if that will ever happen) linux is going to be well known, and I don't mean that just by the name, but people will actually from time to time use a computer that has linux installed.

Then and not until then will my mother think "why do I need this windows for anyway?" and might try linux out on the home computer. Then the kids start getting used to it (from home, school and most important, friends) and the adoption to linux REALLY hits, because no household will pay $$$ for an operating system if they know one that's usable for free. Not to mention the applications.

Alongside, user friendly distros such as ubuntu, mandriva and feodora will grow even easier to use (as a matter of fact, I think ubuntu is easier (and faster) to install than windows XP or 2000).

As long as ...

[140Mandak262Jamuna]

What kind of conclusion is that, "Linux will remain second as long as MSFT has the first mover advantage"?

It is like saying Tiger Woods will remain number one as long as no one comes along who is better. Or this guy will live as long as he does not die.

You need to go to Harvard to come to lame conclusions like this? Nah, you need to go to Harvard to write escape clauses like this. If Linux become dominant you just declare, "MSFT no longer has the first mover advantage, so I am right". If Linux fades to obscurity, you can go "See, I told ya, Linux will never become numero uno"

Re:As long as ...

[bhmit1]

It is like saying Tiger Woods will remain number one as long as no one comes along who is better.

Not quite. It's like saying Tiger will have more fans forever because he started with more fans today. And that fan base will ensure that Tiger is always better than everyone else.

The flaw in this article is that they assume:

• Companies won't demand open standards
• Every version of MS software will continue making significant improvements
• MS will not start to get caught with the viral GPL license issues the way open source developers continuously get caught with patents. The nice thing about patents is that they eventually expire, GPL doesn't.

What will really happen is the law of diminishing returns will kick in, and MS users will have even less of a reason to upgrade each time as more eye candy and unneeded features requires more hardware. At some point, the features that MS gives over linux will not be worth the cost of MS. Additionally, as formats open, and applications move to the web, the ability to leverage the monopoly will continuously reduce. The best thing MS has going for it now is application support and the bundling that is done by all the major PC builders. If they lose either of those, they will lose their grip on being number 1.

Rich Get Richer

[Doc Ruby]

Short version in English: Harvard says that because MS has more market share, it will have more market share.

Isn't that the thinking that kept IBM in control of computing in the 1970s?

Re:Rich Get Richer

[Klaidas]

Umm, it's not more, it's almost all market share.
And well, when Vista comes preloaded with almost all PCs, and there's no other version of Windows availible, guess who continues XP's generation?

And the moral is?

[FishandChips]

As with economists, you could lay all of America's business professors end to end and still never reach a conclusion.

Linux does not aim to be best, second best or ninety-third best. Take Debian: it aims to provide a free universal operating system. How well it does, in the perception of others, is only incidental to Debian's core purpose. So, looking at all this in terms of winners and losers or best and worst is largely illusory. Linux is doing just fine and does not have to hit some arbitrary bar - such as overtaking Microsoft's market share - to continue to do just fine.

Why does this kind of thing ilicit so much emotion

[antifoidulus]

here? You would think that to some people, knowing someone used a Microsoft product was akin being spit on. Why? I'm a mac user but that doesn't mean I really give a damn if someone else uses Windows, Linux or anything else. Hell, I use Linux at work. It's no skin off my back, my OS doesn't stop working because someone is using Windows.

I consider my operating system to be a tool, not a way of life, not something that defines me. Maybe that is why I never understood OS evangelism. Can someone please explain to me that when someone says "Linux will not be the most popular desktop operating system in existence" Linux users feel the need to sling such insults as "numbnuts"(which by the way is not very mature and not likely to win you very many converts) towards them?

I think you are wrong.

[jellomizer]

When ever I talk about purchases of computer and OS to bigwigs. It basically breaks down as LInux is a Free OS and could be considered good enough to do what they want to. What usually sowers the deal with Linux is the fact that the company usually has some software that is for windows only and moving off it is out of the question. Many times it is a CAD Program, other times it is some old custom app that cannot be replaced (Cheaply) and the people who made it are long gone. And on some other situation companies just went threw a painful migration from old Unix to Windows and they are not willing to go back to a Unix like platform for a long time (Even though Linux and newer Unix have far more to offer then their 1989 SCO box).

Re:I think you are wrong.

[Fred_A]

What usually sowers the deal with Linux is the fact that the company usually has some software that is for windows only and moving off it is out of the question.

I would dampen this somewhat by rephrasing it as "moving to Linux right away is out of the question". However a lot of companies with in-house packages are considering moving said package to an OSS/Free platform. It is however understandable that it's not done overnight. But in five years, I expect that a number of the ones I work with will have made the switch.

Commercial packages (such as the CAD software you mention) are a different problem which mostly depends on the market penetration of the OS. Although even then, some stuff will never be natively available to our platforms. While in some cases emulation through Wine can help, it's not something I'd wager my business on.

Those two cases are indeed often forgotten by the enthusiasts who blindly believe you can easily replace windows by Linux on any desktop.

And on some other situation companies just went threw a painful migration from old Unix to Windows and they are not willing to go back to a Unix like platform for a long time

I don't have many examples of those, but the few I've met usually hate themselves for the switch because the main factor was the price of Unix workstations and they willingly ended up with what they knew would be a less versatile tool to save money. And now they find out they could have saved even more money by sticking with Unix. Sucks to be them. ;)

Most of us don't care

[pembo13]

A lot of us aren't in OSS for the the ride to the top. I personally couldn't care if 1% of the population used OSS or 99%. As long as I have the freedom to use the software that I want when I want to, then things are fine with me. And _that_ is one of my peeves against the Microsoft Corp.: by the very nature of their marketing/functioning the people who use their software tend to be drones in that they know not how to function with anyone else doesn't have the dam 4 colored Windows logo all over them.

I like Linux and the majority of OSS tools that I use because I prefer them to their Window's based counterparts, with a few exceptions. I have found that explaining to someone that Linux is "better" than Windows is like explaining gold is better than silver - they have a jewelry box full of a silver and their minds just aren't willing to absorb new information on that topic - and why, they think they are happy with what they have. All that will happen is that eventually, I will not know enough of Windows to troubleshoot their machines anymore

Re:From TFA-Tortise-hare.

[tomhudson]

In case people haven't noticed, linux has not only caught up, but surpassed Windows, in terms of stability, modularity, customizability, ease of install, maintainability of the code base, etc.

That last one - maintainability of the code base, is a killer. There will be no Windows after Vista. Even Microsoft has alluded to as much.

BTW - That "etc" I mentioned includes REPUTATION. What is the reputation of linux vs windows? Ask any virus-writer.

No mention of vendor lock and switching costs

[140Mandak262Jamuna]

Looks like these two researchers are still using lessons learnt in the marketplace for actual physical objects and applying it to non-physical, intellectual products. The entire article introduces a term demand-side learning . But does not mention the words "vendor lock" or "switching costs".

If you are selling garden hoses, the cost of switching to a competing brand is just the replacement cost of a garden hose. If a company is switching software from one vendor to another, the switching cost is considerably more than just plain cost of new software. Like changing the garden hose requiring you change all the plumbing fitting and pressure valves in your home! The first mover advantage is directly proportional to the switching cost. Where are Lycos and Hotbot now? All vendors know that and they strive hard to increase the switching costs, from AutoCAD, Ansys, Fluent, Cadence, to Oracle, MSFT every dominant vendor in the market tries as hard as possible to make it inpossible to switch.

The reason why garden hoses, light bulbs and tires have low switching cost is because of standardization. Standards defined by independant third parties, not by the manufacturers themselves. People, consumers and corporations are beginning to understand the issue, as seen the recent moves by Massassuchetts to mandate ODF as the archival format for its documents. It is inevitable that people will see the advantages of interoperability and standardization. The first mover advantage will diminish as consumers level the playing field by demanding interoperability and standardization. At that time the "second mover" into these fields will be OSS with value added services.

Where it all boils down to

[LinuxDon]

You can make the entire discussion as complex as you want, but there is only one reason why Linux doesn't succeed on the desktop market: Most commercial application are written for Windows, among them are a lot of specialist applications like ERP systems and to name another example "analysis software which interoperates with an advanced metal detector to detect explosives in the ground". With these kind of applications you can't just switch to an alternative, because there are just too few that match your needs and often NONE of them support Linux. The only way for Linux to succeed in these kind of settings is to make Wine work flawlessly. While Linux suits my home needs and server needs -very well-, it's useless on the desktop at the company I work for.

I think a simple economic model gives answers

[Kjella]

If you measure things on a complexity of say 0-100, then there's only a limited range of that which is profitable. That is, there's no money in making notepad clones and there's no money in extremely complicated features noone is able to use. However while there is OSS software that's trying to make money, a lot of it does not. Even in the darkest post-OS/2 days when Windows was completely dominating Linux evolved in a market that was essentially dead. That kind of development can't be stopped.

That is why I think OSS software will slowly consume normal COTS software, because they will keep going after the commercial companies say "Well, we've now added every feature with a tolerable ROI". I'm not quite sure about the timescale, but I think the OSS software base is only in its infancy. Imagine 10, 25 or 50 years down the road, how many software packages have matured to a point where they're everything a user expects from a word processor/graphics editor/media player etc., feature-complete and bugfree.

Re:I think a simple economic model gives answers

[RAMMS EIN]

``and there's no money in extremely complicated features noone is able to use.''

On the contrary, that's where the big money is. As you point out correctly, commodity software is slowly being overtaken by OSS; I think this is inevitable in a level playing field: open source will be developed as long as people are interest, no matter how unprofitable it is, but proprietary software will have a difficult time luring users away from open-source software. And open source tends to drive the price towards 0.

The only way to make money, then, is in markets where there is no open source software. That's what Windows does, currently: there is no open source implementation of the win32 API that is good enough to really pose a threat to Windows. But other than that, the money is mostly in custom software. Small money for simple things, and big money for big things. You bet you can rake in a lot of money for developing and supporting a large, complex application that drives a whole business. And someone will be raking in that money, until that application is replaced with commodity software - which may never happen.

Re:From TFA

[twms2h]

I think GNU/Linux was there first...it just didn't have the marketing that Windows had.

If you think that was all it lacked, you have been living in a world very different from mine.

Re:From TFA

[alanQuatermain]

The point is not whether a Linux ditribution existed prior to Windows 95, but whether it existed prior to Microsoft's software being installed on some high percentage of the world's IBM-compatible computers. The latter has been the case pretty much from the word go. When my dad bought a PC, he bought DR-DOS 6 to go with it, but never got around to installing it because the hard drive actually had MS-DOS 5 installed on it at the factory (the hard drive factory, that is -- apparently not uncommon back in 1991-ish). Until relatively recently, it's been fairly rare to find an IBM-compatible computer that does NOT use a Microsoft operating system. Sure, a few people back in the day would run DR-DOS, a few less would run something akin to BSD, although I have an idea the first PC-compatible BSD was for the 386, so that's got to be at least six or seven years after MS started shipping PC operating systems.

Sure, Linux distributions were available before Windows 95, but the same argument made by TFA existed back in 1993 too -- Microsoft's software was entrenched, it was the de facto system on most IBM-compatible computers, and it's reasonable to assume that a high proportion of the people using those would stick to what they have/know, rather than switch to something entirely new. Therefore, despite the existence of Linux, most folks stayed with DOS & Windows 3 until Win95 came along, which was a funkier Windows which didn't need you to mess about with DOS to get it running.

-Q

Re:Harvard is Wrong

[honkycat]

They are just plain stupid.

Great insight, thanks for clearing that up for us.

The cost of switching is a lot higher than just the "Microsoft tax." Most companies are heavily invested in particular software packages (CAD, accounting, payroll, etc). These are very specialized packages that often must be guaranteed (and often certified) to meet specific regulatory requirements. Unless the companies behind those packages can be convinced to migrate, there really is no option to switch for the company. When you factor in the costs like these, it's a no-brainer to stay on the Microsoft wagon. Compared to the overall cost of doing business, the software costs are negligible.

Those trends are very misleading.

[NDPTAL85]

I don't think those trends say much of anything. Did it ever occur to you that perhaps there's more searches for Linux because people need more help using it and getting it to just run properly? How many searches for "Macintosh" is someone with a Mac going to do if their Mac is running perfectly from the start? Just how many of those "Linux" searches are desperate newbies to Linux who are in week 3 of trying to figure out how to get wifi or wpa encryption support on their laptops or to get their monitors to run in the right resolution or to get CUPS printing working? One linux user might initiate dozens of search queries for anything linux related. A Mac user on the other hand would be searching for non-computer related stuff cause their computer is already working as it should be.

Apple is selling Macs hand over fists. The Apple Stores are always jam packed, the company is sourcing a third provider for notebooks and can't currently meet the demand they have for MacBooks and MacBookPros. The overwhelming majority of people who use Macs don't even know what UNIX is or that Mac OS X has a Unix core running underneath. Apple just doubled its laptop marketshare to 12% of the US market. Add in Apple's installed base all over the world and its pretty clear to me at least that not only are there more Mac users than Linux users but its going to stay that way for quite some time. Even if you account for the fact that it is nearly impossible to track Linux installs because Linux can be downloaded for free you should also admit that a lot of those downloads are probably never installed. I myself run Kubuntu Linux on a spare laptop and before I settled on it I downloaded at least 8 other distros. Some people download more distros some download less but the point I'm making is that only a fraction of the Linux downloads that are happening end up being actual installs. And sometimes none of them end up being permanent installs as the OS is still too rough around the edges for many users and they either go back to Windows or OS X.

Re:Why does this kind of thing ilicit so much emot

[babbling]

It does matter which operating system other people use, because companies like Microsoft and Apple take steps to ensure that as a Linux user, I can't communicate properly with users of other operating systems. For example, I tried to apply for a job recently, and the government department that I was trying to apply to sent me a Microsoft Word .doc file. These files are in a secret, proprietary format that Microsoft won't tell people how to open. They want to ensure that only Microsoft Word will open such files.

Another example, there's a radio station that I like to listen to online, and because they only offer Windows Media streams, I had to break the law (due to software patents) to play them on my Linux computer. Breaking the law isn't something I enjoy doing, and it shouldn't be something I have to do in order to not be excluded just because I am not using Microsoft software.

The problem isn't that people aren't running Linux, it's that they're running software from companies who are trying to exclude me (a Linux user) from being able to communicate other people (Windows and Mac users).

The Header is wrong

[ps3udonym]

They didn't predict that linux will remain second best at ALL. In fact, given the current conditions Linux will WIN!

"Having obtained this basic result, we investigate the conditions that will warrant that Linux ends up forcing Windows out. We do this by modifying the model in two ways. First of all, we look at the effect of having buyers such as governments and some large corporations committed to deployment of Linux in their organizations. We call such buyers strategic. In addition to cost-related reasons, governments back Linux because having access to the source code allows them to verify that sensitive data is treated securely. Binary code makes it hard to figure out who has access to information flowing in a network. Companies such as IBM, in contrast, back Linux because they see in OSS one way to diminish Microsoft's dominance. We find that the presence of strategic buyers together with Linux's sufficiently strong demand-side learning results in Windows being driven out of the market. This may be one main reason why Microsoft has been providing chunks of Windows' source code to governments."

Currently we have large stratigic buyers who are buying into linux in a big way. China is no small fish. This sounds more like the situation as it stands now. What he IS saying is that Linux will NEVER displace Microsoft "On it's own merits" because the superiority of the OS doesn't defeat Micosofts current large install base. We DO need things like the Free Software Foundation, it is critically important for linux that governments see the advantages of the development model and buy in.

You can see from the actions of Microsoft that they have known this data for quite some time, or at least are reacting in much the ways that the study recommends. So this IS a fight and the winner is by no means certain. One thing that I got out of this was that while Linux can accually defeat Microsoft, MS can't acctually defeat Linux and force it to dissapear. Think about that for a while.

Re:From TFA

[RAMMS EIN]

``Slackware is free. Windows is not. Microsoft can afford to pay thousands of employees. Slackware cannot. It's a tradeoff, and in fact "users" (that would be brought by marketing) hurt OSS. Developers are the lifeblood of free software. Users who are the lifeblood of commercial software. They are fundamentally different models.''

It's an interesting view, and OSS and proprietary software definitely have different dynamics, but I don't agree that "\"users\" (that would be brought by marketing) hurt OSS". Many of the behaviors a user could exhibit are the same for OSS and proprietary software: bug the developers, complain to friends and acquaintances that the product sucks, helping other users, recommending the product, writing about it in columns, etc. In either case, users may or may not be required to pay for the software (which can help maintain it), but it's never given that they will: a lot of software is free for use, and if it's not, there are always ways to avoid payment.

The big difference is that users of OSS can and will help maintaining and improving the product.

So, if anything, more users is an even bigger boost to OSS than to proprietary software; at least that's how I see it.

Conclusions Flawed but Very Interesting.

[twitter]

The summary is not as much fun as the article, which declares Microsoft's future dependent on FUD, sabotage, intentional waste and dumping rather than code quality. The whole summary reasoning boils down to, "It will be like this tomorrow because it's like this today." Even M$ knows that's not true. What M$ and IBM did to DEC used to keep Bill Gates up at night, and still might despite all of his ill gotten wealth. The authors have much more interesting things to say and do not really conclude M$ will always be around. The authors, while they do overplay the importance of an undefined "network effect" don't make such a gross error.

The authors don't really understand free software development but they do understand what M$ must do to stay alive. They understand the M$ network effect, which is difficulty working with people who don't have the latest and greatest M$ crap, but completely miss the free software networking effect and much of free software's social benefit. The more free software does, the more it will be used and the more it will grow. It's a power function, not dependent on large organizations and we are still at the very start of the curve.

One of the key flaws I found in the author's reasoning was this:

However, with a monopoly, the efforts to develop new software and improve the platform are directed towards one system only and this may turn out to be better from a social welfare perspective.

That's seriously flawed for two reasons. First there is no such thing as a "Linux Monopoly". It's only freely publish standards that make it look like a coherent whole and it's only M$ intentional ignorance of those standards that keeps both systems from interacting freely. The second, they seriously underestimate the size of the free software community and it's growth potential. The free developer community is and will allways be larger than the non free community. The whole point of the non free monopoly is to charge people money to participate. Free participation will never cost more than time and effort. GCC comes with most GNU/Linux distributions and there is a fantastic library of source code for every purpose no further than a network request away. The cost of a full version of M$ Visual Studio is close to $800, after you have paid the OS tax, and you need to buy a new one for each programmer every year or so. How economists could miss such a basic part of their model as cost of raw materials is beyond me, but part of it is a flawed assumption that free software is dependent on government and business support:

This questions the social desirability of policies aimed at guaranteeing Linux's survival. ... This [corporate] support is important because there are tedious portions of the code that would rarely be developed spontaneously by members of the Linux-developer community.

Wile corporate and government participation are welcome, studies don't bear out the necessity of their involvement. Companies and governments are going to increasingly use free software because of the tremendous flexibility and cost savings. There are hosts of things you just can't do with non free softare and most programmers spend all of their time making things work. Most programmers would be just as happy or happier with free software as long as it does the job.

Recognition of all the evil things M$ must do, while common here, are welcome from economists and business types. Formal recognition of the SCO and other FUD attacks, dumping by "piracy", the Halloween Documents, even sabotage of free software by "encouraging forks" are nice to see in print from a "respectable" organization. Remarkably, nowhere is there a statement that M$ has or must improve the quality of their code. Their conclusion is telling:

We conjecture that there are multiple equilibria and that the use of FUD to mold perceptions about future value becomes crucial.

Essentially, M$'s future is depends on lies. That's not a very bright future. Admission to that fact is all it takes for them and all of their intentional waste to dissapear.

Economic modelling is very accurate!

[Russ Nelson]

Economic modelling is very accurate! It was used to successfully predict 12 of the last five recessions.
-russ