Democratizing production through open source knowledge: from open software to open hardware

Generative hacking opens the market

The GPL has contributed to the transformation of the software production field and, as Nissenbaum (2004) argues, is centrally embedded in the ascent of digital capitalism. Access to the source code, a precondition for software freedom as defined by Stallman, also inspired the development of open source software licences. Like free software licences these require a sharing of code, but they have fewer restrictions on commercial use. As Söderberg notes, open source often allows redistribution under GPL but may not require it, making these licences attractive to software producers (2008). The results of this dialectical movement of knowledge include the incredible success of open source software within the market economy: two-thirds of major websites use open source Apache servers (Netcraft, cited in Cammaerts, 2011). As Weber (2004) identifies, major software companies are now core contributors to open source projects, recuperating the processes that hackers originally linked with radical politics – drawing on what Terranova (2000) calls ‘free labour’. This restructuring of the labour process in turn restructures knowledge: open source software developers maintain collective knowledge through code repositories outside individual institutional control. Their own knowledge about how to use and apply code from these repositories is becoming increasingly significant, destabilizing the in-house software design processes that used to characterize software development (Weber, 2004) and inviting a more heterarchical organizational structure for software production (see Stark, 2001).

As has been outlined here, in software, the circulating process of challenge and recuperation can be traced historically through the relationship between generative forms of hacker culture including free software, and open source licences as they are applied in industrial software production. This process has been less discussed, however, in other areas of open culture and open knowledge production. As the next sections of this article explain, as software and hardware design converge, efforts are being made to develop norms for the production of open hardware design. These efforts reveal a different movement of knowledge and another set of expectations about democracy that are salient for considering the long-term socioeconomic impact of democratic production of media. The next section examines the points of connection and divergence between cultures of software and hardware hacking as a means of identifying further instances of democratic participation through collaborative production.

Small-scale production and the ‘long tail’ of hardware

The major transformations of software markets promulgated by open source software licences were as much a result of the non-material quality of software as they were of the actions of peer-production communities of practice. The creation of repositories of software code do indeed create commons of knowledge that can be accessed by anyone – but the fact that the digital information contained in these repositories can be duplicated perfectly at vanishingly low cost is also important. In this context, free and open source software licences that specify free access either to parts or to whole sets of source code create an abundance of free and open code. Using free or open source software code neither consumes resources on the part of duplicators nor does it lessen the amount of code available for others to use. Because code is notated, records remain of previous modifications, and it is possible to identify which code is open source.

In contrast, and posing significant problems for the promise of expansion of commons knowledge, hardware remains a material product with manufacturing costs. Open hardware advocates note that these costs are dropping as many forms of electronics hardware are becoming software-controlled (Rubow, 2008), as simple electronic switches can now be printed or etched on to various materials, and virtualization software allows some hardware to be controlled by software, but other constraints remain. Material constraints, for example, include the fact that modifying the circuit board of a digital device transforms the device and its function, but does not provide its original designer with any knowledge or information about the nature of that transformation. Open designs – in the form of circuitry diagrams and schema for designing chipsets – are available in repositories such as OpenCores (OpenCores, 2011). However, there are few means of identifying hardware objects as open-source, or of making the labour of hardware hackers visible on the devices they modify. This is turn makes it difficult to find space to challenge the existing regimes of intellectual property. How then can hardware hackers gain access to the equivalent of software source code? Various modes of organization have emerged, but these are mostly ad hoc and rarely develop shared norms of governance such as the free software and open source licences. As a result, there is a variety of ways in which open hardware products engage the market.

Buechley and Hill (2009) describe the Lilypad, an MIT project that developed a flexible electronic chip board adapted from the Arduino and intended for use in the manufacture of electronic textiles. After the end of the MIT project, the chipset became commercially available through the SparkFun manufacturer. This produced a ‘long tail’ of personal consumption that created small but significant markets, supported and expanded by online communities (Buechley and Hill, 2009). In licensing terms, the project designers initially applied a Creative Commons licence to the Lilypad’s design schematics and basic board designs and made them available to anyone who wished to use them, provided that the original designs were acknowledged and the results remained open source. The personal and institutional networks of the MIT researchers were the primary means through which the designs were originally distributed. According to Buechley and Hill, several open source extensions of the chipset have been developed, one of which, a wireless radio connection, has now been integrated into the LilyPad kit and is also widely commercially available. Other extensions made using the open source designs were purely for personal use of designers or artists, with the personal permission of the original designers. This example demonstrates how an academic project can be transformed into a commercial product primarily through the use of an open source licence (in this case, the Creative Commons licence).

Another example of the articulation between open hardware production and the market illustrates the role of existing communication infrastructures and social relationships in effectively establishing frameworks for sharing knowledge about hardware. At the end of 2010, Make magazine sponsored a competition for ‘green’ DIY projects. The winning project, the Global Village Construction Set, proposed ‘an open source, low-cost, high performance technological platform that allows for the easy, DIY fabrication of the 50 different Industrial Machines that it takes to build a sustainable civilization with modern comforts’ (cited in Make magazine, 2010: Your Green). The project maintains the designs and schematics of these large machines on a wiki. The project’s goal is neither mass-scale production of industrial machines nor a proliferation of vastly different industrial machines, but instead a standardization of machine designs allowing a set of pre-determined modifications. Yet the knowledge of the design schematics is expanded primarily through Make magazine’s readership. Make magazine also publishes instructions for how to create basic electronic devices using existing, commercially available open hardware products such as the Arduino micro-controller. These instructions describe how to make an electronic product, but instead of being located in a central repository, they are primarily distributed as part of O’Reilly’s magazine content. Publishing access to design schematics solidifies the importance of the magazine among its core subscription base. Access to open source hardware schematics or designs thus generates value for the emerging DIY culture, without inciting major changes in hardware production processes beyond small-scale DIY and art projects. Again, there is increased participation in hardware production, but ad hoc forms of knowledge exchange are supported by the kinds of commercialized media that have helped to define markets within niche communities of practice.

Hardware freedoms and open source hardware licences

Some open source hardware practitioners argue that it is important to develop standardized means of identifying and collecting open source hardware designs and schematics. Underlying these efforts are expectations that better information about open hardware designs and transparency will lead to a more robust economy of innovation parallel to the one that occurred in open source software. An organization called the Open Hardware and Design Alliance (OHANDA) has begun a project to extend the norms of open licences, developing Hardware Freedoms from the Software Freedoms described above. The project attempts to create a parallel legal structure for hardware designs mimicking that of software code, and is an explicit attempt to apply the normative frames that characterize open source software production to hardware design. The freedoms read as follows:

The efforts to define hardware freedoms in the same terms as software freedoms is evidence of an interest in democratizing hardware production on the model of software, particularly drawing on the ‘viral’ nature of free software licences such as the GPL. As the hardware freedoms themselves suggest, many interlocking layers influence the extent to which hardware is or can be made open, both in terms of its potential design and redesign and in terms of its flexibility of use. The OHANDA project shares the goals of free software advocates in that it seeks to extend the principles of copyleft to hardware design. It also attempts to apply these principles to a formal process that will align the process of knowledge exchange about hardware designs with the process of sharing open source software code.

OHANDA’s proposal is to develop a trademarking system whereby the packaging of a product based on open source hardware will carry a logo indicating that the designs are available online. As a first step, this would require the designer of the hardware to agree to use a ‘copyleft’ licence based on the hardware freedoms outlined above. The designer would then have to register with OHANDA to receive a tracking number for their design. The tracking number would be included on the trademark logo, so that an erstwhile hardware hacker or hardware chip designer could enter it into a computer and find the design in an online repository. This repository would also include information on the designer, the product and the licence. OHANDA provides a flowchart (Figure 1) as a means of explaining their proposed process:

OPEN IN VIEWER

Figure 1.

OHANDA flowchart showing open source hardware relationships

The assumption behind this project is that open hardware production is empowering for producers. The website reads: ‘Empowered by the freedoms, users may develop the product further, register as producer, share his/her design artifacts applied with copyleft license and be connected to the derivatives of the product’ (OHANDA, 2009). More profoundly, the aims of the project are directed at establishing the virtuous cycle of sharing code and unlinking knowledge and labour from the market that encouraged the expansion of open source software production. The main goal, according to one of the OHANDA founders interviewed for this article, is to alter the hardware production ecosystem, lowering barriers to participation in the design of hardware, and potentially also lowering costs for hardware producers, who could use open hardware designs to lower the cost of producing more specialized or bespoke products (Neumann, 2009).

The actual hardware designs being identified with OHANDA, however, do not fit neatly within this virtuous cycle. For instance, one of the first products to apply for an OHANDA licence mark was a line of portable, easy to use cooking-stoves called ‘Good Stoves’, designed by Dr N. Sai Bhaskar Reddy and intended for use in a variety of contexts where resources are limited. The stove designs are available on the internet, but are also available as design schematics on paper. The easily modifiable designs specify easily available materials such as scrap sheet metal. This is an example of open hardware licensing being employed to facilitate knowledge exchange within a development context. Another early licensee was a proposal from the C-Base, a technology and cultural hub in Berlin, for open hardware to be used in a moon exploration project initially targeted at a competition run by Google. The C-Base is an internationally known hub of playful and provocative hacking. Like many of the centre’s projects, the Open Moon project is partly serious and partly frivolous: proponents intend to demonstrate that open source hardware is robust enough for a space voyage, but they wish to produce this hardware within a collaborative, non-hierarchical organization. Tellingly, as of this writing the Open Moon project has withdrawn from the Google contest while continuing with open sourcing hardware. Neither of these OHANDA licensees is primarily motivated by commercial efficiency and market success.

Along with the examples of Make magazines’ machines and MIT’s Lilypad, these OHANDA licensees illustrate the variety of ways that open hardware articulates with the market: in some spheres it might be completely outside the market, for example when hardware designs are shared by artists or scientists as the result of publicly funded work. In others it might create markets like the niche markets for DIY electronics parts aggregated by Make magazine. In still other spheres existing markets could be disrupted, as Neumann implies above. Or perhaps open hardware could address the movement of monopolies of knowledge outside existing markets: for example, repositories of designs for water filtration systems and other useful devices suggest the emergence of ‘Open Hardware for Development’ (OH4D) practices which could parallel similar efforts to develop ‘ICT for development’ (ICT4D).

Open hardware licences as boundary objects

Open hardware licensing schemes have the potential to be enormously challenging to the structures of knowledge that characterize the software and hardware industry. At the same time, distilling cultural norms into licensing codes that are then applied as stand-ins for those complex social structures has some limitations. Partly, this is because licences, as codified norms, have the potential act as what Star (2002; Star and Griesemer, 1989) refers to as boundary objects. Boundary objects are texts or organizing conventions that are used for coordination and alignment between members who work in the same sphere but hold divergent goals and exist in different social worlds. Star and Griesemer’s original (1989) notion of boundary objects refers to their coordinating activities in science, but subsequent work (notably Star, 2002) has explored how boundary objects – most often, standards – operate to coordinate and align other types of work including information systems design.

Locating and analysing the appearance of boundary objects makes it possible to understand how communication occurs across communities, and thus how knowledge moves and develops. Star writes, of boundary objects: ‘they have two important properties: they are loosely structured in common use, and become more tightly bound in particular locations. They are thus both ambiguous and clear, at different moments, for different purposes’ (2002: 118). This kind of contingency of structure seems to characterize open hardware licences. However, at present, the OHANDA process works weakly, at best, as a boundary object: the process of applying a free software-like licence to a design placed in a repository does not address the way other communities of practice might share knowledge (the LilyPad designers making products for their own use or the Make magazine readers reading design schematics for fun). As a result, licences related to open hardware now proliferate: the Open Hardware Roadmap, produced and maintained by American open hardware advocates, lists five different open hardware licences, which do not include the OHANDA hardware freedoms (Open Hardware Roadmap, 2011). Many of the licences mentioned adhere to an Open Source Hardware Definition agreed by the hardware design community in 2010, but the Roadmap’s authors still express concern that small companies producing hardware (mostly for the amateur market) will ‘theoretically not have much legal recourse if their designs get patented, made/sold, or used in a closed system’ (Open Hardware Roadmap, 2011). Licences are unevenly evoked as necessary for sharing common knowledge, but also for protecting individual innovation. Thus, the dialectics still hold: some knowledge is more easily recuperated, for example chipset designs were able to be manufactured by the company that commercialized the MIT project’s Lilypad, but specific knowledge about the designs still had to be communicated by the researchers themselves, through individual social networks. Thus, powerful institutions as well as ad hoc forms of knowledge exchange both configure the ways that knowledge about hardware is transferred.

The struggle to match open hardware’s production process and cultural expansion to the parallel processes that occurred with open source software reveals that the creation of democratic communication spaces is contingent on the structures and cultures of capitalist production, as well as on the more resistant or emancipatory norms encoded in free and open software licences. The expansion of free software production has created its own structural dynamics that have resulted in codified licences facilitating a democratic transformation in software production. To a certain extent this has depended on a culture that views software development not only as labour but also as expression. For members of this culture, open hardware licences act as a way of ensuring the freedom of hardware designs and thus the freedom of the designer to express him- or herself. Yet encoding these cultural norms into open hardware licences also recapitulates the process by which the labour of software programmers has been recuperated by the software production market.

Some parts of the proposed OHANDA process reveal the challenges that opening hardware can pose both to markets and to proprietary and enclosed structures of knowledge. Increased openness based on licences only democratizes certain aspects of the production process. The production of bespoke designs by a small factory is perhaps evidence of a shallow democratization of production in that it broadens the nature and type of participation in hardware design (see Gristock, 2008), but this is not democratic in the same way as the knowledge gained by an artist developing a new idea as the result of public support, and sharing it in her project documentation, or a DIY practitioner wanting to share a pattern if only to receive commentary from a community of practice. Indeed, the shallow democratization of industrial hardware production is already challenged: by re-engineering practices that are essentially piracy including the rise of ‘Shenzen’ mobile phones, which are re-engineered copies of brand-name electronics, is evidence of this oppositional democratization.

If licences and other boundary objects codify norms from particular cultures while superficially engaging with other cultures and institutional actors, aligning values and practices becomes difficult. The dialectics identified in the case of open source software have motivated a codification of open source hardware, even though the products and devices being licensed and marked are determinedly non-commercial. At the same time, the commercial open source hardware kits developed by Arduino and projects like LilyPad are used by artists and educators in several ‘long tails’ of individualized or small-scale production. The various kinds of knowledge being shared suggest that instead of one community of practice employing modes of sharing common to open source practice as developed in software production, there are several different potential locations at which open source hardware projects confront (or fail to confront) markets. In addition to industrial hardware design, which confronts open knowledge as piracy and imitation, these include long-tail markets of amateurs, designs produced as the result of public investment in art or innovation, and open hardware for development.

The initial OHANDA efforts encode the values of a certain group of open source software producers who hypothesize a parallel transformation of hardware production based on their experience and observation of software production. For other social worlds, such as DIY practitioners, global South innovators like the Good Stoves founder, or Shenzen phone manufacturers, the organizational structure and expectations of market success implied by the OHANDA effort could be less meaningful. In this broader conception, flows of knowledge begin to effectively operate in several different directions, breaking down even further the necessary – but suspended – dialectic between marginal or resistant knowledge and the hegemony of neoliberalism and the market.