1Free/Libre/Open Source Software (FLOSS) has become an important economic issue, and is changing the computer industry, with the decline of some businesses, such as the Unix producers (SCO bought by Caldera, SUN bought by Oracle, etc.) An increasing number of companies are getting involved in the communities of development (Lakhani, Wolf, 2005), potentially changing the structure and the aim of the volunteer-based organization, leading Fitzgerald (2006) to speak of “OSS 2.0”. The aim of this paper is to propose an analytical framework to better understand the roles played by firms in the communities and to see at which level industrial business strategies can match with the communities’ organization.
2So from the perspective of the analysis of roles proposed by Gleave et al. (2009), this work sets out to complete the studies of firms’ behavior in FLOSS communities, firstly by defining the role of firms on the basis of their core markets and secondly by studying the questions raised by firms’ behavior for the commitment of developers to communities and for the long-term viability of the latter. It seeks to answer some research questions raised by Fitzgerald (2006), on the “elaboration of business models” and the “Value for Money v. Adhering to Acceptable Community Values” issue. The article starts from a discussion of the literature regarding relationships between firms and FLOSS, in particular an analysis of the reasons why firms get involved in the communities (section 2). We propose a discursive analysis leading to a classification of the different businesses built using FLOSS (section 3) and explaining how firms integrate FLOSS communities into their industrial strategies (section 4). Conversely, in section 5 we discuss the role of firms in FLOSS communities. As a conclusive section, we seek to develop serious studies and tools for measuring and assessing the involvement of firms in FLOSS communities.
Previous work on firms and floss communities
3In the last decade, an abundant and growing literature has discussed this question. We believe that this literature can be split into two approaches, two points of view based on two data sources, the first looking at the production side and the second at the market side to explain firms’ interest in FLOSS.
4The production side starts from the communities and evaluates the level and manner of firms’ involvement in them. The research agenda, as put forward by von Hippel and von Krogh (2003), is to understand how firms get involved in a specific community and what these communities can offer them. From an organization science point of view, the question is how agents organize themselves to manage distributed innovation and under what conditions firms can capture part of this innovation for business purposes. Here, we are close to the analysis of firms’ social roles in online communities (in the sense of ‘social roles’ as defined by Bonaccorsi et al. (2006). The advantage of such studies is the availability of data from communities.
5These roles are not directly studied, as these works examine the roles played by individual people. Some of these individuals belong to companies, and the roles of the companies are deduced from the roles of the users. However, their exploitation leads to crucial results for understanding the links between business and open source. In 2005, Lakhani and Wolf (2005), analyzing a survey of 287 communities (i.e. people active in FLOSS development projects), showed the importance of business participation in FLOSS communities, as a majority of their respondents are skilled and experienced professionals working in IT-related jobs, with approximately 40 percent being paid to participate in the FLOSS project. Henkel (2006), studying the “embedded” Linux system, showed that business involvement pursued several strategies and that firms did not reveal all the codes they produced but rather carefully selected their contributions. Dahlander and Wallin (2006), in a study of the “GNOME” graphic interface project, based on the theory developed by Teece (1986), argue that by hiring developers who participate in this development project, firms try to control a complementary asset important for building their products and services. Iansiti and Richards (2006) identify, amongst the various FLOSS projects, a “money-driven cluster” where IT vendors’ motives are economic. In this cluster, significant investments have been made in projects that will serve as complementary assets to drive revenues to vendors’ core businesses” (p. 26).
6However, these authors were looking at already well-established communities, where the software developed is shared by numerous actors, people or firms. This can hardly explain why some companies, like MySQL AB, which owns the entire eponymous database software, open source it and yet remain responsible for most of its development, as if it were the core asset of their business . On the other hand, from a strategy and management sciences point of view, some scholars have looked at the business side and explained the use of FLOSS products in terms of the characteristics of the market and firms’ positioning on this market.
7The two main issues here are the definition of a business model (what do you sell when choosing FLOSS?), and the links between a business model and involvement in communities. Chang et al. (2007) have surveyed the literature on the different “FLOSS business models” and classified them “into five types: (a) Support Contracts; (b) Split Licensing; (c) Community; (d) Valued-added closed source; (e) Macro R&D Infrastructure”. They have looked at the advantages and disadvantages of each model, proposing a case study for each, but without investigating involvement in the communities, and the market conditions under which each model is most efficient. Surveying Italian firms, Bonaccorsi et al. (2006) have proposed a definition of “FLOSS based business” and different reasons for firms to participate in FLOSS development. But they have not really explained the link between the kind of business model and the degree of involvement in communities.
8The first to establish a link between these two aspects were Finnish scholars. Surveying Finnish firms, Dahlander and Magnusson (2005, 2008) showed that to a large extent, the variety of firms’ involvement in FLOSS can be understood in the light of the position of software in the firm’s business model (as core activity or not). Again using data on Finnish firms, but at product level, Koski (2007) remarked that “it seems that factors other than those typically found to explain differences in entrepreneurial innovation behavior such as firm size and age account for the differences in the product and license type strategies of the software companies. [His] data indicate that the firm ownership structure has a major influence for the software firms’ product-level business strategies” (p. 123).
9If these studies prove the link between the market and involvement in FLOSS, they suffer from some limitations. They are rather descriptive and do not propose explanations for the variations in firms’ behavior in similar markets: why do Asus or Dell install Linux on their computers without contributing to its development while HP and IBM do contribute? This may be because these authors look at the level of firms without looking at the differences between the different branches of the industry, as West (2003) did on the server/operating system market (what he called the “platform market”). The present work therefore belongs to this second category of approaches. It proposes a systematization of West’s approach to the whole computer industry.
Floss as a strategic tool in firms’ hands
The basic conditions of the computer industry
10There is a wide diversity of actors in the industry in terms of both products and size. Successive waves of innovations and company strategies have led to a progressive reshaping of industry borders and structure. For example, Internet has impacted software production, pushing firms to include more services in their offers, designing new ways of selling software-based applications, such as SaaS (software as a service) (Cusumano, 2004, pp. 86-127, Campbell-Kelly, Garcia-Swartz, 2007).
11However, the foundations of the industry (the basic conditions) have remained unchanged since they were described by Steinmueller (1996); Cusumano (2004): IT products are built by assembling hardware and software units in a given architecture, and these products (isolated or integrated into networks) are used as parts of information systems and solutions.
12On the basis of such technical organization, it is possible to distinguish three large types of “vertical specialization”: i. component producers, ii. computer and IT device suppliers, iii. the market for applications. All these segments are concerned with software production, as even chipset manufacturers have to deal with the operating systems embedded in the machines containing their components. However, since the beginning of the 1970s, some firms have specialized in software production and, since the beginning of the 1980s, in software publishing (packaged software). According to Cusumano (2004, chap. 2), the application market can be divided into service and product, and on the product side into business specialized offers and global, platform offers:
- The “packaged software” business solution providers: combining standardized goods and customized services, they have been highly successful in the field of professional dedicated solutions, both for management needs (ERP such as SAP) and for technical software (middleware applications, compilers, development tools like those from Ilog) or branch-specific applications (like software for computer-assisted design or computer-assisted manufacturing from Dassault Systèmes or others).
- The platform producers: on the one hand, software publishers have broadened the scope of their offer by supplying a variety of application tools that can be combined with their core product or by producing multiple versions of the latter. This enables them to better meet users’ specific needs while keeping production costs down. The archetypal example of such a “platform strategy” is Microsoft, which now offers different versions of its operating system for servers, corporate users or private individuals, as does its open-source competitor Red Hat. The same kind of strategy can be observed for Oracle, which sells professional applications developed on its database technology, and which has recently bought BEA and SUN, after other takeovers, to enlarge its applications portfolio . Another illustration is given by Symbian in the field of operating systems (OS) for mobile applications.
- The architects: on the other hand, service companies, especially the large ones, such as IBM and Cap Gemini, have tried to master a wide range of software technologies and products which they can combine and adapt to the constraints and the existing equipment of their customers. They intervene as “architects” of their customers’ information systems. Smaller local service companies have a similar position, also providing infrastructure services but on a smaller scale (single server maintenance instead of a large IT infrastructure) or aimed at specific professional needs (like IT infrastructure maintenance services in the food industry).
13Users play a double role, deriving from both their economic and technical standing. Depending on the market, and especially their bargaining power in it, the users are more or less able to select the (technical) offers. At one extreme, users and contracts in the global service/architects market are related to large structures, with substantial buying capacities and generally endowed with significant technical skills. So they are likely to influence economic and technical choices.
14We will distinguish three main types of users according to their relation to the product and the technology (Zimmermann, 1995; Kogut, Metiu, 2001; von Hippel, 1988, 2002). The first is the category of “Naive customers or users” (denoted N) who are not endowed with noticeable technical skills and do not individually weigh very much in economic terms. They are overall sensitive to prices and even if they may react to new characteristics of products or to branding issues, they are not capable to translate their needs and satisfaction into technical terms. The second is the category of “Kogut-Metiu Users” (KM)  who are not able to contribute to software development but can generate new features or innovations by revealing their own needs. KM users are sensitive to price and quality arguments. The third category is that of the “Von Hippel Users” (VH) who may act as “sources of innovation” (von Hippel, 1988, 1986), capable of contributing to software development by proposing improvements or modifications, either developing them themselves or at least designing the technical specifications.
Firms’ strategies regarding FLOSS
15In a market where the dominant user is naive, firms can only differentiate on prices, as it is the only signal understood. The more competent users are, the more firms can differentiate vertically on quality or horizontally via the creation of niche products for specific needs and users.
16If we focus on computer and software and services, user competency is a key parameter for the structure of the industry and for explaining how firms consider software and how they invest in development. It can be summarized in the following way (Jullien, Zimmermann, 2009, for detailed analysis): the more skilled people are, the more the strategies increase because it is easier to differentiate horizontally. This explains the industrial appropriation of FLOSS, its variance from one market to another and the competitive advantage it gives.
17This may be the sector where the link between market structure and FLOSS business practices is the clearest. In the server market, producers have habitually provided proprietary solutions with proprietary Unix . The rise of PC servers has allowed some users to avoid such a bundling problem; moreover, using Linux or another free Unix means a cheaper offer (vertical advantage) reusing Unix programs (content) portfolio. Here, suppliers are dealing with highly-skilled VH clients that have obliged them to adopt FLOSS. The competitive advantage of FLOSS is its openness, allowing it to be tuned to specific needs, as proved by Google, which runs more than 40,000 servers with “customized versions of Linux” .
18On the notebook segment, where users are mainly naive, competition is generally based on prices. When Asus entered the market with its eee-PC, it used Linux for price reasons, because Microsoft Windows Vista was too costly in terms of resource needs and price to be competitive. Since then, observing the success of this market, Microsoft has designed a specific, downgraded version of Windows XP for these computers, and today, Linux market share has dropped to around 10% . In this case, free software is considered as freeware and does not seem to provide any competitive advantage. It has simply been used to force the monopoly to drop its prices.
Software and service
19We have already shown that Linux succeeded where users were VH. For price reasons, of course, but also for quality reasons: people could choose a system better adapted to their needs because it is more flexible, as did Google. For a platform editor, the attractiveness of its platform is a growing function of the applications available on it. The asset of a FLOSS platform is to allow the creation of a community of application providers, who will adapt their product to their platform. And as these applications are not at the same level of maturity, since they may be incompatible, there is a need for somebody to organize this compatibility, helping firms to select the applications they need. In a word, the “3A services” (assistance, assurance and adaptation to use).
20Red Hat is the best example of such a strategy, with on the one hand, the Fedora community to foster “innovation” , the availability of applications on a Linux platform, and on the other hand, the Red Hat company, when you want to buy assurance and assistance .
21A growing source of revenue comes from “3A services”. Currently, the main evolution for those firms is to switch from a demand pull strategy (functionalities are developed to stimulate/create the demand) to ‘on-demand’ development (development when required and paid for or carried out by the users). Therefore, FLOSS is used to increase the business feedback from users, considering openness as a way to reduce transaction costs and as a signal of quality. This explains why open source business products are developed mainly in “business” software (ERP, computer infrastructure software like compilers), where there are many users prepared to pay for configuration, maintenance or assistance services. We will distinguish two kinds of structures, again in terms of the computer skills of the users:
- When users are VH software professionals, we find one firm organizing cooperation around its brand-name product. The producer approves the contributions, ensures stability of the tool and helps developers to use it. If some individual contributor becomes important (in terms of contribution volume/quality/innovative aspect), s/he may be hired by a producer, with reduced recruitment costs and risks (ACT, MySQL and some small services companies use this method). By contributing to innovation, the developers (and possibly companies using the tool), are therefore guaranteed that their needs will be taken into account more quickly and integrated into the product (which is a fundamental factor in reducing costs, according to von Hippel (1988)).
- When users are more KM, firms are more service-oriented. The open source asset is more of a flexible, adaptable input, developed by a consortium of information system consultants sharing and co-developing the tools their business is based on, with sometimes a two-level organization, where a software producer and editor of a tool deals with information system resellers. The text-book examples are Compiere ERP and CRM, or Zope Content Management System (CMS), where there is an editing firm, which sells its services and products as in case one, but mainly to “partners”, service companies, as shown in figure 1. In the second case, resellers act as service-providing, architecture companies, which we shall discuss in the following paragraph. In both cases, the interest of FLOSS is its flexibility, allowing it to be adapted to the client’s needs, and the license, which guarantees that the product will always be available, as the modifications performed by other firms decrease the costs of development for each member (as they are shared) and facilitates and accelerates a global presence for the product.
The Compiere “ecosystem” (taken from: http://www.compiere.com/partners/)
The Compiere “ecosystem” (taken from: http://www.compiere.com/partners/)
22When naive users are dominant, firms do not seem to be able to carry out direct business based on FLOSS products. Even if some FLOSS offers exist for this segment, such as Open Office or Firefox, their market share remains small . We do not know of any firms dedicated to them, and we believe that firms’ support is more a consequence of platform or hardware providers’ strategies: they need these commodities for their platform to be adopted by VH or KM users, and FLOSS is a means to create a consortium to develop it (as SUN does by supporting Open Office development).
23As Horn (2004) pointed out, the assembly of different components requires access to the source codes (problem of compatibility), and their adaptation to different needs (of users and of other components). They must be available in the form of open-source software (therefore legally modifiable). The competitive advantage in using FLOSS, in addition to price, is therefore the ability to offer an assembled set of components with greater interoperability, which should increase the quality of the final product, on a market where the quality of services is one of the recurrent problems (De Bandt, 1995). Revenues are generated from assembly and adaptation services, as is the case for any traditional service company. The only uncertainty about the model concerns the availability of the components: who will develop them and who will maintain them? Moreover, the customers of these companies may already have (proprietary) programs installed that need to be taken into account. In the end, an open source strategy could even be a guarantee of means (maximum use of free software), but not a guarantee of the results (use of only free software), unless the customer requests this, since in this situation, s/he has the last word.
24Two kinds of firms use FLOSS today: newcomers who specialize in FLOSS architecture, using FLOSS as a vertical (price) and horizontal differentiation asset , and incumbents, such as IBM for its service activities. Traditional service firms like Cap Gemini are more agnostic with regard to the technologies used and the intellectual property regime involved. They will generally follow the customers’ demand, which depends on their ability to keep up with the development of the project. These customers are most often large organizations, skilled computer users who are receptive to the opportunity to integrate the most advanced software components, developed under FLOSS licenses. So they are becoming increasingly involved in FLOSS as the market grows and matures . Global service firms’ websites are quite explicit about this strategy .
Dependence on users’ skill
25What seems clear after this rather qualitative analysis is that the users’ skill level is important to understanding the degree of involvement of firms in FLOSS. When users are naive, firms may use FLOSS, but only for price reasons, in the same way as they might use freeware. The more VH the users are, the more complex strategies regarding FLOSS are, and the greater firms’ involvement and participation are. In some cases, when users are VH, firms may even produce FLOSS and animate the community like Ada Core Technology for Ada 2005 and MySQL AB for MySQL data bases do. But in any case, FLOSS is regarded as open source software. This means that firms use FLOSS for technical reasons (sustainability, flexibility) and for innovative reasons (increasing the speed and quality of feedback).
Communities and the firms’ strategies
26Firms will behave differently according to the significance of FLOSS software or the FLOSS community in the specificity of their offer: is it a complementary or a core asset for this offer? According to the theory (Teece, 1986), if it is a core asset, then the firm should invest a lot to manage it, if not, it should not invest at all, buying or using it as it is (as an “off-the-shelf” component).
Naive users, the freeware strategy, the community as a commodity
27As the aim of using FLOSS is then to propose the lowest price possible, firms will not invest more in FLOSS development than the effort needed to adapt the software to their product(s): otherwise it would increase their costs. So FLOSS may be seen as a free commodity, a freeware.
Firms’ FLOSS strategy according to the skill of the users
Firms’ FLOSS strategy according to the skill of the users
Business packages for VH users, the community, a core asset
28As far as business packages are concerned, the specific asset of the producers lies in their package knowledge and in their capacity to manage the dynamics of evolution. This makes the open sourcing of a software the specific asset of the firm which owns it: on the technology markets where the customers are computing developers, revealing the code facilitates cooperation. The producer organizes the collaboration in a “symbiotic” relationship (to borrow the term used by Dahlander and Magnusson, 2005). By providing their own innovations, developers (possibly companies using the tool) are thereby assured that their needs will be taken into account more rapidly and integrated into the product, a crucial point in reducing their costs (von Hippel, 1988). From the producers’ point of view, this cuts R&D costs as the users provide them with new feature requirements and, more original, implementation; on the other hand, only the firm that incorporates the contributions is capable of verifying and guaranteeing their correct functioning and helping customers to use it. So a FLOSS-based package model means that the firms which publish the software remain heavily involved in its development in order to control it. As their core competence lies in the management of the software published, firms should only invest in the software they publish, and the involvement of salaried developers in other projects should not be encouraged.
Services to VH and KM users, the community as an complementary asset
29In between these two extreme cases of involvement in communities are service firms. The only uncertainty in their FLOSS business model arises from the availability and the quality of FLOSS components: who develops them and who maintains them? They therefore need to evaluate these components and monitor their evolution . This need for evaluation and control increasingly depends on the importance of the component for their business and that of their clients. We may even formulate the hypothesis that the more skilled the users are, the more the firm must master the technology, because of the growing level of complexity of the feedbacks and demand.
30And to be able to integrate knowledge and innovation from the open-source communities, open-source firms have to develop internally efficient capabilities of absorption. This is an essential condition to capitalize and internalize the communities’ contribution and the users’ feedbacks to improve their own product quality. Dahlander and Magnusson (2008), working on the relations between firms and open-source communities, show that those firms need “to develop sufficient absorptive capacity to benefit from external developments, not only to identify useful external knowledge, but also to assimilate and apply it”. This is what has been called a “commensalistic approach” (Dahlander, Magnusson, 2005). This corresponds to the more general assertion from Cohen and Levinthal (1989, 1990) about the necessity for a firm to make internal efforts of R&D a prerequisite for the absorption of external technology.
31We believe that this reflects a change in the technologies used and therefore in the complementary assets these firms need to manage, not really in the core competences. Traditional architect firms are not involved in FLOSS development, as they do not use these technologies. But they may have other processes for monitoring the evolution of the complementary asset - the technologies they use. They may participate in publishers’ training sessions, or conclude a “global alliance” with their key partners, as Cap Gemini does .
Roles of firms in floss communities
32These results are consistent with Teece’s theory (Teece, 1986; Teece et al., 1997), if we consider the FLOSS community and the evolution of a technology as the asset (see table 2). If firms see that as a source of technology they do not invest. They do invest a lot if it is the core of their business to guarantee efficient integration of the innovations coming from this community. In between these two extreme cases, the community is seen as a complementary asset, as firms try to participate to follow and control the innovations coming from the users.
Firms’ investment in the asset “community”
Firms’ investment in the asset “community”
33Thus firms may assume the different “roles” of the FLOSS organization onion model  (figure 2): core developer (like MySQL), developer (IBM with Apache), bug fixer or reporter (Compiere “partner” companies) or user (Asus), and this mainly depends on their users.
Onion model based level of involvement in FLOSS communities targeted by firms regarding their business and the skill of their users
Onion model based level of involvement in FLOSS communities targeted by firms regarding their business and the skill of their users
34This raises a series of questions about the relationship between firms and communities which we shall now discuss. Two main aspects need to be addressed. The first is related to the interaction between two kind of organizations: firm and community. In terms of FLOSS model efficiency, what competitive advantage does a firm derive from participation, and what consequences does it have for firm organization (the management of their employees involved in the community) and for community organization (in terms of governance).
Integrating firms into studies of the FLOSS ecosystem
36The origin of the open-source rationale remains that of developer-users pooling their development efforts for their own needs, with the aim of creating better access to efficient tools for everyone. Whether this voluntary collaboration is initiated by individuals or by a firm, the adhesion of user-developers is a key condition of its success.
37The management of the relation between an open-source firm and the related communities is thus of crucial importance, as a recent literature has already emphasized. Ãgerfalk and Fitzgerald (2007) have observed that to preserve the co-existence and co-operation of two types of organization based on very different albeit not contradictory rationales, firms must, in a nutshell, “not seek to dominate and control process”, “provide professional management and business expertise” and “help establish an open and trusted ecosystem”. They view such interaction as osmotic rather than parasitic, as firms’ resources reinforce the sustainability of communities . These considerations could easily be extended from the sole framework of the communities to the whole user base.
38On the other hand, with regard to community governance, do organizations weigh the same as individual volunteers, or, by their size, their financial capacities, do they influence the evolution of the communities (in terms of organization, goals, etc.)? For instance, what does “IBM is treated the same as any other member of the Apache Group, albeit one with deeper pockets and more developer resources to spare” (Fielding, 1999, p. 2) mean in practical terms?
39Actually, if we look at the level of involvement of firms, summarized in table 12, it can be related to different structures of the firm-community relationship, or the kind of structuring of communities by firms:
- single firm community, when one firm controls the kernel and thus makes the main decisions about the evolution of the software and must animate the user community.
- consortium, when firms dominate and cooperate to develop a platform (Apache, Linux). In this case there is usually a foundation where the main decisions are taken, and which can hire the people the most involved in the project.
- open communities, where firms do not seem to play a dominant role.
40In the third case, firms have no real power, but can be more disruptive for the stability of the community: if they use without contributing, they may discourage the main contributors (Foray et al., 2007). If the dynamic of the community decreases, firms will not invest to maintain the community and may forbid their employees from investing themselves more.
41This study is of importance because Open Source initiatives are numerous, in various industries (Balka et al., 2009), but FLOSS remains the movement which has had the biggest impact on its industry and is thus the case of reference in the open source IP model for innovation production.
Firms in FLOSS communities: an ambivalent attitude
42The intervention of firms in FLOSS can affect the functioning and organization of the communities in different ways. Obviously, this depends on the nature of the community, particularly whether it existed before the arrival of commercial players. It also depends on the type of license in force or required by the firm.
43The most natural license for a FLOSS community is certainly the GPL (or its equivalents like the French Cecil) which guarantees the pooling of developments eand protects against attempts at appropriation. There are a few notable exceptions, but these can be explained by the particular context in which they arose. For example, BSD came into being a long way from concerns about free software, in the institutional context of Berkeley University. The aim was to create a common base for a new generation of Unix without opposing possible proprietary developments.
44For the commercial firm, on the other hand, the GPL is often considered contradictory to the achievement of business purposes. This is why the advent of the Open Source Initiative (OSI) was a historical watershed, heralding a new era with the entry of commercial firms into the domain of FLOSS and the emergence of hybrid strategies aiming to combine the rationales of two worlds that appear, at first sight, diametrically opposed (Coris, 2006). Thus Netscape, one of the very first companies to “free” a proprietary code, for its browser Communicator, introduced two new licenses: NPL (Netscape Public License) and MPL (Mozilla Public License), making it obligatory to publish all modifications to the initial code, but without excluding the possibility of including proprietary modules without their source codes (Faucon, Smets-Solanes, 1999, p. 81).
45When a company invests in the free domain by “freeing” a program or a category of software products (like MySQL, or Sun with Star Office which became Open Office), it induces a mode of organization in which it continues to play an important role. As we stated above, MySQL set up a double regime of intellectual property (GPL + some kind of license approved by the OSI  ). Sun chose to abandon its own SISSL (Sun Industry Standards Source License) in favor of the GNU LGPL  from the 2.0 version of Open Office, an indispensable condition for maintaining a large community of contributors for a flagship product capable of competing with the “cash cow” Microsoft Office by targeting a mass market, catering to a wide variety of user skill levels. Here the aim is to weaken the monopoly power of Microsoft rather than generate a direct flow of revenue. In that, it is widely shared by the community of developers and users. Oracle has chosen to follow the same path, while clearly reaffirming at the same time its presence closely linked to the product.
46Of course, the involvement of a firm in a FLOSS project can be seen from two opposing viewpoints. On the one hand, it contributes what may be substantial resources to the project. On the other hand, it can be seen as a kind of misappropriation of the project for commercial or image-related purposes, which can in return have a negative effect on the commitment of voluntary contributors. Thus Mateos-Garcia and Steinmueller (2008) examined aspects of hierarchical structure and authority in open source projects in relation to the diversity of developers’ interests and motivations. For these authors, the presence of firms in these communities can have a regulating effect, in that “these firms may become mediators between large markets of users (with a set of defined usability needs) and the technically ‘savvy’ developer community. By participating in development (and therefore obtaining authority in a project), creating usability labs and sponsoring the introduction of usability features into the open source development process, these firms may favor the alignment of the seemingly divergent interests of developers with those of the potential users”.
As a conclusion: measuring the role of firms
47Investigating more deeply the role of firms in communities would require the collection of serious data to measure the role of firms in FLOSS communities. The main issue is that organizations do not contribute to the projects in themselves, but via people hired to do so. As stated above, the works that exist on this topic are based on the behavior of firm’s employees in the community, and they therefore use the data produced by the community. This suffers from some limitations.
Collecting data on the behavior of firms
48The way scholars operate here is by looking at the data produced by the communities (e-mail lists, source code produced and signed, etc.) People are considered to belong to a company when they sign with a company address.
49The main limitation of this approach concerns the exhaustiveness and coherence of the data collected: some developers do not contribute using their employer’s address, and others do so without their employer’s agreement. However, when firms commit themselves to a community (as IBM can be said to have done with Linux), this technique can be used to track their involvement and their evolution over time.
50But as Gleave et al. (2009) pointed out for individuals, looking at what firms do does not provide information on why they do it or whether this is driven by the management: in a word, on firms’ motivations.
Surveying firms’ strategies
51Looking, at the same time, at firms’ business may provide a good validation of the theory of firms’ roles we have developed here: the link between their involvement and their market.
52We therefore argue that the study of firms’ behavior should be completed by questionnaires or interviews, as Shah (2006) did to track individuals’ motivations to participate. One point that we believe deserves particular study concerns the missions that firms give to developers participating in a community: are they hired to participate, are they hired because they already participate, do they have a specific goal (to reach a certain level, to work on a certain part of the project), or is participation the only thing required?
53This is important for understanding not only the extent of employees’ involvement, but also what firms contribute to the community and thus to the FLOSS ecosystem as a whole.
This means that any developer/contributor wanting to contribute to the official MySQL product has to transfer her copyright to MySQL. http://forge.mysql.com/contribute/cla.php. Once owning the whole copyright, the firm can manage a dual licensing scheme, distributing the product under the license it wants, either GPL or more classical closed license. So a customer who does not wish to reveal further enhancements of the source code must keep these enhancements to herself.
“Ellison saw that if Oracle played its cards right, the confluence of the database, the Internet, and the Web browser could displace the operating system as the focal point of computing and erode Microsoft’s industry dominance”. Brent Schlender, CNN Money, 1999.
In reference to the notion of “frontier-users” put forward by Kogut and Metiu (2001).
See West (2003) for a full discussion of FLOSS strategies in that sector.
See http://en.wikipedia.org/wiki/Google_platform for a summary of the Google platform and useful links on that topic and Stross (2008) for a whole presentation of Google.
See the presentation of Fedora at: http://fedoraproject.org/wiki/Overview
Also very well explained on Red Hat’s website: http://www.redhat.com/rhel/
Some sites estimate Firefox market share around 20% (see http://marketshare.hitslink.com/browser-market-share.aspx?qprid=0 http://marketshare.hitslink.com/browser-market-share.aspx?qprid=0) but they probably overestimate it, as they look at the browser used to visit sites, thus favoring the browsers used by Web intensive users, who are more skilled than the average user, and may use more FLOSS.
As explained by Slatter (1992), one of the main strategies for newcomers to the technological market is technological differentiation. Basing its offer on new FLOSS products can be seen as a way for new service companies to differentiate themselves.
In 2005, Gartner forecasted that “[in] 2008, 95 percent of Global 2000 organizations would have formal open-source acquisition and management strategies” (http://www.gartner.com/it/page.jsp?id=492152). In their 2008 study, they said that “Adoption of open-source software (OSS) is becoming pervasive, with 85 percent of companies surveyed currently using OSS in their enterprises and the remaining 15 percent expecting to in the next 12 months” (http://www.gartner.com/it/page.jsp?id=801412).
See Capgemini’s for instance: http://www.us.capgemini.com/services/serv_overview.asp?Serv ID=30
See Foray et al. (2007) for a discussion of that particular point.
Where evaluation is concerned, it is worth noting that the main service companies in France have published tools to evaluate FLOSS, and they use this as a commercial argument (we would say a signal) to their customers about their ability to evaluate these products. See, for Atos, http://www.uk.atosorigin.com/en-uk/services/solutions/systems_integration/technologies_expertise/open_source/default.htm, and for Cap Gemini, http://www.capgemini.com/services/technology-services/open-source/solutions/.
http://www.capgemini.com/collaboration/alliancepartners/ for classical alliance, and http://searchsystemschannel.techtarget.com/news/article/0,289142,sid99_gci1261207,00.html for those made with the open source world.
This model is described by Herraiz et al. (2006) and has been proposed by Crowston and Howison (2003). In a career in a community, the successive tasks a developer may perform are to use the program, to use the mailing list(s), to report bugs and to fix them, to be a core developer.
The study of FLOSS participants’ motivations has already generated a vast literature, reviewed by Lakhani and Wolf (2005); Shah (2006); Scacchi (2007) among others. A person’s role evolves over time (Jensen, Scacchi, 2007), as do the reasons why s/he participates (Demazière et al., 2006). Whether it is called the “process of joining” (Herraiz et al., 2006), “role migration” (Jensen, Scacchi, 2007) or the “co-evolution of systems and communities” (Yunwen et al., 2004), the evolution follows a “canonical path”, the “onion model” (Herraiz et al., 2006) from none to central responsibility for the future of the community. This is close to the sociological notion of “career” (Becker, 1963), as pointed out by (Demazière et al., 2006; Shah, 2006).
With firms, the career is becoming twofold: the community participant’s career is connected with a more traditional one, the professional career (Vicente, 2008). This has potential consequences on the design of the canonical path, but also on the motivations of “free” developers.
To have some responsibilities in a community, you have to “participate”, to commit to this community, and to behave as expected by the community. Firms change this into more strategic involvement: one can be hired to participate in a community, but with a level of involvement bounded by the employer (e.g. in a service firm). Can you still talk of “freedom” (of expression, of choice, of contribution), which is “one of the roots of the FLOSS culture” (Scacchi, 2007) and a well known incentive to participate, when people are paid to follow their employer’s strategy of participation? As already stressed by Fitzgerald (2006), firms’ behavior may contradict the values of the communities and thus affect their efficiency as ideology diminishes the effectiveness of FLOSS teams (Stewart and Gosain, 2006).
FLOSS employees may face a contradiction, a “tension” (Thévenot, 2007) between the commitment to the community and to their employer.
This also involves the impact of communities on firms’ organization and management (we thank our anonymous referee for this remark). How do firms have to organize themselves to capture the feedback from the communities? This point is extensively discussed by the open innovation literature, as defined by Chesbrough (2003); Chesbrough et al. (2006), or more recently by Baldwin and Von Hippel (2009), but the analysis proposed by Teece et al. (1997) on how firms must organize to settle “dynamic capabilities” could also provide a starting point.
MySQL calls this the FOSS exception, which “allows developers of FOSS applications to include Oracle’s MySQL Client Libraries (also referred to as “MySQL Drivers” or “MySQL Connectors”) with their FOSS applications. MySQL Client Libraries are typically licensed pursuant to version 2 of the General Public License (“GPL”), but this exception permits distribution of certain MySQL Client Libraries with a developer’s FOSS applications licensed under the terms of another FOSS license listed below, even though such other FOSS license may be incompatible with the GPL.”, http://www.mysql.com/about/legal/licensing/foss-exception/
LGPL for Lesser General Public License is a less restrictive type of GPL that allows the use of libraries of proprietary programs in an LGPL development and vice versa.