Open-source: untangling a web of assumptions

Photo Nordine Benkeltoum / Lecturer, LM2O (Centrale Lille), Research Associate, Centre de Gestion Scientifique (Mines ParisTech) / October 19th, 2011

The free software movement when viewed from afar remains poorly understood but on closer inspection reveals a web of surprises. Who knows that around half of all "volunteers" are actually being paid for their contributions? The frontier between commercial and non-commercial activity has become somewhat blurred and the idealized vision of a utopian community actually hides an extremely wide range of actors and entities.

ParisTech Review – The idea of free software is often framed by the notion of “communities.” How, and along what lines have these communities evolved?

Nordine Benkeltoum - The theme of community is often evoked by developers and firms that have come of age within the open-source framework. The reality is that the term free software is more diffuse than is sometimes acknowledged and is the product of a rather loosely-knit network of contributors many of whom vary widely in terms of composition and scale.

A common assumption is that these communities consist of of thousands of developers when the reality is quite different since most of them are based on two or three individuals… These communities are composed of a diverse range of actors. Some consist of thousands of volunteers who share a passion for 2D video games maybe, or system and device drivers to give another example. Interoperability issues are also a major area of focus. At the firm level the preoccupations are rather different and consist largely of source code sharing and the creation of more standardized platforms for the software and services sectors. These objectives often rely on laying foundations for sound governance at both the individual as well as corporate level.

Governance structures for the leaders of the Free Software movement (OpenOffice.org, Mozilla, Linux kernel, etc.) have undergone rapid transformations since the early days of open source phenomenon. Control over Linux kernel development bears virtually no traces of its initial structure and ever since giants such as IBM made the decision to throw their weight behind Linux, with close to $1 billion of investment, the playing field has changed dramatically.

Over the course of my book on the subject, I attempted to provide clear reflection of the patchwork approach of the open-source environment and have focused primarily on the structures as well as internal dynamics of the development process. I chose to avoid statistical samples as to be truly representative this would have limited me to communities of five individuals. Instead I chose to work outward and “elaborate” rather than “verify” a theory, what could be called a theoretical sample (see Eisenhardt in Academy Of Management) has been created through investigating five cases studies and has led to what I believe to be a more nuanced understanding of prevailing trends.

Which examples were you able to draw from?

The first was the European consortium OW2, which began life as a collaborative project between INRIA (the French National Institute for Research in Computer Science and Control), Bull, and France Telecom, and is an independent open-source community committed to providing the best and most reliable enterprise computing infrastructure software. The primary focus is on middleware, a software layer which according to Jean-Pierre Laisné can be defined as “the technical infrastructure dedicated to providing an interface between the operating system, applications, and individuals”. In all honesty enterprise solutions of this type remain a rarity in the open-source world, and are of little significance statistically, and yet the example has relevance, not least for its originality, and can only serve to enrich the model as a whole.

The second case represents the contribution of the business world to the open-source movement and is no doubt familiar to most readers as the world’s leading open-source office suite, OpenOffice.org. In the early 2000s Sun Microsystems announced its decision to make the source code for its proprietary StarOffice suite freely available. At the time this was seen as a strategic move. The project governance was conducted with the explicit aim of challenging Microsoft’s dominant position in the market by offering an alternative. After Oracle’s acquisition of Sun the original development community was split in two with the creation of the Document Foundation, and a fork was created.

Freeworks forms the foundation of the third case study and this Paris based association composed of artists, developers, and other creative types takes a “copyleft”, as opposed to copyright, approach towards creative expression. Their work has particular relevance as it demonstrates just how far the philosophy of open source has moved beyond its original application to software development.

The fourth case examined is that of Kexi, a community of developers dedicated to creating a stable data management system (along the lines of MS Access) that makes use of a graphical interface running on the KDE desktop environment.

The Paris-based Mandriva is the object of the fifth case study and as the creator and maintainer of the Mandriva Linux distribution presents a particularly rich source of information. They provide concrete example of a business model based on a mix of both commercial and non-commercial activities.

As can be seen by the heterogeneous nature of these five case studies my decisions on which areas to explore were based largely on a desire to create a pentagon-shaped model through which we can better view the comingling of various strands of the open-source model, viewing it more as a complex ecosystem where communities evolve by freely passing between five separate but inter-related poles. A small group of developers might for example take the decision to form a company but there is an incredible diversity in the form it might take and most firms are extremely malleable in terms of the way they are organized which takes us back to your original question and the notion of community. The reality is that the motivation and background of contributors is more mixed than most open-source commentators are willing to admit. The anti-establishment origins of the movement and the creation of communities tend to be somewhat over-emphasized to the point where we can’t see the woods for the trees, and the true picture is being lost in this process.

What role do structural models play in bringing some order or centralization to the development process?

Open-source software development has been compared to a bazaar but while this may have rung true at the beginning most communities have created some sort of hierarchy of individual contributors, usually based on the quantity of source code contributed, technical skill, and communication. Public perception has yet to catch up with the reality that the largest and most successful communities are in fact governed according to strict operating procedures. A single piece of software is chopped up into various ‘projects’ which are then delegated to groups of developers who are accountable to the project director. According to those on the inside, the process closely resembles what goes on in commercial software development. In my opinion there are two principle differences however: 1) the majority of the source code is freely available; 2) a progressive model for innovation has been created which tackles problems from two sides, guaranteeing stability and continuity up until the point a “fork” is created, such as was the case with OpenOffice.org and a new piece of software is derived from previous work.

In the free software movement the rules of the game dictate that when there is a conflict between the designer and the programmers, each is free to leave the other’s company and take the shared code in a different direction. This is easier said than done however and most would agree that a significant leap must be made to make up for the information gap between coders and designers and it is often more efficient to abandon previous work and design a new program from scratch. Perhaps this is why forked software is so often the fruit of individuals or groups of programmers that already have an intimate familiarity with a given body of code and its structure.

Popular imagination draws a distinct line between commercial and non-commercial activity in the software industry. This has not prevented Microsoft from dipping its toes in the movement and since 2009 the corporate giant has contributed several drivers to the Linux community. Has the line, so clearly drawn in Bill Gates’ seminal letter from 1976, begun to blur? How do the two spheres interact with each other?

The frontier between commercial and non-commercial activity is constantly shifting in the open-source movement. The free software movement can no doubt be seen as an alternative to the licensing and payment structures developed by corporate entities such as Microsoft. Free software can trace it roots to the anti-establishment philosophy of movements led by such organizations as the Free Software Foundation (FSF), established in 1985, and to loosely affiliated groups of hackers who can be distinguished by their willingness to push the limits of technology and are often at the cutting edge of software development. What all these groups have in common is a shared philosophy that “information should be free.”

It is somewhat understandable that the legacy of a heroic past, led by gurus and organizations such as the FSF, has been hard to shed as the world of free software has evolved but in actuality the movement now maintains significant links with the business community. Open-Source is no longer subversive and plays a large role in corporate strategy. In some cases free software exists as a generic solution for large industrial groups and allows them to increase their bottom line. Open dialogue between enterprises revolves around creating standards for various software components that have been commoditized, and no longer provide the basis for competitive advantage. Database technology provides a useful example as the architecture is well understood by all and software-makers have more to gain through knowledge-sharing as this both lowers overall costs and frees them up to focus on the development of proprietary components, with more restrictive licensing and higher added value.

One explanation for the commercial success of the open-source model has enjoyed is the way in which users are co-opted into the development process thereby providing solutions to problems that if left to the market might never be solved. In fact, it defines the very essence of the free software movement as the users become co-developers and rapidly accelerate the debugging process of any new software. No closed-source developers can match the pool of talent available for free software and in any case, they often have a direct or indirect interest in producing “buggy” software as locking the client in can lead to increased profit margins over the long term. Of course, open-source vendors are no less susceptible to drawing value from complexity, and many ways encourage its creation through the structure, the changes, and the complexity of the programming languages, etc. employed in their offerings. The example of OpenOffice.org is particularly relevant as the expertise required to even comprehend, let alone change or interpret, millions of lines of code is in the hands of a very select few.

We can identify certain communities that are clearly oriented toward producing code to be deployed and utilized in large-scale projects. To take an example, Thales makes liberal use of free software in its mission critical systems for air traffic control and close to half of all flights are now reliant on a system that traces its origins to the open-source movement. Other communities are less interested in commercial activity and are largely supported through grants or benefactors.

While some of the software giants have largely resigned themselves to an open-source future and have expanded their role in the sector through multiple initiatives (e.g., Microsoft’s Port 25 and CodePlex) others have resisted this trend and are investing little in free software despite the fact that many of their offerings depend on it. The Mac OS is based on free software and the Safari browser integrates a forked version of KHTML (developed for the KDE project) but Apple has contributed little to the open-source community and the exchange has largely been one-way.

Can we draw any conclusions about the kinds of individuals that make up the talent pool of contributors to open-source projects?

Moving beyond well-worn stereotypes of the community, such as the fact that it is almost exclusively male, the sociological context under which development takes place throws up some significant surprises. While a large number of contributors are IT professionals who draw from their expertise to write code in their free-time there are others who come from worlds with no direct relation to software development and I have crossed paths with a wide range of characters: doctors, hospital practitioners, lab technicians, even a musician …

Another important point that often gets lost in the conversation surrounding free software is that a large number of contributors actually make a living off their participation in open-source communities. In a study by the European Commission it was revealed that around half of all contributors receive financial remuneration for their work.

Detractors of the open-source model are fond of the bazaar metaphor but while this may have been legitimate in the early days, the statistics clearly show this is no longer the case. Nevertheless, the notion lingers in the minds of popular imagination despite the fact that many open-source communities are comprised of highly qualified professionals. Drawing from a concrete example, the independent enterprise solutions provider ScalAgent Distributed Technologies, which specializes in asynchronous middleware design, was founded by a graduate of the elite Ecole Polytechnique, a university professor, and a PhD all of whom specialize in distributed technologies. In terms of competence it would be hard to imagine a better combination for success.

In fact, a number of communities have blurred the boundaries between what constitutes free software and the world of professional, and highly lucrative, development based on shared code. Intermediaries exist to facilitate interaction and numerous individuals move quite freely between the two worlds. The Mozilla Foundation was established in 2003 and its rise from the ashes of the publicly traded Netscape Group provides a perfect case study. When AOL acquired the company it continued to support web browser development and the foundation exists as a non-profit entity (receiving large donations from some of the biggest names in the industry) and relies on significant numbers of volunteers. To handle contractual arrangements and revenue generating activities a wholly-owned subsidiary was created with the inauguration of the Mozilla Corporation in 2005 and maintains significant links with internet giants such Google and others.

Finally, we must acknowledge the significant number of communities that continue to emerge from public institutions, universities, and most notably research centers. Their contributions are both direct (source code contributions) and indirect (as they are often the only sources of income for free software developers who are employed in these organizations). Richard Stallman was actually working a day job at MIT’s Artificial Intelligence Laboratory before he launched his career as the philosophical figurehead of the free software movement with the launch of his revolutionary GNU Project.

Is there any truth to the notion of a pyramid-like structure under which those at top create value through the toil of unpaid volunteers that make up the foundation?

This is somewhat simplified version of the actual reality. Moving beyond the world of smaller communities we certainly find examples of large-scale networks under which the designers and project leaders receive financial remuneration. But they are not the only ones to benefit as financial gain depends primarily on levels of competence as well as the significance of contributions in the form of source code. The question is almost a non sequitur as the frontier between professionals and amateurs has blurred and between the two poles we find an extremely diverse range of activity that resists explicit classification.

A far more pertinent question would be to interrogate the relationship between developers and users, and the way the open-source philosophy integrates users into the development process. This is where most of the scientific literature is focused as in most cases what really distinguishes one sector from another is not so much technical sophistication or methods of remuneration but rather the relationship that develops between a product and its user base.

I do not want to imply that treating users as co-developers is the only reason for the success of open-source development but it is certainly one of the most salient which, alongside the notion of breaking projects down into individual modules, helps create value and leads to better software. But once again, it is important to avoid getting too bogged down in stereotypes, one of the most tenacious being the idea that software produced from this collaboration between software developers and users is leading towards ever more disruptive innovation.

The actual reality is rather different and forms the subject of an empirical study I conducted alongside Armand Hatchuel (Deputy Director at the Center for Scientific Management, Mines ParisTech and member of the French Academy of Technology) to evaluate the success of software based on the user-developer approach. We began with a thorough survey of close to 300 cases from which we were able to derive the variables that would form the basis of our investigation. From this point we aggregated the evaluations of 125 of experts on the 25 pieces of software chosen for the study to build an assessment model at which point we were able to extend the procedure to 152 pieces of free software. The findings suggest that most innovations introduced by user-developers are technically oriented and based on a problem-solving approach with particular focus on interoperability issues. This flies in the face of the received wisdom that users are the source of disruptive usage-oriented innovation. When we reflect on the results of our study they are entirely logical. What users really want are solutions!

The blueprint for free software has matured and well-established networks for communication are firmly in place. Have we reached a stage where the management model is applicable to other parts of the IT industry and perhaps other sectors of the economy?

Change has been afoot for some time and open-source has become well integrated with the computer industry as a whole. Some companies have been quicker off the mark than others and most corporate strategy is now oriented toward drawing from the most desirable aspects of the model such as sustainability and greater technological independence.

The one major problem that continues to bedevil free software however is the question of finance! The proprietary model, and its strict licensing restrictions, will continue to play a major role in the industry due to the simple fact that certain software continues to require significant investment in R&D and the costs must be recouped somehow. To take an example, CATIA, Dassault Systèmes flagship product management lifecycle suite relies on code that remains a closely guarded secret and will remain so. Other companies take a more patchwork approach integrating free software with substantial layers of proprietary code. A shift of perspective has been made and the worlds of free and proprietary software are no longer mutually exclusive and in most cases have become complimentary sides of the same coin.

References

BOOKS

More on paristech review

By the author

  • Open-source: untangling a web of assumptionson October 19th, 2011

www.paristechreview.com

This content is licensed under a Creative Commons Attribution 3.0 License
You are free to share, copy, distribute and transmit this content

Logo creative commons

5 quai Voltaire 75007 Paris, France - Email : contact@paristechreview.com / Landline : +33 1 44 50 32 89