Microsoft at the power point [why Linux is slandered at FR]

The Economist ^ | 13 September 2003 | Economist staff

[chilepepper]

Governments like open-source software, but Microsoft does not

IN MAY, the city of Munich decided to oust Microsoft Windows from the 14,000 computers used by local-government employees in favour of Linux, an open-source operating system. Although the contract was worth a modest $35m, Microsoft's chief executive, Steve Ballmer, interrupted his holiday in Switzerland to visit Munich and lobby the mayor. Microsoft even dropped its prices to match Linux—a remarkable feat since Linux is essentially free and users merely purchase support services alongside it. But the software giant still lost. City officials said the decision was a matter of principle: the municipality wanted to control its technological destiny. It did not wish to place the functioning of government in the hands of a commercial vendor with proprietary standards which is accountable to shareholders rather than to citizens.

Worryingly for Microsoft, Munich is not alone in holding that view. Across the globe, governments are turning to open-source software which, unlike proprietary software, allows users to inspect, modify and freely redistribute its underlying programming instructions. Scores of national and state governments have drafted legislation calling for open-source software to be given preferential treatment in procurement. Brazil, for instance, is preparing to recommend that all its government agencies and state enterprises buy open source.

Other countries are funding open-source software initiatives outright. China has been working on a local version of Linux for years, on the grounds of national self-sufficiency, security and to avoid being too dependent on a single foreign supplier. Politicians in India have called on its vast army of programmers to develop open-source products for the same reasons. This month, Japan said it would collaborate with China and South Korea to develop open-source alternatives to Microsoft's software. Japan has already allocated ¥1 billion ($9m) to the project.

Why all the fuss? Modern governments generate a vast number of digital files. From birth certificates and tax returns to criminal DNA records, the documents must be retrievable in perpetuity. So governments are reluctant to store official records in the proprietary formats of commercial-software vendors. This concern will only increase as e-government services, such as filing a tax return or applying for a driving licence online, gain momentum. In Microsoft's case, security flaws in its software, such as those exploited by the recent Blaster and SoBig viruses, are also a cause of increasing concern.

Government purchases of software totalled almost $17 billion globally in 2002, and the figure is expected to grow by about 9% a year for the next five years, according to IDC, a market-research firm (see chart). Microsoft controls a relatively small part of this market, with sales to governments estimated at around $2.8 billion. But it is a crucial market, because when a government opts for a particular technology, the citizens and businesses that deal with it often have to fall into line. (In one notable example, America's defence department adopted the internet protocol as its networking standard, forcing contractors to use it, which in turn created a large market for internet-compliant products.) No wonder Microsoft feels threatened—the marriage of open-source software and government could be its Achilles heel.

Policymakers like open source for many reasons. In theory, the software's transparency increases security because “backdoors” used by hackers can be exposed and programmers can root out bugs from the code. The software can also be tailored to the user's specific needs, and upgrades happen at a pace chosen by the user, not the vendor. The open-source model of openness and collaboration has produced some excellent software that is every bit the equal of commercial, closed-source products. And, of course, there is no risk of being locked in to a single vendor.

That said, open-source is no panacea, and there are many areas where proprietary products are still far superior. Oracle, the world's second-largest software company, need not worry (yet) about governments switching to open-source alternatives to its database software. But Microsoft is vulnerable, because an open-source rival to its Windows operating system exists already, in the form of Linux.

If Microsoft is indeed squeezed out of the government sector by open-source software, three groups stand to benefit: large consultancy firms and systems integrators, such as IBM, which will be called in to devise and install alternative products; firms such as Red Hat or SuSE, which sell Linux-based products and services; and numerous small, local technology firms that can tailor open-source products for governmental users.

As a result, the company has been fighting back. Microsoft and its allies have sought to discredit open-source software, likening its challenge of proprietary ownership to communism and suggesting that its openness makes it insecure and therefore vulnerable to terrorism. The firm also created a controversial slush fund to allow it to offer deep discounts to ensure that it did not lose government sales to Linux on the basis of price. And Microsoft has paid for a series of studies, the latest of which appeared this week, which invariably find that, in specific applications, Windows costs less than Linux.

More strikingly, Microsoft has been imitating the ways of the open-source “community”. Last year, the firm launched a “shared source” initiative that allows certain approved governments and large corporate clients to gain access to most of the Windows software code, though not to modify it. This is intended, in part, to assuage the fears of foreign governments that Windows might contain secret security backdoors. Microsoft has also made available some portions of the source code of Windows CE, which runs on handheld PCs and mobile phones, to enable programmers to tinker with the code. Tellingly, this is a market where the company is a straggler rather than a leader.

Jason Matusow, Microsoft's shared-source manager, says that developing software requires leadership and an understanding of customer needs—both areas where proprietary-software companies excel. As for proposed legislation that would stipulate one type of software over another, it is anti-competitive and could leave users hamstrung with products that are not the best for their specific needs, says Robert Kramer, executive director of the Initiative for Software Choice, a Microsoft-supported lobby group. Microsoft will advance these views next week in Rome, where it is hosting the latest in a series of conferences for government leaders. But the signs are that many of them have already made up their minds.

[Dominic Harr]

Now it is interesting to consider IBM releasing OS/2.

IBM makes their money off of hardware sales, for them, the OS is just one component.

I can see that benefitting them.

That is an interesting idea. One I almost wonder if IBM wouldn't -- or hasn't -- considered.

[Nick Danger]

I don't see MicroSoft competing head to head with Linux by using MS's top of the line OS's (at discounted prices) as being a winning strategy. Much better to co-opt the Open Source movement and cut Linux down in one bold lightening strike move with Windows 95, imho.

There is an old management book, now out of print, called Up the Organization. It's by the guy who used to run Avis.

One of the stories he tells in the book concerns the threat posed to Avis by the discount rent-a-car guys like Budget. Somebody thought they should start their own down-market rent-a-car outfit... have two brands... compete with Budget on Budget's turf. The CEO asked the finance guy what he thought of that idea. "I call that pissing in the soup," he replied.

Microsoft has $50 billion in the bank. It will be difficult to persuade the people running it that they don't know how to play this game.

Linux represents a challenge. All companies face challenges. IBM faced one in the form of the VLSI microprocessor. It was going to make most big boxes go away, in favor of things the size of a breadbox that sold for under 5 grand. That was bad news for a company that made its living selling big boxes. A 370/168 sold for $7 million back then. That whole business was going to go away. But it was reality, and they faced it.

At the time the Apple ][ started appearing on people's desks in offices, the big computer companies were IBM, Burroughs, Sperry-Univac, NCR, Honeywell, and Control Data. All but IBM are dead. The big mini companies were DEC, Data General, Prime, and Wang. They're all dead, too.

IBM isn't dead. They faced their technological challenge. Those who faced it with them didn't make it. It didn't hurt that IBM had $40 billion in cash when it started, which was probably worth more than Microsoft's $50 billion is today.

But Microsoft has a fighting chance... at least as good as the one IBM had. Their flagship product is going to get mowed down by a New Thing that costs a whole lot less. All they have to do is weave their way through a minefield, and come out the other side. It can be done. IBM did it.

[Nick Danger]

That is an interesting idea. One I almost wonder if IBM wouldn't -- or hasn't -- considered.

They have considered it. OS/2 had become a cult, the way Amigas were a cult. When IBM chose to drop it, the "cult" made a proposal to take it over, to keep maintaining it. IBM said no, but my recollection is that one of the reasons is that Microsoft still has some hooks in it... OS/2 started out as a joint development effort. I think there might be some big banks that still use it. I've heard that, but I don't know for sure.

[general_re]

All they have to do is weave their way through a minefield, and come out the other side. It can be done. IBM did it.

And they did it despite the fact that DOJ occupied Armonk for a good long time. Or maybe it was because of that ;)

It's a good parallel in a lot of ways - people forget nowadays, but at one time IBM was just as hated and feared as MS is today. There was a whole generation of CS folks that came out in the 60's and 70's, for whom "anybody but IBM" was more than just a suggestion, it was a way of life - and from them, you got the rise of UNIX and companies like Sun and Apollo and Silicon Graphics. And Microsoft, of course.

And now history repeats itself - "anybody but Microsoft" is the mantra these days. If Microsoft has any advantage, it's that they saw what happened to IBM, and have a path to follow, and they don't have Justice all over them the way IBM did. Watching the downfall of IBM, but not its subsequent rise, is, I think, what gives Microsoft its greatest blessing and its biggest curse - their rampant paranoia about the competition. A blessing in the sense that it keeps them constantly moving forward; a curse in the sense that it makes them heavy-handed when they don't really need to be, which is partly why they're hated so much. And they're hated just for being number one, same as IBM was...

[Coral Snake]

I don't think that IBM can release the OS/2 code because there is substantial M$ and Adobe as well as other third party code in it. (the LAST thing OSS needs is another SCO mess going on with M$ as the plaintif should the "wrong codes" be opened.) What they could do however now that I thought about this some more is use the code from the OS/2 CLI and Workplace Shell that they DO own to give LINUX a more OS/2ish appearance for Businesses that they are trying to move from OS/2 to Linux.

[tortoise]

Free Republic is hosted on Verio? *shudder*.

Yeah, I've thought about offering to move them to our fiber more than once, particularly since we have extensive fiber networks in the area where the servers are now. Cheaper, faster, and not Verio. (And by "our fiber", I literally mean that; we own gigabit fiber networks that are among the fastest segments of the optical switch fabric that is the US Internet.)

If they ever get sick of Verio, I could hook them up. I wonder how much bandwidth they are actually getting for their $2,700/mo. That buys a pretty fat pipe (no oversubscription and burstable to very high throughput in fact) straight through the core these days.

[tortoise]

Clustering has been available since NT 4.

In all fairness, Windows clustering remains an utterly useless piece of crap to this day. It was a checklist feature toy that Microsoft sort-of put together and never really got working correctly. Massive supercomputing clusters are a very mature feature in Linux, but it never worked right on Windows and MS did actually hire some of the guys who designed the Linux clustering software to try and get Windows clustering running correctly. I'm sure they could have gotten it working on Windows eventually, but Microsoft was mostly paying lip service to the concept and it looks like they've dropped it.

[Bush2000]

Except that MicroSoft already has such a competitor.

And introducing its own obsolete OS as another open source competitor would further dilute its market share.

[general_re]

It started out as simply a means of basic failover support and load-balancing, and that's still mostly what it is - high-performance computing is still down the road a ways. But they've hardly dropped it - it still exists, it's still being developed, and it's still being supported. The clustering support in Windows Server 2003 is a marked improvement over previous versions, as a matter of fact.

[Bush2000]

In all fairness, Windows clustering remains an utterly useless piece of crap to this day.

Thanks for the FUD -- but you couldn't be more wrong (nothing ever changes).

Windows Clustering Delivers

[chilepepper]

I used to have a very low opinion of IBM but over the years have become a big fan of theirs (probably started when I bought one of their laptops -- and am typing this on my updgraded one, a T21). To me IBM epitomizes the giant corporation that has re-tooled itself for the new realities of our economy.

Now IMHO they were able to do this because IBM corporate culture allowed them to. It seems to me, as regimented as IBM is (or was), their heart was in the right place and they never took their eye off the future. While they are clearly not the *pre-eminent* high tech company these days, they are not very far off being so. I find the IBM approach to things to be very professional, measured and honest.

I cannot say Microsoft engenders the same sentiments. MS is more of a pyramid scheme than a company, and their "technological breakthroughs" are laughable. Microsoft is *very fragile* IMHO, and money by itself won't be enough to retool it unless the MS culture allows it. MS *WAS* able to adapt to the network and the WWW, but my impression is that from the moment the internet forced its way onto the MS radar their intention has been to layer it with their products in such a way as a way to strengthen their monopoly. Make no mistake -- MS was in many ways blindsided by the WWW -- if they had a really good R&D group they would have know what was coming, and LED the technology, not show up later and MONOPOLIZE IT. IMHO MS lacks both in "class" and honesty, and it would be disasterous to underestimate the importance of these things in realigning a corporation in a vastly different economy.

IBM all these years was more interested in the technology of what they do -- becoming an empire was a result of *THAT*, because they started from the premise

"aha, here is a new technology -- get the designers and scientists on it and if we do things this way we will have the best hardware and will become preeminent"

The Microsoft approach, IMHO, is more like this:

"aha, here is a new technology -- get the lawyers and MBA's on it and if we do things this way we will sell this many more systems a year and will guarantee this much cashflow this many more years in the future from upgrades and if we do things this way we will lock out competitor X and competitor Y (Lotus, WordPerfect, Netscape, NetWare, ...)

[Snuffington]

It's not Linux that is the problem: it is the idea itself. ... Why is it that they invariably come from the States? This is because private ownership stimulates innovation.

I agree with you in some ways, disagree in others, and would like to head in a different direction to explain why.

You're absolutely right. The idea is the problem. Not a specific idea. The entire nature of ideas.

Through patents and copyrights, our current economic model assigns ownership to ideas. In effect, it treats ideas like physical property - a "fixed pie" view of idea holding. The problem with this is that it's a fiction. Ideas don't work that way.

In defense of those who developed the law as such, they weren't idiots. They didn't misunderstand the nature of ideas. They were trying to come up with a practical compromise - a system by which defense of intellectual "property" could produce some of the same benefits as defense of physical property. It is also important to note that, whatever the problems or abuses, this has been a largely successful effort.

The important point here is that it was a matter of practicality and societal benefit that crafted this sort of law. There is no grand philisophical ground to stand upon in claiming ownership over an idea. On its face, it's an absurdity. In order to be useful, ideas need to get out. When they get out, with or without the intent of those they get out to, other people now share them.

Further complicating this is the notion that more than one person may arrive at the same idea independently. This is handled somewhat strangely in the copyright and patent world (e.g. the use of clean rooms and "virgins" in software and hardware development, or the inability to copyright a book title, etc.).

Why has this all become such a crisis at the beginning point of the 21st century? Not because ideas have changed. Not even really that laws have changed. The problem is that technology allowing the sharing of even extremely complicated ideas is becoming universal. In digital format, I can send anything from the detailed blueprints of a brand new microprocessor, to the new fall fashion lineup from the GAP to almost anyone in the world. I don't even have to be sneaky or malicious in intent. The technology was designed to make that sort of idea sharing ubiquitous. Tracking this back to its roots - from the printed word, to the printing press, to the telegraph, telephone, and internet, one can see that, far from stifling idea generation and growth, new ideas are exploding as quickly as the capacity for transmitting them allows. Idea creation follows the technical capacity for sharing ideas far more closely than the laws crafted to claim ownership over them.

Does this mean we should abolish copyrights and patents? Not at all. It means we need to examine the practicality of such laws in light of the technical realities of today.

Some of those now defending our current laws in defiance of the technical reality are openly suggesting that we cripple our technology to limit its capacity to share intellectual property (e.g. Orrin Hatch). Others suggest we pack the courts with even minor cases of intellectual property piracy (e.g. the RIAA). I would say these paths are far more likely to harm growth and innovation than the alternative. A better path would be to find new compromises between the reality of ideas, and the benefits of property protection. Protect intellectual property where it is practical and promotes social good. But don't pretend intellectual property can be treated exactly like physical property all the time.

[Southack]

"And introducing its own obsolete OS as another open source competitor would further dilute its market share."

Then perhaps Windows 95 is not as **obsolete** as it is made out to be, if its old dead code is that much of a marketshare threat to frontline MS OS's almost a decade later.

I mean, if I had the choice of either keeping a customer on a MicroSoft platform (where I could at least sell them major money-makers such as MS Office) or losing said customer to Linux, then giving away Windows 95 as Open Source would be a far easier choice to me than would be the option of discounting to near $0.00 a newer OS like Windows XP.

Are you actually an MS employee? I ask only because it would actually mean something to my own business model if MicroSoft itself is content to lose platforms to Linux rather than find ways to maintain their current desktop lead.

[tortoise]

Thanks for the FUD -- but you couldn't be more wrong (nothing ever changes).

That's awfully disingenuous and your link actually proves my point. I'm talking about high-availability computational cluster containing dozens or hundreds of servers, and the best you can do is come up with is a distributed failover solution that maxes out at a few servers. Geez, talk about apples and friggin' oranges.

Get your head out of your ass and try again: Give me some examples of massive distributed application clusters with dozens of nodes that is more than just a glorified failover solution running on Windows. I'm not even a particularly big fan of Linux, but Windows application clustering is effectively non-existent and nobody even pretends otherwise. On the other hand, Linux application clustering is pretty advanced and comparable to the Big Iron clustering solutions in many ways.

Windows may get REAL clustering some day, but it doesn't have it right now. It is worth pointing out that this requires some substantial support in the OS kernel to work well that Windows doesn't have currently, which is also one of the reasons clustering matured so rapidly in Linux over the years. You can hack a communications framework on top of any OS for a limited number of applications, but more advanced capabilities (cf Mosix) pretty much require kernel level support.

[tortoise]

The clustering support in Windows Server 2003 is a marked improvement over previous versions, as a matter of fact.

Well that's good to know. Truly competent clustering requires an operating system with a ton of hooks expressly for that type of usage. You can fake a little of it on top of the operating system, but "real" application clustering (i.e. not just basic failover or simple loosely coupled SIMD) requires non-trivial kernel-level support.

I am acquainted with some of the Linux clustering guys who were hired by Microsoft to help them with their problems. According to them, the problem with clustering on Windows is that the internals of the OS are very poorly suited for clustering compared to just about any flavor of Unix, so the problem is reduced to doing gross hacks without tweaking the OS internals too much, or doing wholesale rewrites of certain OS internals which could break a lot of things. Microsoft chose the former route for compatibility's sake, and has a very slow roadmap to eventually get the OS in the state it needs to be so that it doesn't disrupt things too much. In the mean time, they are doing marketing-checklist-only "clustering" so that they can claim namespace until their OS is actually capable of real competent clustering.

[Bush2000]

I don't have time to educate you about Windows Clustering. It's real. It exists. And it works.

Clustering

[Nick Danger]

giving away Windows 95 as Open Source would be a far easier choice to me than would be the option of discounting to near $0.00 a newer OS like Windows XP.

It's not obvious why this would be true. The marginal cost of making the next copy of Windows XP is no higher than another copy of Windows 95. If I'm going to get pushed to the wall on price anyway, I'd as soon they had XP. At least that way I have a prayer of selling them the current rev of Office or something. Plus I won't have ten million First Time Users calling me on the phone about the brand new Windows 95 machine they just bought from Dell... and you know Dell would do that. Or worse yet, the dreaded call from Citicorp that they just bought 10,000 of the damned things, and now want a support contract on them. Say no, and Oracle becomes the standard for new Enterprise systems throughout the bank.

This same decision making process is what killed all the pre-microprocessor computer companies except IBM. They'd look around at those damned PC's, and to them it looked like if they went into it, every one they sold would come out of the hide of their Big Box business. And they were right... it would have. So they covered their eyes, and pretended that PC's would go away, assuring themselves that, "There is no reason for any individual to have a computer in his home."

There is an almost Tao-like quality to this process. It is precisely the size and success of the established firm that leads it to put itself out of business. During the whole first half of the invasion by the New Thing, it looks like any adoption of the New Thing would only cost more business than it could possibly bring in. So the established firm sits there, watching the ants at the picnic devour its toes. But it lets that go, because it would cost an arm and a leg to save the toes. Then the ants start in on the legs. And it lets that go too, because each leg saved would cost an arm as well. Finally, with the legs gone, it falls into the now-huge swarm of ants, and disappears. This happened to one company after another. And it's not because the people running them were stupid. Every step they took was logical and prudent when they did it. Microsoft could easily go the same way. The boneyard is full of companies that did.

[tortoise]

I don't have time to educate you about Windows Clustering. It's real. It exists. And it works.

Once again, your link is not to anything that resembles true distributed application clustering. Stop giving me links to glorified failover software and start showing me evidence that Windows supports real application clustering.

I'll give you some specific targets that the rest of the universe uses when NOT talking about simple failover solutions: Give me evidence that Windows clustering can do what, say, MOSIX can on Linux (e.g. transparent process migration, memory ushering, etc). Show me that these standard distributed application clustering capabilities currently function under Windows and I will concede that you are correct and I was wrong. I'll be waiting.

[js1138]

Microsoft does not answer the phone to support OEM OS installations. If Dell installs the OS, Dell supports it. I know this because I build machines and install OEM Windows apps, and have to support them. I also support some machines purchased from Gateway, and since their answer to all problems is System Restore, I have to handle those problems also.

[Southack]

giving away Windows 95 as Open Source would be a far easier choice to me than would be the option of discounting to near $0.00 a newer OS like Windows XP. - Southack

"It's not obvious why this would be true. The marginal cost of making the next copy of Windows XP is no higher than another copy of Windows 95. If I'm going to get pushed to the wall on price anyway, I'd as soon they had XP." - Nick Danger

I'm approaching the solution from a different angle.

MicroSoft's option is to either compete for users who are currently switching over to Linux (e.g. the City of Munich in the article for this thread, along with their 14,000 desktops), or else concede that entire segment of the market over to Linux.

As this article points out, MicroSoft attempted to compete.

But there are two ways to compete: one way is to lower the price of your current frontline software (e.g. Windows XP).

The other way is to offer one of your 2nd tier products for free (e.g. making Windows 95 into Open Source code so that it is just as open and in the public domain as is Linux).

If MicroSoft chooses the first way and discounts its frontline software, then revenues from selling Windows XP will plunge as customer after customer starts demanding City of Munich-style discounts on pricing.

Ouch. Why would I do that to myself?!

The second way is cost free. Windows 95 code is already dead. Microsoft has locked it up in a vault and no longer sells it. If I suddenly make it "Open Source", then customers who want "Open Source" free software will now have an option to choose from **either** Linux or from MicroSoft. So now I'm effectively competing with Linux.

Moreover, my first tier clients are still going to be paying top Dollar for my fresh releases such as for Windows XP, as I haven't had to discount its pricing if I choose this option.