Weekly Edition
Return to the Press pageSponsored link
Serve your customers, not your servers, with VERIO Linux VPS. Full-access test-drive here.
| Weekly edition | Kernel | Security | Distributions | Search |
| Archives | Calendar | Subscribe | Write for LWN | LWN.net FAQ |
Windows will beat Linux threat, say academics (TechWorld)
(Log in to post comments)
more information
Posted Sep 11, 2006 18:20 UTC (Mon) by stevenj (subscriber, #421) [Link]
The article is a bit thin on the actual assumptions made by the authors. See also this interview with the authors.The academic paper was published in Management Science, vol. 52, issue 7 (July 2006). Apparently that entire issue was devoted to open-source software. The table of contents is:
- "The Promise of Research on Open Source Software," by Georg von Krogh and Eric von Hippel.
- "Understanding the Motivations, Participation, and Performance of Open Source Software Developers: A Longitudinal Study of the Apache Projects," by Jeffrey A. Roberts, Il-Horn Hann and Sandra A. Slaughter.
- "Motivation, Governance, and the Viability of Hybrid Forms in Open Source Software Development," by Sonali K. Shah.
- "Exploring the Structure of Complex Software Designs: An Empirical Study of Open Source and Proprietary Code," by Alan MacCormack, John Rusnak and Carliss Y. Baldwin.
- "Strategic Interaction and Knowledge Sharing in the KDE Developer Mailing List," by George Kuk.
- "Location, Location, Location: How Network Embeddedness Affects Project Success in Open Source Systems," by Rajdeep Grewal, Gary L. Lilien and Girish Mallapragada.
- "Two-Sided Competition of Proprietary vs. Open Source Technology Platforms and the Implications for the Software Industry," by Nicholas Economides and Evangelos Katsamakas.
- "Dynamic Mixed Duopoly: A Model Motivated by Linux vs. Windows," by Ramon Casadesus-Masanell and Pankaj Ghemawat.
- "Entry Strategies Under Competing Standards: Hybrid Business Models in the Open Source Software Industry," by Andrea Bonaccorsi, Silvia Giannangeli and Cristina Rossi.
- "Open Source Software User Communities: A Study of Participation in Linux User Groups," by Richard P. Bagozzi and Utpal M. Dholakia.
- "The Architecture of Participation: Does Code Architecture Mitigate Free Riding in the Open Source Development Model?" by Carliss Y. Baldwin and Kim B. Clark.
I don't think they are freely available (I can access them through my university), although it's possible that some of the authors may post PDFs on their home pages. Seems like it may be interesting reading.
Of course, trust the press to jump on the most pessimistic article. (Not to mention that it's always dangerous to draw too many conclusions from the very latest research papers, especially if you are not a specialist. It's still too early to tell whether the analysis in that paper will stand the test of time.)
conclusions aren't as absolute as Techworld makes them sound
Posted Sep 11, 2006 18:45 UTC (Mon) by stevenj (subscriber, #421) [Link]
Anyway, let's not be Slashdot and jump on the most inflammatory quote in the article. If you look at what the authors actually say, their conclusions are much less extreme and absolute than what the Techworld writer seems to imply.
In particular, if you read the interview I linked, the authors state that when you include strategic buyers (large entities choosing free software for reasons other than the immediate cost) or cost asymmetries (if you assume that GNU/Linux development is more efficient than Windows development), then Linux-based systems can indeed displace Windows in their model.
The authors, in the original article, also caution that theirs is a "stripped-down structure with which to work rather than a more or less faithful representation of the actual interactions between Linux and Windows [...] we think that the model’s sparseness brings into sharp focus certain effects that should, at a minimum, be kept in mind in the real-world context (as opposed to not being recognized at all)."
conclusions aren't as absolute as Techworld makes them sound
Posted Sep 11, 2006 20:12 UTC (Mon) by nix (subscriber, #2304) [Link]
Indeed. I thought the TechWorld article must be oversimplifying things, asunder their logic Linux would never have got going at all, i.e., a
contradiction.
more information
Posted Sep 11, 2006 19:21 UTC (Mon) by josh_stern (guest, #4868) [Link]
Thanks for the citation. I had a look at the paper. The essence of the idea behind the abstract is that if we assume that buyers only look at their own self-interest for short term use of an OS (they are "myopic" and not "strategic" in the language of the article) and we assume that the value of the OS is proportional to its market share, and we assume that the rate of improvement of any OS increases with its market share (though maybe open source has a different rate increase function) and we assume that users start out with some positive valuation for MS-Windows in excess of Linux, and we assume a few other things necessary to write dynamic equations, then...
Given all of the above, there is some dynamic but non-zero price Microsoft can always set for their OS so it retains market share. Change any of the above assumptions and all bets are off. The paper is basically a math exercise.
According to my understanding, the model has nothing to say about the world for someone who already prefers Linux regardless of price. To anyone who believes that MS-Win is still better but that Linux is currently increasing in customer esteem at a faster rate than MS-Win, the model says that Microsoft is not following optimal behavior; it say that what MSFT needs to do is lower prices enough so that their market share is at least holding steady and ratio of rate of improvement is at least 1.
more information
Posted Sep 11, 2006 20:33 UTC (Mon) by superstoned (subscriber, #33164) [Link]
let me add to that: linux has a bit of a 'weird' advantage, and i'm notsure this model can catch that: its Free. which, in this case, means it
can't go bankrupt. and as it's development will almost sure go on, no
matter what (well, it might be forbidden to develop free software), it's a
hard nut to crack.
more information
Posted Sep 11, 2006 22:22 UTC (Mon) by drag (subscriber, #31333) [Link]
That and from my very very limited understanding is that economic models that assume one thing or another about a average person's behavior in regards to having limited vision are pretty flawed.
I don't realy know a lot about it, but it seems to me that especially in business if your not a strategic thinker your not a successfull person. Your simply not going to be able to make a small business into a large one just by dumb luck and a limited view point. Now these people may not have the same view point as yourself, but that doesn't mean that they aren't thinkers.
If your big business you may exist on beurocratic stuff and pure momentum alone, but I don't think that will last.
It would be interesting to see a more complex model with maybe a it bit more 'game theory' mixed into, if that is a correct way of looking at it.
Of course the positive spin you could put on this article is that it does a good job of proving that long-term looking intellegent people should choose Linux because it offers more benifits and lower costs over time.
more information
Posted Sep 11, 2006 22:29 UTC (Mon) by josh_stern (guest, #4868) [Link]
Their model assumes Linus is free and doesn't say anything about costs to develop or bankruptcy. It also doesn't explain how Linux could have gotten started and gotten as far as it did - taken literally, it actually predicts that couldn't happen, so this is already a demonstration that the model isn't quite right. Another aspect that the model doesn't deal with is what type of ROI MSFT stockholders expect - i.e. even if they could theoretically drop Windows price as low as required for steady state, they might not be able to do that in reality.
more information
Posted Sep 12, 2006 5:00 UTC (Tue) by gdt (subscriber, #6284) [Link]
there is ... price Microsoft can always set for their OS so it retains market share.
And this strikes me as the flaw in the argument. There is actually a floor to the price Microsoft can charge, and that is the price which spooks Microsoft's shareholders. The model doesn't include this factor at all, but since Microsoft has an unusual relationship with its shareholders it is important to model this relationship rather than to assume the shareholders will be happy with a profit-maximising return.
more information
Posted Sep 12, 2006 16:27 UTC (Tue) by josh_stern (guest, #4868) [Link]
Agree. In another post I remarked on that problem and the problem that the model, if true, would have predicted Linux never rise to its current market share. One further issue is that part of the models assumption is that community based software improvement can play a significant role on MS-Windows as well as Linux and, while I don't question this premise per se, I question whether relying on a lot of community-based software for the MS-Windows platform wouldn't have the effect of further reducing the perception that MS-Win is a more valuable OS - another state assumption the model depends on. Let's look at an example of this in reality: on the one hand, having Firefox/Thunderbird and OpenOffice for MS-Win adds rational value to the platform, but on the other, users knowledge that same working software is available for them on Linux diminishes their perception of a value gap.
I think they don't *quite* get it
Posted Sep 12, 2006 7:04 UTC (Tue) by xoddam (subscriber, #2322) [Link]
From the author interview:
Consider SCO, a small Swiss-based "vulture" firm...
That sympathetic but fundamentally incorrect factoid, and the business of "demand learning" -- something the authors think the "Open Source" development model can somehow do faster than a large proprietary software company that has intimate relationships with its large customers and buckets of money to spend on usability testing -- lead me to think these guys, while not entirely *wrong*, don't quite "get it".
The single advantage they posit for Open Source is in no way inherent to free software -- and when comparing to a responsive and well-heeled proprietary business, doesn't even actually exist. They *do* surmise about "strategic" reasons for users to make a platform switch, but they don't seem to consider what the reasons actually are from the user's point of view (as opposed to Stallman and Torvalds, whose motivations they do briefly ponder). Short-term prices simply don't enter into these decisions -- while long-term data dependency on an untrustworthy vendor *does*.
I also have serious misgivings about what their definition of social welfare might be.
Finally, the social welfare result that a Windows monopoly is not always worse than a Linux-Windows monopoly was also unexpected. This questions the social desirability of policies aimed at guaranteeing Linux's survival.
Ah, shills. The rest of the interview doesn't suggest it, but that's a dead giveaway. Nothing more to see here...
Read more carefully
Posted Sep 12, 2006 14:40 UTC (Tue) by dthurston (subscriber, #4603) [Link]
They *do* surmise about "strategic" reasons for users to make a platform switch, but they don't seem to consider what the reasons actually are from the user's point of view...In the interview, they do consider the user's motivations for switching; they mention governments' desire to see the source code to verify security and IBM's desire to lower Microsoft's share. From the point of view of the model, the exact reasons some users might prefer Linux is probably irrelevant. They also explicitly say that a Linux monopoly is always preferable to a Windows monopoly. I don't think your "shills" comment is fair.
Maybe I should *write* more carefully?
Posted Sep 13, 2006 2:13 UTC (Wed) by xoddam (subscriber, #2322) [Link]
> From the point of view of the model, the exact reasons some users> might prefer Linux is probably irrelevant.
This is exactly what I meant by not considering the reasons. They've
made their model incapable of predicting anyone's platform change without
the network effects of an already growing market share. It has to be
told that change is happening despite its basic assumptions.
Not that it wasn't an interesting intellectual exercise, but trying to
model an actual phenomenon without modelling the factors which bring that
phenomenon about seems pretty absurd to me.
> I don't think your "shills" comment is fair.
That was just a bit of flamebait, but whatever the variable is that their
model labels "social welfare" (I suppose it's some estimate of the
general utility of the software people happen to be using at the time), I
certainly wouldn't call by that name.
As far as I'm aware the biggest reasons for the continued dominance of
the proprietary share of the market are abuse of the Microsoft monopoly
to ensure that no significant vendor retails PCs with GNU/Linux
pre-installed, and patents keeping free software from implementing
certain popular multimedia applications. If either of these factors are
modelled, Casadesus-Masanell and Ghemawat only mention it in passing:
> A few actions that the model suggests Microsoft could do to
> remain competitive are:
> ...
> e. Decrease Linux's demand-side learning. Because the way
> to do this involves some questionable (from a legal point
> of view) actions, we will refrain from suggesting specifics.
Hmmm. Both market factors above can be seen to reduce free software's
"demand-side learning" (which is *NOT* an advantage Linux has over a keen
competitor which does its market research, as Microsoft clearly does):
by abusing the monopoly to keep that 'demand side' from encompassing the
general public, and by abusing patents to keep free software from
satisfying some of the demand.
OK, the patent side of it is not 'questionably legal' as long as software
patents are legal ... but what else could they possibly be referring to?
Read more carefully
Posted Sep 13, 2006 4:53 UTC (Wed) by se8ohThi (guest, #40470) [Link]
> They also explicitly say that a Linux monopoly is always preferable to a> Windows monopoly.
... which actually doesn't make much sense. A "linux monopoly" as such cannot
exist by definition. A monopoly is when something is and can be sold by
a single entity only - which is rather unlikely to happen with linux.
A linux _monopoly_ probably wouldn't be any better - basically the whole
point why a linux _monoculture_ probably would be better is the fact that
it can't be turned into a monopoly easily (if at all) ...
I think they don't *quite* get it
Posted Sep 13, 2006 8:40 UTC (Wed) by Seegras (guest, #20463) [Link]
Consider SCO, a small Swiss-based "vulture" firm...Swiss? Not at all. Utah is definitly not in Switzerland.
I think they don't *quite* get it
Posted Sep 15, 2006 17:38 UTC (Fri) by Alan_Hicks (subscriber, #20469) [Link]
The SCO Group or its parent company the Canopy Group may be incorporated in Switzerland as a tax shelter?
Changing expectations
Posted Sep 18, 2006 21:04 UTC (Mon) by njs (subscriber, #40338) [Link]
> Of course, trust the press to jump on the most pessimistic article.
It's interesting that "Windows might beat Linux" has become a
surprising and newsworthy story.
A point that I believe was missed...
Posted Sep 11, 2006 21:56 UTC (Mon) by filker0 (subscriber, #31278) [Link]
As MS moves on to Vista and Linux (and other Open Source solutions) improves, a vast number of existing systems will be orphaned by Microsoft, yet Linux will continue to run on those systems with a higher efficiency than MS-Windows did. This will start with a few people who are currently in the MS camp deciding to try Linux on their otherwise retired systems, discovering that it runs faster and better than Vista on their new hardware, and some of those will move over into the Linux camp. This will be somewhat corrosive to the Microsoft mindshare.
This, of course, assumes that the progress on the ease of use (as perceived by the non-Linux/Unix savy) continues and that desktop solutions that don't ever require the user to directly log in as root are developed. Looking at how MacOS X handles this, I'm pretty sure it can be done on the Linux side.
One way we in the Open Source community can do is to encourage commercial software developers to port their applications to Linux, and that will involve getting a sufficiently common installation scheme across distributions that a single installer can install on Debian, Fedora, RHEL, SuSE, Ubantu, etc. We also have to convince them that hosting software on a GPL system automatically forces them to make their code GPL. Even if there are free software analogs, people (in general) don't want to have to learn new interfaces all at once, and if the commercial products are available, it will lower the barrier that is inhibiting a larger migration to the Linux desktop.
All of this would alter the numbers from the original article.
No convincing required.
Posted Sep 12, 2006 3:58 UTC (Tue) by xoddam (subscriber, #2322) [Link]
We also have to convince them that hosting software on a GPL system automatically forces them to make their code GPL.
Please! Even assuming that you meant to write "doesn't force", no-one who has been paying attention thinks so. No further convincing needs to be done; the factors which keep much commercial software Windows-only are market forecasts (there is no sign that the Windows market will cease to grow, even if its share of the total installed desktop base is gradually shrinking) and the perceived costs of porting.
At some point (though not just yet) a further factor will be that the free software desktop has incumbent free-beer competitors for most of these Windows-only apps. But this can only be considered a temporary effect, as if there is ever any significant demand for a free software application on Windows, it will be ported sooner or later.
That the platform has a particular license is *not* an issue. Anyone who thinks that free software still 'needs' to convince proprietary vendors to jump ship from the Microsoft platform is still living in 1998.
No convincing required, except where it still is...
Posted Sep 12, 2006 14:04 UTC (Tue) by filker0 (subscriber, #31278) [Link]
Yes, I mean't "doesn't force", and the reason that I wrote (or at least meant to write) what I did wasthat I've recently attended a presentation given by a .NET evangelist (not a Microsoft employee) who
was implying that the risks of offering your product on Linux was that you'd be forced to open-
source your intellectual property, thus destroying your competitive advantage.
Several people at the seminar took him to task on this, but he wouldn't back down, and some of the
"suits" at the presentation remained convinced of the correctness of his assertions. The
misconception is still out there, being promulgated by people who have a vested interest in keeping
the MS monopoly in place.
I believe that, unless there are radical changes in the way MS does things, their own software will be
their downfall, not pressure from Linux or other open source alternatives. I was just pointing out
that I feel that the article doesn't take into account the mindshare shift as the weight of MS Vista
and upcoming versions of MS applications obsoletes systems that will continue to run Linux very
well for many years into the future. I got off onto a tangent about commercial applications...
No convincing required, except where it still is...
Posted Sep 12, 2006 16:29 UTC (Tue) by zotz (guest, #26117) [Link]
"Several people at the seminar took him to task on this, but he wouldn't back down, and some of the "suits" at the presentation remained convinced of the correctness of his assertions. The misconception is still out there, being promulgated by people who have a vested interest in keeping the MS monopoly in place."
I would ask someone who maintained such a position where I could get my hands on all of Oracle's* code since they have ported their stuff to linux and thus, according to the view that running code on GPL OSes forces you to GPL your code, then the Oracle code should be GPL licensed by now. Surely?
This should clue in anyone with half a brain that is interested in the truth. What can you do with those who aren't interested?
* Substitute as needed.
all the best,
drew
(da idea man)
http://www.ourmedia.org/user/17145
All great news, really
Posted Sep 11, 2006 23:46 UTC (Mon) by bojan (subscriber, #14302) [Link]
Around year 2000, every time I suggested something should run on Linux, I'd be dismissed straight away by "serious business" folks. These days, I get hired to port strategic, core business applications to Linux or move complete production environments from other OSes to Linux. A lot has changed already.
Duopoly would be an awesome result, IMHO. If I could walk into any computer store and have an "official" choice between Windows and Linux machines, I'd be very happy. But here is the real kicker (quote from the interview):
> 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.
That's why Microsoft has been so active in trying to convince people that adopting an "open source first" policy is a no-no. We can see that in the tech press daily. My bet is that Microsoft either sensed this already or did such research themselves.
A link to the orginal article:
Posted Sep 12, 2006 5:58 UTC (Tue) by zorgan (subscriber, #4016) [Link]
PDF of the research article.I have only skimmed through it, but it looks interesting (and far more balanced than the headline of the techworld article is suggesting).
As an interesting side note, they measured a distinctively positive correlation between Linux adoption rates and innovation/technology indices (among 51 countries), and a highly negative correlation between Linux adoption and piracy (as predicted by their model).
A link to the orginal article:
Posted Sep 12, 2006 7:13 UTC (Tue) by TimM (guest, #5436) [Link]
... and a highly negative correlation between Linux adoption and piracy (as predicted by their model).Surely this is the real story. Increased Linux use prevents piracy ;-)
Please, nobody reply pointing out that correlation does not imply causation. I'm aware of the difference. Pity the general public is not.
A link to the orginal article:
Posted Sep 12, 2006 13:59 UTC (Tue) by gravious (subscriber, #7662) [Link]
Hmm,
It has been plainly obvious to me, since the late nineties (at least) when I first started thinking about the intersection between software and ethics, that software piracy means a different thing to a Microsoft than it does to a Borland - say. A software company with a near monopoly in the market will preserve that market share in the face of piracy and will not lose money even though each pirated copy is theft. For a company like Borland though it matters a lot if you pirate their software because the lost revenue will not be recovered through any network effect because as being in second place to begin with they had to adhere to some standard whether it was Ansi C or Intel x86 assembly code instructions or whatever. Witness Netscape - a striking example, its browswer was even given away free! The comapny basically said, 'pirate this', and they still got their ass kicked (regardless of what Joel on Software says about re-engineering).
It appears to me that FOSS is the only way to to prevent strangulation. Software is too easy to copy and copy perfectly at that. A Microsoft has a fundamentally different kind of monopoly to say an AT&T or a Standard Oil or a De Beers (http://www.theatlantic.com/doc/198202/diamond) or a Government - well okay, governments are a different kind of monopoly again and the only ones that seem to be able to deal with really aggressive monopolists always assuming mr. goverment is independent and fair which he is not. So Microsoft has a fundamentally different type of monopoly.
Another interesting factoid is that every single _large_ software company has it's own OS (or more). Apple - Mac OS(n). IBM - PC DOS,OS/2,AIX. SGI - Irix. Microsoft - DOS,Windows (3x,9x,NT,CE),OS/2,Xbox...Novell - Netware, Novell DOS, SUSE. Sun - SUNOS,Solaris. HP - HP/UX. DEC - Digital Unix...SCO, AT&T, NEC, Atari, Commodore, the list goes on and on. The only exceptions to this rule seem to be Adobe, Autodesk, Symantec, Oracle and a number of others. But as you see, to be a truly large software you must have an OS you can call your own... But once Microsoft commands the OS space, this commanding threatens every single _other_ large software related company (granted Apple, Sun, IBM, SGI, NEC, HP all do hardware as well) in fact you can see that the companies that don't do hardware also don't have an OS except Microsoft until recently. Microsoft monopoly in the OS space kills the software arms of the large hardware/software behemoths and stifles innovation in the purely software space unless you are very specialised (digital/media, autocad, cam, big iron databases, security - more or less any space that is not general and that Microsoft has not gotten around to taking an interest in.
This leads me to believe that FOSS is the only non-governmental way (that is, legal measures) to move on from the dire situation we as innovators find ourselves in - though FOSS requires the legal system to work to its advantage. I predict, regardless of what this research paper finds, that Microsoft will get more and more into hardware and services, UNIX will get thrashed more and more by Linux* and there will be other GPL OSes that we are not aware of that will compete with both Microsoft and Linux but will not be UNIX based - RockBox springs to mind. Microsoft's dominance will come to be seen to be an aberration. There are too many curious and inventive and altruistic people in the world. When a research paper takes into account people's moral side when making decisions then it will serve as a viable model for forecasting.
Any thoughts on this?
Anthony
* when I say Linux - substitute your own favourite FOSS OS here :)
A link to the orginal article:
Posted Sep 12, 2006 16:36 UTC (Tue) by nix (subscriber, #2304) [Link]
Your argument that every large software company must have its own OS is more than slightly dented by the 'except Oracle'.
(Of course, in some respects the Oracle RDBMS *is* an OS; at least it contains most of the components of an OS kernel inside itself, for all that it doesn't export them to the outside world. But still.)
A link to the orginal article:
Posted Sep 12, 2006 17:55 UTC (Tue) by oak (subscriber, #2786) [Link]
Doesn't Oracle have it's own version of (RedHat?) Linux with kerneltuned to best performance for Oracle?
As to the previous comment, didn't all those companies have also
their own HW (except Novell?)? In that case having your own OS
was kind of mandatory as back then there was no free OS (like BSD
or Linux) one could adopt and in which community participate.
A link to the orginal article:
Posted Sep 13, 2006 19:07 UTC (Wed) by AJWM (subscriber, #15888) [Link]
Although Oracle won't run on bare hardware, it pretty much treats the OS as a souped up BIOS and does its own filesystems, process monitoring, shared memory management, etc. layered on top of the native OS. Then there are all the Oracle applications that live on top of the RDBMS.
Almost the same with Autodesk -- they have a ton of apps that live atop AutoCAD, along with their proprietary file formats and own dialects of programming languages (Autolisp, eg).
Perhaps "platform" would be a better word than "OS" in the grandparent's context.
A link to the orginal article:
Posted Sep 14, 2006 11:30 UTC (Thu) by nix (subscriber, #2304) [Link]
Well, yes, all flexible-enough systems turn into platforms eventually. (Emacs, Mozilla...)