Learning from LiMux

The LiMux project–Linux in Munich–is approaching the finish line.   LiMux was the largest municipal open-source adoption project in the world when it began in 2003, and was widely viewed as a test case for public-sector adoption. But the path proved difficult–the original 5-year transition became 8 (and counting).   Now, some 9000 of the city’s 12-15,000 desktops run Linux, with the remainder scheduled for migration in 2012.

Florian Schieß, deputy head of the project, recounted some of the difficulties to Heise online in 2010:

Schießl explains that the process ran smoothly in some departments, but failed to progress in others, because the “technical back end structure hampered or even declined cooperation”. They found implementation errors in basic server protocols and proprietary tools which were not compatible with any other software-based management solution. “The lack of open standards for interoperability and the domination of lock-in interfaces was awful,” bemoans the Linux advocate. He notes that this only becomes apparent when you try to make yourself independent of a single vendor, and stop being a “happy slave”.

In this context, it seems worth revisiting Volker Grassmuck’s great 2005 account of the project, which we published as part of a collection called The Politics of Open Source Adoption (SSRC 2005), aka POSA.  Now, I will admit that a six-year old account of municipal open-source politics in Munich may not have the mass appeal that we strive for on this site–and I’m momentarily at a loss for a Hollywood movie tie in.  But the bigger picture remains, I would argue, of more general and current interest:

Our project began with the observation that accounts of the F/OSS movement, to date, have been oriented mostly by the improbable fact of F/OSS’s existence. We propose that, at this stage of F/OSS development and advocacy, we can begin to ask a different set of questions—not how open source works as a social and technical project, or whether open source provides benefits to a range of constituencies (in terms of cost, security, etc.), but rather how open source is becoming embedded in political arenas and policy debates. … It is our argument that the limits of F/OSS adoption reflect, in part, a limited capacity within the F/OSS movement to document, compare, and draw lessons from these processes.

Now, that was 2005, and–given writer’s lag–the chapters really spoke to the situation in 2003-2004.  I haven’t stayed closely engaged with these issues since then.  But I have wondered: has this situation changed?   Is there now a more formalized body of knowledge or best practices that makes F/OSS adoption–both the political decisions and the actual implementation–easier in the public sector?  Relatedly, how has the evolution of proprietary systems (now not just Microsoft but also web services like Google apps) changed those calculations?

The POSA project made a modest effort to create such a body of work by tapping the (hypothetical) pent-up interest of F/OSS advocates in documenting their experiences.  We tried it the easy way: we put the report on a wiki and invited editing and expansion.  We pushed for contributions for about 6 months, including offering prizes for the best contributions.  But the incentives around the wiki didn’t work and–despite a few additional contributions–POSA 2.0 never materialized.

But POSA 1.0 left some very good accounts of the landscape in 2005–of which Grassmuck’s history of LiMux and, I think Rishab Ayer Ghosh’s piece on EU strategies have aged best.  Here’s Grassmuck:

Munich considered dependency on Microsoft to be acceptable as long as substantial independence remained in crucial areas, such the ability to set up backoffice services without MS products. Unilog [a consulting firm hired by the city] observed that this independence had become increasingly problematic. Operationally, Microsoft’s product policy had forced Munich to migrate its operating systems even though there was no functional reason to do so. Strategically, the current generation of MS products, because of their high degree of integration, creates pressure to deploy additional MS products. It is, for example, nearly impossible to run clients securely under XP without the corresponding MS backoffice services. Microsoft did garner high scores for the reusability of existing staff know-how and for its de facto standard data formats….

Microsoft informed City representatives of a recently concluded general contract with the Federal Interior Ministry in which it guaranteed that it would not abuse its monopoly on pricing and contractual conditions over a period six years. This level of cooperation also promised better responsiveness by Microsoft in implementing public administration standards and interfaces into MS products, and in answering the security demands of the public administration. Even though Microsoft did not discount its products outright, the new arrangement implied a savings of several million Euros. These savings came in several forms—especially better synchronization of update costs with the actual migration of workplaces. By continuing to use the XP configurations until 2010, costs for a follow-up migration after 2007 could be dropped. This offer raised Microsoft’s score on both financial and qualitative strategic criteria…

At about the same time, a number of internal Microsoft documents were leaked to the press, including a confidential e-mail, written a year earlier, that offered a glimpse of the company’s strategy to combat its free competitor. “Under NO circumstances lose against Linux,” wrote Orlando Ayala, then the top sales executive at Microsoft Corporation, in an email sent to senior executives, including [Steve] Ballmer, both vice presidents, the company’s top lawyers and the general managers of Microsoft operations in Asia, Europe, Africa and the Middle East. As reported by the International Herald Tribune, Ayala laid out a strategy to dissuade governments from choosing free alternatives to the ubiquitous Windows operating system, especially on desktop computers. He told executives that if a deal involving governments, educational systems, or other large institutions looked doomed, they were authorized to draw from a special internal fund to offer software at a steep discount—or if necessary, for free. The “Education and Government Incentive Program” was intended to “tip the scales” toward Microsoft in these deals, but the fund was to be used “only in deals we would lose otherwise.”

Among these documents was another e-mail written by the head of Microsoft’s services department, Mike Sinneck, and sent two days after Ayala’s memo. Sinneck gave details of Microsoft’s “Business Investment Funds,” which earmarked $180 million in 2003 for discounts on consulting services—especially in the server market where GNU/Linux is the company’s strongest competitor….

In an interview with IHT, the chairman of Microsoft operations in Europe, Africa and the Middle East, Jean Philippe Courtois, defended the discounts by pointing to the similar tactics of its rivals, e.g. Sun Microsystems’ free distribution of StarOffice. But the anti-trust assessment would be different for Sun than for Microsoft. Under EU law, a company that holds a dominant market position is prohibited from offering discounts that are designed to exclude competitors from the market. With Microsoft already the target of several antitrust investigations by the European Commission, the message the internal documents sent to Munich politicians was clear: a decision in favor of Microsoft under these conditions would be vulnerable to antitrust complaints by competitors.

The surprise visit by Ballmer to Munich Mayor Ude served to draw even more attention to the strategic decision and its potentially far-reaching effects. The external effect was to create a wave of sympathy for the Munich decision makers from many sides, urging them to stick with their original choice of F/OSS. Internally, it brought those still hesitating in the City Council firmly behind the anti-Microsoft decision. Since the Council had already announced that the qualitative strategic criteria regarding long-term sustainability would be decisive, Microsoft‘s price reductions were largely moot. Without the Ballmer visit, Hoegner concedes, it would have been more difficult to garner the complete support of the City Council.

Wrapping all this up, Robert Latham and I drew some provisional conclusions about the politics of open-source adoption:

• Open Source actors are becoming more adept at forum shopping—the pursuit of a policy goal in one venue to influence an outcome in another. Although the direct benefits of F/OSS advocacy in some forums are not always evident—the example of WSIS figures prominently here—advocacy efforts in official venues contribute to a larger arc of legitimation that feeds back into other adoption contexts. FOSSFA’s use of WSIS to legitimize its status in Kenya’s ICT debate, documented below, is an example. Because developing countries, especially, often look to the UN system for guidance on ICT policy, the effects of UNDP or UNECA endorsements can extend beyond the rather modest ICT programs of these organizations.

• Intermediaries are crucial. Technical staff needs to be involved at the outset of any F/OSS adoption or migration, but they are rarely sufficient to achieving it. This is not simply a matter of the lack of attentiveness, on the part of technically-adept F/OSS developers, to the needs of ‘average’ users. Successful F/OSS migration depends heavily on the ability of intermediaries to (1) translate technical issues for wider constituencies, and (2) connect F/OSS’s characteristics to other social and political values, including but not limited to cost. This mediation extends F/OSS coalitions outward, and also has an important feedback effect as non-technical constituencies begin to buttress (or require) technical staff support for F/OSS. This can be significant in organizations where technical staff is internally split about the costs and benefits of F/OSS adoption. Any number of actors can fill these roles, from charismatic and broad-minded software technicians to professional consultant/intermediaries, such as eRiders.

• F/OSS is likely to be explicitly integrated into party politics where the state plays a prominent role in technology-centered development. In such contexts, F/OSS offers a powerful and distinctive technology agenda that aligns most frequently with left-leaning critiques of globalization. Brazil is the most prominent example of this alignment—as well as of the party-identified internal opposition that it generates.

• F/OSS adoption requires a substantial private sector for software support and other technical services. Richer, better-organized migrations (e.g., LiMux) can contract multinational F/OSS providers such as IBM and Suse to manage the transition, and also undertake efforts to promote the growth of local support services. Poorer candidates, such as Kenya or the civil society sector described by Coleman, face a chicken-and-egg problem in which the service sector is often unable to provide the technical and human resources to support large-scale F/OSS adoption.

• Cost estimates are often unstable or incomparable across contexts. Because the marginal cost of software production is near zero for both F/OSS and proprietary software providers, proprietary bidders have a significant capacity to match F/OSS’s cost advantages in licensing. For large-scale software migration, F/OSS and proprietary bids often end up on similar terrain, dominated by the cost of services and by guesses about the future (e.g., about the behavior of proprietary vendors and/or the development trajectory of F/OSS). Morevoer, the ‘value’ of major F/OSS characteristics, such as freedom from vendor lock-in or capacity to scrutinize the code, is not readily quantifiable. There are relatively few cases in which F/OSS adoption is unequivocally cheaper than proprietary software, and these are confined primarily to contexts of ‘first deployment’ of IT infrastructure, where transition costs (from, e.g., MS Windows) are low. Think Extremadura, not Munich. In decisions at the margin, social and political preferences can play a decisive role.

• Large institutional and municipal actors (e.g., Munich, Paris) are beginning to make strategic use of F/OSS as a bargaining chip in negotiations with commercial software vendors. Microsoft and, presumably, other large commercial vendors have dedicated resources to undercutting competitive bids from F/OSS providers, but there are clear limits to this strategy. Small actors will not be able to benefit from this form of market power, but they will benefit from the network effects that large scale F/OSS adoption generates.

• Clumsy intervention by Microsoft is the surest road to F/OSS adoption. A very short cognitive distance separates perceptions of Microsoft’s aggressive dominance of its software markets from perceptions of a bullying, US-dominated model of globabalization. Overt Microsoft involvement in local and/or national politics—as in the cases of Brazil and Munich documented here—can rapidly close that distance and activate a much more explicitly political set of commitments to F/OSS, rooted in local or national desires for autonomy. Microsoft’s characterizations of F/OSS as akin to communism or of software choice as best left to the unfettered market generally reinforce this process.

• There are many software markets in which Microsoft is a non-issue, such as in the healthcare sector. F/OSS politics operate differently in contexts where private monopoly power is not the primary foil. Moreover, F/OSS advocacy has a different opportunity structure when extensive government regulation is the norm, not the exception. In many of these settings, F/OSS must define its comparative advantages not against the proprietary software model per se, but against more powerful articulations of open standards, designed to promote interoperability. F/OSS advocates are generally geared for the battle over the desktop, but F/OSS presence in more specialized fields is often weak.

So where are we, almost seven years later?  Pointers welcome.

Happy Holidays, everyone!