We may differ in our views about software, but one thing Fred and I have in common is that we both lived for extended periods in Europe. Lately Fred has taken to taunting me about what he calls the "flop" of the big European drive to switch their local governments from Windows to Linux desktops. He sent me a couple of news stories from ZDNet UK that purport to document this trend:

Bergen (Norway) puts "desktop Linux project put on ice"

Birmingham (UK) Linux project "has flopped"

London borough council "ditches Linux plans"

To keep Fred happy I took a look at these stories. But I have to say I wasn't totally convinced by them. Yes, there have been some big Linux desktop projects that crashed and burned. But some of the biggest and most talked about Euro government open source projects – like the ones in the Extremadura region of Spain or the German city of Munich – are still moving ahead. Moreover, it looks like the flops that did occur were due more to the ineptitude of the bureaucrats in charge than any fundamental problems with Linux itself. At Birmingham for example the migration proposal degenerated into a food fight between Ubuntu and Suse partisans before finally succumbing to a vigorous counterattack by Microsoft.

A common theme in the failed European projects is that government officials started off with big dreams about migrating thousands of PCs away from pricey, proprietary Windows to free and open source Linux, then walked away when they realized how complex it would be to navigate the dense thicket of competing Linux distributions and alternative open source application stacks. At the end of the day they chose to take refuge in the simplicity of the single-vendor solution, even at the price of selling their souls to Redmond. In other words, politicians made promises they couldn't keep and then decided to bail on said promises. This has been known to happen before, but it has nothing to do with Linux.

But even though these stories fail to persuade me that large-scale desktop Linux is a pipe dream, there is one thing about them that disturbs me: none of these Euro open source projects – not even the ones that are still moving forward – seems to have been driven by any fundamental economic or technical need. They are all based on top-down politics rather than bottom-up user demand. And that cannot be a good thing for Linux. On the contrary, open source can only win by going to places where users are spontaneously seeking new solutions to problems that existing technologies haven't solved. New technologies rise to fame by creating demand in places that the opposition has chosen to ignore, not by blindly butting up against the walls of long-established legacy citadels.

Each of these European cities has been using Windows-based PCs for decades and has thousands of them deployed. These are not "green field" sites. Over the years they have sunk huge amounts of capital not just into the purchase of Windows and Office, but even more into the training and accumulated experience of the tens of thousands of users who sit in front of the keyboards all day long. And yet these municipal authorities have persuaded themselves that it makes sense to dump all that in favor of open source alternatives which use very similar technology to accomplish nearly identical tasks in ways that are only superficially different.

By attacking the heart of the existing Windows installed base in large organizations rather than going after new markets, Linux champions are barking up the wrong tree. This is not the kind of battle that the open source camp should be choosing to fight, because the odds are stacked against it. I mean, sure, you can argue that Firefox is better than Internet Explorer (I certainly would), or that OpenOffice is "almost as good" as Microsoft Office (hmm, not so sure about that). But you can't argue that they are radically different from each other in the way that, say, Google is different from the Yellow Pages or that Salesforce.com is different from SAP's on-premises CRM suite. Putting user interface issues aside, most open source fans would admit that the initial desktop Linux experience can be a little bit rockier than the typical Windows experience, mostly because of differences in the number and quality of available drivers. And if we bring the user interface back into the debate, it would hard to deny that Apple has historically held the advantage over both Microsoft and Linux. In short, it's tough to see why an open source desktop PC should offer a significantly different or better experience to a large population of legacy Windows users than Windows and Office do.

So what is going on here? Why are the Euro cities tilting at the Windows-to-Linux windmill instead of doing something really creative like dumping all their clunky back-office ERP and HR suites for on-demand applications delivered over the Internet? The official reason they usually give is that open source desktops will save them money. The algebra is simple. Microsoft costs big bucks, Linux and OpenOffice are free, therefore we win. QED.

Of course this argument overlooks the fact that the most prominent Linux distributions, Red Hat and Novell Suse, are not free. But it turns out most of the large European public sector Linux projects that have actually gotten past the talking stage have chosen some form or other of Debian. While IBM, Novell or Red Hat are often seen hovering in the vicinity when these projects first get started, they usually don't make the final cut. It's cheaper and politically more profitable for the elected incumbents in City Hall (or the H ô tel de Ville or the Rathaus) to write a modest check to a local consultant to put together a custom Debian build than to fork over a bucket of cash to some <cough> American <cough> open source vendor.

Inexpensive Debian notwithstanding, my neighbor Fred always throws a tantrum when he hears the "Linux is free" argument. He invariably tries to get me to check out some of the numerous anti-Linux case studies Microsoft has helpfully posted on its web site. These studies supposedly reveal all the budget-crushing hidden costs incurred by Linux migrations. But honestly, good neighborliness aside, who is going to take Microsoft's word for this? Not me, that's for sure.

I would be much more inclined to believe a cost comparison between Windows and Linux if it came from one of the migrating Euro burgs. Unfortunately none of them has bothered to release any kind of detailed financial analysis of its experiences with open source, at least not that I can find on the Internet. However, a little time spent hacking around on Google did turn up this very interesting status report published last March by Munich (in English, thankfully) about its deservedly famous Linux project. The report doesn't have any comparative cost data (which is perhaps why it hasn't been picked up by the Microsoft propaganda machine), but it does give a detailed breakdown of the city's budget for its uber-ambitious multi-year move to Linux.

So what does the report tell us? The Linux story in the Bavarian capital goes all the way back to 2001, when the City Council, according to this document, "demanded examination of alternatives" to Microsoft. At the time the city was still running Windows NT 4 and was worried about NT's lack of compatibility with new hardware (e.g. USB) as well as with future versions of Office (the last version of Microsoft's desktop suite to support NT 4 was released in 2001).

Windows 2000 or Windows XP (released in 2001) would have been obvious upgrade targets. But the city fathers declared that they wanted to pursue "more independence from software producers" and "more equality of opportunities". Munich's IT staff also promised that open source would offer better protection from virus attacks than Microsoft. Last but presumably not least, they expressed the hope that the city would reap lower costs in the "middle term," although an official stated that TCO was not a factor in the decision:

· "We do not have a goal to compare total cost of ownership. Microsoft stopped supporting NT 4.0, so we must migrate."

The city adopted a migration budget of 35 million euros, or roughly $50 million dollars at today's exchange rates (not including the PCs themselves). The goal of the migration is to move 80% of the city's 14,000 PC users to desktop Linux and an open source application stack based on OpenOffice. As of mid 2007 apparently only a few hundred Debian desktops had been installed, but the city still hoped to complete the project by 2009. In other words, Munich will have moved from NT 4 to Debian Sarge in about the time Microsoft will have taken to get from Windows 2000 to Vista SP2. The city's budget works out to nearly $4,500 per desktop user. The largest item in the budget by far is for training – 38% or around $19 million. That's about $1,700 per user, a not insignificant sum. Presumably this money will be spent to train people how to use the KDE desktop and, above all, OpenOffice. Another interesting item is the $7.5 million for the migration of some 13,000 Microsoft Office macros, forms and templates to their OpenOffice equivalents. That's about $577 per object converted. Nice work if you can get it!

The Munich document doesn't cost out the hypothetical Microsoft alternative it rejected – an upgrade to, say, Windows XP and Office 2003 – so we can't compare the TCO for the two solutions. A Microsoft upgrade would certainly have cost a pretty penny. There would have been the new Windows and Office licenses, plus no doubt a recurring Software Assurance fee. On the other hand, less money would have been spent on consultants and format conversions. I'm also guessing that the added protection from virus attacks the city hoped to gain by adopting open source may have been vitiated by its decision to keep using Windows NT years past the end of security updates for that OS.

At the end of the day, although it remains an obvious truth that Microsoft's fundamental business plan is to leverage its dominant position in desktop PC computing to squeeze every possible dollar out of its customers (you and I would do the same if we were in their shoes), it's still next to impossible to make a believable economic case for Munich's Linux migration. In fact, based on the city's published documents, it's clear they never even tried to make such a case. The city's decision-making process was simply not based on any conventional notion of ROI, time-to-payback, or cost effectiveness. No, Munich's choice was motivated by something other than money.

If not money, then what? A cynic might suggest that the Bavarian politicians were merely looking to shift some of their generous IT budgets from an American software company to favored local consultants and public sector employees. For all I know there may be a grain or two of truth in that thought. After all, Steve Ballmer isn't registered to vote in Munich. That said, one could argue that putting money into open source is a better way of dispensing the taxpayers' hard earned kopecks than squandering it on "bridges to nowhere" or any of the other absurd "earmarks" that American politicians routinely slip into legislation in the dead of night. If the German pols really plumped for Linux just to get themselves reelected, then maybe a few American pols should take inspiration from them.

But I don't believe the Munich Linux project is an example of political corruption, however high minded. A deeper analysis of the public statements made by officials in Munich and other European cities embracing open source suggests to me that their true motivation is not self-interest but something more like a desire for self-purification. These pious Burghermeisters really believe that adopting open source will win them the admiration of their fellow citizens. In sum, it's politically correct and it's hip and it's morally ennobling. It's right up there with saving the whales, hugging the trees and rolling back global warming. And it's a low-cost way of reminding us how morally superior the Old Continent is to the New. Of course European municipal leaders aren't the only people who choose open source because of what it says about them. Here in San Francisco, running Ubuntu on your Dell is a fashion statement by urban techno-hipsters. (Interestingly enough, Google Trends shows that among the cities producing the most queries for "Ubuntu" are San Francisco and that other capital of fashion, Milan, along with cool Barcelona and Linus-friendly Helsinki.)

All of this is well and good. As long as uncool commodities like money and productivity are not the primary goals, people should use whatever software makes them feel good about themselves. If there is no trans fat or second-hand smoke involved and no excessive carbon footprint, if no animals or children are harmed, what's not to like? Of course in situations where a net positive return on investment is required, for example in big ticket IT projects in private sector companies, you will not find many people undertaking just-for-the-fun-of-it Windows-to-Linux desktop migrations.

What you will find are plenty of people installing RHEL or SLES on new servers, along with software stacks that include things like Apache, Tomcat, PHP, JBoss, MySQL, SugarCRM and Alfresco. To be sure, these same people are also installing plenty of SQL Server and Exchange Server, along with the inevitable VMware to keep those pesky old NT apps alive. But so what? Linux and open source aren't going to take over the world. It's time to admit that.

The reality is that most PC users working in established large to mid-size organizations in the developed world are simply not going to give up their Microsoft desktops. It just doesn't pay to reinvent the wheel if the wheel you already have isn't broken (even if you hate the Microsoft logo on it). What does pay is leveraging innovative new technology to move to a completely different and more productive paradigm. Desktop virtualization – such as offered by VMware, Citrix or the new Israeli startup Qumranet – is an example of such a paradigm shift. It reduces costs and solves a whole host of management and security problems that a traditional client-side install of Linux with a heavy office suite doesn't even begin to address.

I'm not sure if the gentle dreamers in Munich will wake up in time to the fact that they are investing in an obsolete computing paradigm. But I am tempted to say to them: Wake up and smell the virtualization!