{"id":7863,"date":"2019-03-22T02:04:38","date_gmt":"2019-03-22T02:04:38","guid":{"rendered":"http:\/\/peerproduction.net\/editsuite\/?page_id=7863"},"modified":"2019-04-03T21:42:48","modified_gmt":"2019-04-03T21:42:48","slug":"open-source-beyond-software","status":"publish","type":"page","link":"http:\/\/peerproduction.net\/editsuite\/issues\/issue-13-open\/peer-reviewed-papers\/open-source-beyond-software\/","title":{"rendered":"Open Source beyond software: Re-invent open design on the Common’s ground"},"content":{"rendered":"
By Kosmas Gavras<\/strong><\/p>\n Download as PDF<\/a><\/p>\n Open source software (OSS) has become a dominant mode of production in a number of areas such as server software, operating systems and scripting languages (Lerner and Tirole, 2005; Chesbrough and Appleyard, 2007). Since the last decade, several studies have focused to a wider applicability (von Krogh and von Hippel, 2006; Nuvolari and Rullani, 2007) of the OSS organizational and production model (von Hippel and von Krogh, 2003; Osterloh and Rota, 2007). Benkler (2002, 2006) draws from OSS and early P2P sharing networks, to describe a third mode of production, the Commons based peer production (CBPP), extending beyond software to open content such as Wikipedia and Openstreetmap among others. Additionally, an equally important direction is the extension of the open source model to the world of tangible objects (Raasch et al., 2009; Balka et al., 2009a, 2009b; Shirky, 2005).<\/p>\n Initially, a significant number of free and OSS theorists have objected to, or at least have been sceptical of, the openness parallelization between bits and atoms (Stallman, 1999; Raymond, 1999a; Maurer and Scotchmer, 2006; Ackermann, 2009). However, in the course of time the open source hardware\u2019s (OSH\u2019s) potential was realized, as among other advantages, the OSH will represent the only possibility to run freely OSS in the near future (Stallman, 2015). In the mean time a lot of major OSH projects have emerged. Indicatively, some of the most prominent approaches are located in the fields of: Electronics (Arduino), Mechanical (Farmhack, Ateliers Paysan, Open Source Ecology), Mechatronics (RepRap, OpenBionics), Non electronic – nor mechanical (Wikihouse, Opendesk, Openstructures, Open Architecture Network, Hexayurt).<\/p>\n Von Hippel makes a parallelization between OSS and open design (OD) as the immaterial phase of OSH: ‘Hardware is becoming much more like software up to the point you actually fabricate an object.’<\/em> (Von Hippel in Thompson, 2008).<\/p>\n Despite most of the OSH\u2019s theoretical background was transcribed from OSS\u2019s principles and licenses, the main theoretical stake persists. What is the ‘source code’ of OSH? The profound answer that design is the ‘source code’ of hardware is not solid and specific enough to establish an analytical framework (Vardouli and Buechley, 2014; Fuller and Haque, 2008). The purpose of the research is to provide a critical insight to existing OSH and OD theoretical definitions and describe potential new dimensions based on the following starting points:<\/p>\n In order to describe the ‘source code’ of OSH, we will analyze a set of case studies against existing OSH definitions and in relation with the above-mentioned points. Specifically, we will highlight inefficiencies or inconsistencies of existing OSH theory in the given context and suggest ways to overcome the recognized limitations. The article is structured as follows: In the next section we describe the existing literature context that operationalizes the research framework. The stimulated research questions as well as the research approach and methodology are developed in the third section. Successively, the fourth section contains the case study analysis. The fifth section discusses the resulting limitations and outlines an emergent OSH and OD direction based on the instrumental potential of the case study research. The final section provides a conclusion of research results along with directions for future elaboration.<\/p>\n The term OD does not imply a single and cohesive meaning (De Mul, 2011; Vardouli and Buechley, 2014). In theory and practice it is often used interchangeably with different but frequently complementary meanings, such as the collaborative, cooperative or participatory design (Habraken, 1972; Kaspori, 2003; Salingaros et al., 2010; Manzini, 2015; Ratti and Claudel, 2015), the modular design (Fuller and Haque, 2008; Kostakis and Papachristou, 2014a), or the freely and globally shared design (Open Architecture Network, no date; Open Architecture License, 2016). Actually, all of those meanings represent major dimensions of an evolving OD definition in the framework of digital Commons and digital fabrication.<\/p>\n Analytically, the Design Global, Manufacture Local (DGML) model (Kostakis et al., 2015) highlights a new productive model with transformative socio-economic implications that is based on the convergence of globally shared OD Commons with desktop manufacturing. The voxel-based fabrication describes a modular design methodology along with an experimental manufacturing technique, which is based on the discourse between digital and analog materiality (Gershenfeld 2007, 2012; Gershenfeld et al., 2017; Kostakis and Papachristou, 2014a; Hiller and Lipson, 2009). At last, participatory design has a history that precedes the domination of the open source model. During the decade of 1970s the approaches of Habraken (1972), Friedman (1975) and Alexander et al. (1977) targeted to the user empowerment in the design process. Despite the aim for increasing user participation, there was always a common and clear separation line between what Manzini (2015) defines as expert and diffuse design. In the case of Habraken (1972) it was the distinction between the design of structure and infill that reflects the bipole of expert and user. Alexander et al. (1977) and Friedman (1975) constructed a pattern language and a computational design process respectively, as non-technical vocabularies for user participation. In general terms, the principle is common: the expert designs an almost closed system that the user interacts with. On the other hand, approaches that draw from the open source model (Kaspori, 2003; Salingaros et al., 2010) describe emphatically the demise of the expert design in the context of what Von Hippel (2017) coined as collaborative free innovation.<\/p>\n Despite the existence of wide literature on various partial aspects of OSH and OD, there are very few attempts to provide an integrated definition of openness in the sphere of hardware. In general, the most systematic and comprehensive quantitative method up to date regarding hardware\u2019s openness evaluation is the open-o-meter (OPEN!, 2018). Open-o-meter is a numeric scale, consisting of eight equivalent Boolean values which cumulatively express the total openness of OSH. Analytical documentation of the openness scale (Table 1) is presented by Bonvoisin et al. (2017). While the openness levels refer to the hardware, the evaluation criteria examine exclusively aspects of immaterial documentation or OD content. In this way open-o-meter is, beyond an evaluation tool, the theoretical foundation of OD as the open hardware\u2019s ‘source code’.<\/p>\n <\/a><\/p>\n Table 1. Open-o-meter. OSH product characterization criteria (Bonvoisin et al., 2017). [Click image for larger view.]<\/strong><\/p>\n In order to develop a critical stance on existing theoretical framework, it is important to trace open-o-meter history back (Table 2). Open-o-meter is a further specification of theoretical freedom principles (study<\/em>, make<\/em>, modify<\/em>, distribute<\/em>), and forms (transparency<\/em>, replicability<\/em>, accessibility<\/em>) of openness, referenced in Open Source Hardware Association\u2019s Statement of Principles 1.0 (2011) and Balka et al. (2010, 2014) respectively. Successively, Open Source Hardware Association\u2019s principles are based to the Open Source Definition (Open Source Initiative [OSI], 2007).<\/p>\n <\/a><\/p>\n Table 2. OSH theoretical framework as an evolution from OSS definitions. [Click image for larger view.]<\/strong><\/p>\n (abbreviations: Open Source Initiative [OSI])<\/p>\n Consequently, open-o-meter framework leans towards OSS rather than free and open source software\u2019s (F\/OSS\u2019s) political standpoint. As a proof of this tendency, open-o-meter defines commercial usability of design documentation and physical product, as one of the basic parameters of openness. In fact, there is a profound contradiction between commercial usability and the notion of openness, which is more evident at the immaterial side of the design-construction continuum. How can design documentation be commercially exploitable and freely editable and available concurrently? Some of the most known examples of F\/OSS (GNU\/Linux, Apache server, Mozilla Firefox) thrive in the absence of commercial appropriation. Additionally, it is proved that total openness, as expressed by the sum of transparency, replicability and accessibility excluding commercial usability criterion, increases the level of contribution (Balka et al., 2014) even if some forms of openness play a greater role than others.<\/p>\n At this point, it should be stated that from an ideological point of view the author stands with the position of the Free Software Foundation (FSF), rather than that of OSI, regarding the social imperative of software (FSF, 2016). Since at the technical level, free software source code qualifies as open source code (Stallman, 2016) and vice versa, it could be supported that open source code is just one -the technical one- of the constituents of F\/OSS definition. In this sense, hereby the term open source code, OSS or open source design does not necessarily point to the OSI definition, unless otherwise explicitly stated. Additionally, taking into consideration the framework of this paper, OD is a commonly acceptable term, comparing with the limited usage of the term free design.<\/p>\n To get back on track, previous research works employing a quantitative assessment of hardware\u2019s openness have covered the field of electronic products (Balka et al., 2010, 2014), as well as the field of non-electronic but mechanical hardware (Bonvoisin et al., 2017). To the author\u2019s knowledge, there is no published quantitative study evaluating openness of the OSH subset that is neither electronic nor mechanical. This absence is explained by the initial lack of OSH practices being neither electronic nor mechanical. Due to the proximity of electronic hardware with OSS movement, the earlier OSH initiatives were concerning electronic devices (Gibb, 2014) or later mechanical (Vallance et al., 2001). It is characteristic that TAPR (2007) – one of the first open hardware licenses – had considered only electrical or mechanical artifacts as potentially open hardware.<\/p>\n To synopsize, the existing theoretical background of OSH is transcribed from OSS definitions and licensing schemes which are not structural properties of the source code. Even if the license profoundly affects to some extent the subsequent form and development of the code, it is not an inherent property of the code itself.<\/p>\n If the licensing schemes represent exogenous properties of open source code, which are the structural properties? According to Conway (1968:31), ‘organizations which design systems are constrained to produce designs which are copies of the communication structures of these organizations’<\/em>. In other words, the structural properties of code or design mirror the structure of the organization that developed them. MacCormack et al. (2006, 2012) proved empirically the ‘mirroring hypothesis’ in the field of software industry, highlighting source code\u2019s modularity. Modularity is defined as the decomposition of a product or design to independent modules. Analytically, the more distributed, informal and open organizations tend to produce the more modular codebases. On the contrary, the more centralized, hierarchical and closed organizations produce more monolithic software. As a summarizing outcome, OSS is structurally different from proprietary software in terms of modularity (MacCormack et al., 2006). The difference in modularity between the opposite ends is surprisingly large, reaching a factor of eight between software of similar size and function (MacCormack et al., 2012).<\/p>\n Therefore, there is a causal bidirectional mechanism between, on the one hand, the dispersed geographical distribution of open source communities, the open contribution spirit and the informal authority, and on the other hand, the emergent modular code\u2019s structure (MacCormack et al., 2012). The causal linkage is bidirectional as ex ante<\/em> modular code\u2019s structure is prerequisite for attracting and facilitating a diverse and distributed community of developers which in turn enhance the overall modular effect in the course of time (Baldwin and Clark, 2006; MacCormack et al., 2012). Preliminary design choices are proved critical regarding the potential successful evolution of OSS projects. Baldwin and Clark (2002) explicated further the importance of modularity as a structural constituent of the open source model:<\/p>\n To summarize, modularity is a fundamental property with causal role at the open source model that can not be omitted when transferring the open source methodology to other fields, including OSH. As a conclusion, the freedoms to study<\/em> and modify<\/em> depend to a great extent in code\u2019s modularity.<\/p>\n While modularity is an important parameter of an evolving OD definition, we realize that it is meaningless to examine the OSH solely as an OSS offspring, but only in a wider perspective including the technological advancements related with hardware’s material nature. For instance, the digital fabrication revolution makes feasible the distributed – even personal \u2013 manufacturing of a genealogy of customized objects, virtually at the same unit cost as if producing identical copies (Rifkin, 2014).\u00a0According to Carson (2010), manufacturing is fading as a centralized, closed and capital-intensive activity.\u00a0As a proof of concept, the digital fabrication revolution and hackerculture have already led to the emergence of a rapidly increasing global network of local makerspaces (Niaros et al., 2017), answering to the user\u2019s need for custom products (Von Hippel, 2005) among others. But manufacturing (being the second part in the design-construction continuum) depends to a great extent on the design input.<\/p>\n What kind of design will form the necessary data input to a localized, on-demand, and customized fabrication? A monolithic, closed, centrally produced design aiming to cover the greater homogeneous market segment of user needs, or a modular, open, globally produced design able to be modified at low cost to fit the local needs? Lakhani describes a user-powered design innovation, proactively forcing companies out of the product design space (2007). Consequently, the real potential of digital fabrication will be actualized more effectively in combination with the peer production, in proposing an alternative, holistic emergent productive model (Kostakis et al., 2013, 2015), rather than in re-organizing existing corporate infrastructure (Hermann et al., 2016) according to design principles that are primarily native to open source culture.<\/p>\n To sum up, modularity, beyond being a structural property of the open source mode of development, it is also related with the freedom to make<\/em> in the socio-technological framework of digital fabrication.<\/p>\n Current OSH theoretical framework have tested and been tested against electronic products (Balka et al., 2010, 2014) and mechanical devices (Bonvoisin et al., 2017), but to the author\u2019s knowledge there is not any published study concerning physical products being neither electronic nor mechanical. There is preliminary evidence that current OSH theory can not migrate free ride from one kind of hardware to another. Both electronic and mechanical devices usually include stardardized, ready-made and commercially available components to a great extent. Off-the-shelf components can limit the design and customization potential which are inherent to the concept of user-developed hardware and digital fabrication. On the other hand, non electronic or mechanical hardware, ranging from a piece of furniture to a house, is characterized by higher degree of customization. Additionally, electronic circuits, machines and other hardware do not share a common design language. Electronic design is comprised of 2d layouts; mechanical design requires a kinetic 4d approach; whereas other hardware can be thoroughly described by a 3d model.<\/p>\n Bonvoisin et al. (2017) acknowledge that the broader applicability of open-o-meter is an open question. Beyond breadth, according to Bonvoisin et al. (2017) further research is required to provide an in depth qualitative specification of each openness criterion. The authors of open-o-meter followed a simplistic binary evaluation for each criterion that was practically imposed by the substantial sample size required for statistical analysis. Based on the literature review and the scope of the research, the following questions will be addressed (Figure 1):<\/p>\n RQ1a – theory\u2019s breadth: Is open-o-meter a valid quantitative tool and definition in the framework of neither electronic nor mechanical hardware?<\/p>\n RQ1b – theory\u2019s depth: If open-o-meter is not generically applicable, then which criteria fail to capture actual openness levels of the examined subset? How can these criteria be improved by further qualitative specification?<\/p>\n RQ2 – theory\u2019s context: How is modularity, which has not been expressed directly in open-o-meter, reflected in OSH practice? How can we describe a positive feedback loop between OSH\u2019s practice and theory, based on a new reading of openness in the context of digital fabrication and OSS\u2019s structure?<\/p>\n <\/a><\/em><\/p>\n Figure 1. Research questions placement in the conceptual timeline diagram framed by the literature review. [Click image for larger view.]<\/strong><\/p>\n (image processing: author, image source:author, abbreviations: Free and Open Source Software [FOSS], Open Source Software [OSS], Free Software Foundation [FSF], Open Source Hardware [OSH])<\/p>\n In order to address the research questions mentioned above, we employed an empirical quantitative and qualitative case study research.<\/p>\n Analytically, the hybrid quantitative-qualitative method refers to the first question (RQ1a) regarding open-o-meter theory\u2019s breadth, while following research questions will be tackled with exclusively qualitative approaches. Theory\u2019s breadth will be evaluated by the assessment of how quantified openness levels of the examined subset, actually correspond in practice to the more abstract and qualitatively approached freedoms of study<\/em>, make<\/em> and modify<\/em> (Table 2). The quantitative method of openess\u2019 assessment is the open-o-meter (Table 1), as documented by Bonvoisin et al. (2017). Despite the fact that the original openness scale utilizes only binary value rating, the current research adopts the use of midpoint assessment, as a general measure for those cases where a variable pattern is discerned. Furthermore, the open-o-meter method followed in this article refers solely to part 1 of the original method, as the external and secondary criteria consisting the part 2 table, are mostly constrained by the case study selection process.<\/p>\n Empirical data required for the quantitative assessment and the qualitative inquiry are acquired from web-based documentation and databases. Indicatively, sources that have been used depending on each occasion are the following: websites of the product development communities, archived webpages, web-based repositories, github or other party development statistics, processing of CAD or other files, participation in developers forums and communication networks, and member initiation questionnaires among others. It is important to stress that data input to the quantitative assessment, is based on the ‘macroscopically’ observed presense or absence (or variable pattern) of certain documentation as described by open-o-meter methodology, avoiding any qualitative characterization of the usefulness of each provided document. Data analysis has followed a combination of categorical aggregation and direct interpretation strategies, as described by Stake (1995).<\/p>\n The choice of case study research is based on the need for detailed examination of particular cases and their context. In this way, the shortlisted cases are intentionally not representative of the whole range of the OSH scape, but they are examined as collective – instrumental case studies (Stake, 1995), able to offer a new understanding of OD in the OSH framework. As the selection of cases is of critical importance in any type of case study research (Stake, 2003), the formal selection process and criteria actually reflect the author\u2019s strategy in the given context. Moreover, it is important to note that the aim of the collective quantitative case study research is not to concentrate on a comparative analysis per se, but to illustrate potentially different viewpoints to the research questions.<\/p>\n The selection[1]<\/a> of case studies is based on an already existing online directory of OSH projects, maintained by Wikipedia (2018a) which is a prominent example of CBPP beyond software. Previous studies have employed either conventional internet search, using a snowball effect approach that was prone to be author biased or online directories built-up and biased by OSH projects\u2019 ‘owners’. On the other hand, the selection of a Wikipedia directory ensures a minimum consensus about what is considered by the greater community as open source. The Wikipedia online directory is a dynamic, bottom-up archive of the OSH projects, the launch of which mostly coincides with the recent history of the OSH development[2]<\/a>. The following selection criteria were applied to the initial list, in order to narrow the corpus of research, according to the research questions:<\/p>\n The OSH communities that satisfy the imposed criteria are: the Wikihouse (2018a), the Opendesk (2018) and the Openstructures (2018a). As denoted by their names, Wikihouse and Opendesk produce house and furniture designs respectively. Openstructures\u2019 range starts from furniture and household equipment, ending potentially to houses or even bigger structures.<\/p>\n Opendesk is a for-profit company, supporting an online platform that connects furniture designers with customers and local makers all over the world. It was founded in 2014, and part of the founders is part of the Wikihouse creators. Opendesk practice largely depends on the DGML model, as it is by definition a global platform for local making. Moreover, it is based completely on CNC manufacturing, as all designs are assembled from flat cut wooden profiles. Most of the furniture overpasses the assigned minimum complexity limit but features significantly less parts than other OSH projects, comprising from a few elements to some dozens of elements. Profit making comes from charging 30% of the manufacturing cost as a transaction fee every time a customer orders a piece of furniture. Apart from the transaction fee, there is a provision for a design fee which is calculated at 8% of the manufacturing cost. In any case, all of the furniture designs are freely available for personal fabrication and usage.<\/p>\n Opendesk project collects 4.5 out of 8 points in the openness scale (Table 3). Assembly instructions and bill of materials (BOM) are provided in non-editable formats, whereas contribution guide is a rather vague and closed process. Furthermore, licensing does not follow a uniform pattern, featuring a variety from copyleft to licenses allowing commercial appropriation. As the object of design is considerably smaller comparing to other OSH projects, the shortage of editable supporting documentation (assembly guide, BOM) is rendered less important than the absence of an open contribution process. The top-down enforced modus operandi of Opendesk is based on a standard and fixed relation between designer(s) and object of design, not formally allowing any collective optimization process which usually is very common to every CBPP and OSS project. Actually, modifications are theoretically feasible but for personal use only, as there is not any open platform or any even informal process for forking and then merging back into master branch. To conclude, Opendesk\u2019s one way workflow is almost identical to the conventional non-participatory design process, being characterized by a small scale closed design team or individual designer, retaining a practical ownership (above formal licensing) over artifact by indirectly controlling the potential modifications. Apparently, this is related to the nature of the for-profit organization and it seriously decreases the project\u2019s overall openness. It seems not rational to rate open contribution potential equally with the existence of editable BOM or other secondary documentation. Is it open a project that just freely reveals the ‘source code’ of the hardware but restricts users and other designers from contributing?<\/p>\n <\/a><\/p>\n Table 3. Openness evaluation. [Click image for larger view.]<\/strong><\/p>\n It could be supported that despite drawbacks, Opendesk still offers what is considered by open-o-meter the ‘source code’ (editable CAD files and non-editable BOM and assembly instructions) which seems almost enough to study<\/em>, make <\/em>and modify<\/em> the design individually. In contrast, a detailed examination of the shared ‘source code’ will reveal that it is indirectly but efficiently restricting users and designers from altering the shared furniture designs even for personal use. The criterion regarding editable CAD files (Table 1) in practice seems a rather generic and abstract scheme. The limitation of the evaluation criterion does not lie only in the quality of CAD files, as suggested by Bonvoisin et al. (2017), but to the nature of the CAD files. To further the argument, it is required to make a clear distinction between design and fabrication CAD files. The actual shared ‘source code’ of Opendesk is flat-cut drawings, intended for CNC manufacturing, which are provided in editable format. Fabrication files, even in native file format, are still a derivative of actual design files. It is self-evident that cut-out drawings (Figure 2) by themselves can not help neither to study<\/em> nor to modify<\/em> the design, but only to make<\/em> it. There is no doubt that if personal resources were limitless, an expert designer could use the fabrication outline drawings and the assembly instructions to reverse-engineer the design files, but this can not be the case. Another important note, beyond the analytic distinction between design and fabrication CAD files, is that the cognitive process producing the latter from the former is of utmost importance regarding the OSH development, and definitely subject of openness evaluation. To conclude, community is not only restricted from providing optimization feedback, but it is generally constrained from studying<\/em> and modifying<\/em>.<\/p>\n <\/a><\/p>\n Figure 2. Opendesk shared CAD files in relation to the design ‘source code’. [Click image for larger view.]<\/strong><\/p>\n (image processing: author, image source: https:\/\/www.opendesk.cc\/<\/a>, last visit: 01\/07\/2018)<\/p>\n Even the freedom to make<\/em> should not be considered as a totally independent principle from studying<\/em> and modifying<\/em> an object. Specifically, the three principles are interconnected to a great extent in a synergistic manner as the full potential of making<\/em> is achieved only if you are first of all able to study<\/em> and modify<\/em>. Otherwise, the DGML concept will be deducted to a simple model of distributed ‘mass production’, negatively affecting the real potential of OD and digital fabrication. It is less meaningful and underutilized to use a CNC or a 3d printer just as a medium to manufacture locally identical copies of what is used to be massively designed and produced.<\/p>\n Openstructures began as a student project at Institute without Boundaries in 2006. In September of 2009, Thomas Lommee – designer and former member of the student team – organized an exhibition, showcasing the concept and some initial prototypes, as well as an open call for the collaborative development of the project. Openstructures is defined as an open modular system for building hardware – in contrast with current closed (proprietary) modular systems – which is inspired by the modularity of OSS (Openstructures, 2018b). The centerpiece of Openstructures system is OS grid, an openly shared geometrical grid, built up from the supersposition of a diagonal, polar and regular grid (Figure 3). The OS grid serves as the common compatibility base to design parts and their interfaces which then can be assembled in different combinations to form anything from furniture to houses. Another important design principle is the vertical organization of artifacts in parts<\/em>, components<\/em> and structures<\/em> depending on the functional autonomy and relative position of each element in a posssible greater scheme. Specifically, components<\/em> are made of parts<\/em> and respectively structures<\/em> are made of components<\/em> in a recursive manner.<\/p>\n <\/a><\/p>\n Figure 3. Openstructures modular grid made out of 4X4 cm square bold-lined cells. [Click image for larger view.]<\/strong><\/p>\n (source: http:\/\/beta.openstructures.net\/block_images\/0000\/0242\/grid4.jpg?1268343038, last visit: 01\/07\/2018)<\/p>\n ‘The open modular system has the potential to:<\/em><\/p>\n In contrast with the declared intentions, Openstructures is the less open project (Table 3) in the open-o-meter scale as the contribution guide is the only steadily provided documentation. Basic documentation, as CAD files whether editable or not, is provided occasionally. Supporting documentation, as BOM or assembly guide, is totally absent and licensing does not follow a specific pattern. From a first point of view, the freedoms to study<\/em>, make<\/em> and modify<\/em> are only occasionally exercised.<\/p>\n Despite the lack of consistently open documentation, according to project\u2019s initiators, community is encouraged to participate either one or more of the following ways:<\/p>\n How can it be possible to contribute to an open invitation when there is not steadily provided any hardware\u2019s ‘source code’? A related important observation, regarding Openstructures\u2019 online platform, is that technically each contributor is permitted to modify<\/em> only those elements that he has authored. Consequently, even in these cases that the design files are available and editable, the development is characterized by a conventional, individual and authoritative design workflow. On the other hand, in the CBPP framework, the collective optimization of existing knowledge modules is equally or even more important than making new modules. Digital Commons – with OSS being an eminent example – are structured more like work in progress (Raymond, 1999b), continuously optimizing according to new inputs, rather than as finished products. As a result, an OD platform without hierachical control of what – new design – is published, while being in the correct direction, it is not a per se sufficient condition to trigger the wealth of open source collaborative practices.<\/p>\n Beyond the actual technical and collaborative limitations of the platform, the Openstructures\u2019 most fundamental characteristic is the modular design development. Even if it is not explicitly mentioned, it seems that modular design in combination with the imposed classification to parts<\/em>, components<\/em> and structures<\/em>, was indirectly a method of asynchronous collaboration and optimization. For instance, the process of re-using a part<\/em> (module) from another designer to a new assembly turns the new assembly into a collaboratively produced object. In fact, a closer look to the repository reveals that most structures<\/em> are composed of parts<\/em> and\/or components<\/em> designed by the same author, practically eliminating the potentially emergent collaborative effect.\u00a0Why does the design modularity not lead to re-use cycles in a collaborative innovation environment as declared?<\/p>\n Apart from the collaborative limitations imposed by the online platform and the cases of absolute documentation scarcity, the reason lies in the essence of modularity itself. What is different, when comparing the OS grid system with the voxel-based design described by 3d printing pioneers (Hiller and Lipson, 2009; Gershenfeld, 2007, 2012), is the interface design. The potentially innumerable joints that can be designed, based on the OS grid, render any interoperability between modules unfeasible. On top of that, the hierarchical division of elements (parts<\/em>, components<\/em>, structures<\/em>) from an individual design perspective leads to an a posteriori modularization of a structure to parts<\/em> and components<\/em>, in an analytic rather than a synthetic manner, with limited re-usability.<\/p>\n In regard to the freedom to make<\/em>, even if design files are occasionally shared, the fabrication files are never shared. This fact renders almost impossible for potential users to fabricate complex structures with little effort, efficiently precluding them from a feedback process. On the other hand, beta-testers and generally users are extremely valuable for digital Commons comparing even to developers base (Shirky, 2008). Raymond (1999b) strengthens the argument:<\/p>\n ‘# 6 Treating your users as co-developers is your least-hassle route to rapid code improvement and effective debugging. <\/em><\/p>\n The power of this effect is easy to underestimate. In fact, pretty well all of us in the open-source world drastically underestimated how well it would scale up with number of users and against system complexity, until Linus Torvalds showed us differently.’<\/em><\/p><\/blockquote>\n Fabrication files\u2019 contribution is not limited to the freedom of making<\/em>. Moreover, fabrication drawings are not derived from design files in a deterministic way as each specific design can be materialised in more than one way. To sum up, as the production of OSH does not culminate in the design documentation, design and fabrication files as well as the intermediary process are subject to study<\/em> and modify<\/em> beyond merely making<\/em>.<\/p>\n Wikihouse was initiated by a non-profit foundation in 2011 (Wikipedia, 2018b), claimed to be ‘an open source project that re-invents the way we make homes<\/em>‘ (Wikihouse, 2018b), based on distributed digital manufacturing. Since then, architects, builders and users had constructed pilot wikihouses all over the world, inspired from the basic Wikihouse prototype called microhouse<\/em>. Microhouse<\/em> ensures by definition an advanced level of complexity far from the imposed cut-off limit. According to its declaration, Wikihouse draws a clear parallelization between OSS and architectural design as ‘digital design allows every home to be designed as code; instantly customized to its site and user<\/em>‘ (2018a). Moreover, it is self-defined as a collaborative project: ‘by everyone for everyone<\/em>‘ (Wikihouse, 2018a), following the DGML principle: ‘share global, manufacture local<\/em>‘ (Wikihouse, 2018b).<\/p>\n Wikihouse\u2019s design knowledge is organized from smaller to larger elements in: tools<\/em>, technologies<\/em> and types<\/em>. Types<\/em> are ready-made building designs, hosting technologies<\/em> as add-on systems which are assembled using tools<\/em>. Microhouse<\/em> is a one bed house design (Figure 4), currently being the only shared type<\/em>. Available technologies<\/em> include WREN<\/em> chassis system and OWL<\/em> internal door kit. The latter along with tools <\/em>(mallet and a step unit) are components of limited interest and we will not analyze them further. On the other hand, WREN<\/em> is an integral system for developing and sharing new Wikihouse types<\/em> or modifying existing ones. Its main purpose is to automate fabrication analysis, subdivision and indexing required for the construction of 3-dimensional structural frame out of flat cut panels milled in CNC machines (Figure 5). Taking into consideration the distinction between design and fabrication CAD files as highlighted in the previous cases, WREN<\/em> could be characterized as the intermediary process going from design to fabrication files. Openness levels will be assessed separately for type<\/em> and technologies<\/em> as they do not share the same documentation. Digital repositories of type<\/em> and technologies<\/em> are maintained at Github[3]<\/a>.<\/p>\n <\/a><\/p>\n Figure 4. Microhouse type. [Click image for larger view.]<\/strong><\/p>\n (source: https:\/\/github.com\/wikihouseproject\/Microhouse\/blob\/master\/microhouse_0.5_isoFull.jpg<\/a>, https:\/\/github.com\/wikihouseproject\/Microhouse\/blob\/master\/microhouse_0.5_iso.jpg<\/a>, last visit: 01\/07\/2018)<\/p>\n <\/a><\/p>\n Figure 5. WREN technology. [Click image for larger view.]<\/strong> (source: https:\/\/github.com\/wikihouseproject\/Wren\/raw\/master\/Images\/Connectors-01.png<\/a>, last visit: 01\/07\/2018)<\/p>\n The Microhouse<\/em> type<\/em> is evaluated as the most advanced project in the openness scale (Table 3). According to open-o-meter microhouse<\/em> scores 7\/8 points, the only drawback being the lack of assembly instructions in editable format which seems a minor deficiency comparing to the other criteria. In other words, if open-o-meter was a thorough and reliable method for the assessment of openness, then Wikihouse should be an almost perfectly open instance in-line with the principles of OSS development. Contrary to the expectations, microhouse<\/em> development has considerable structural differences from the well-known OSS examples. Despite the fulfilment of participation guidelines criterion, the development of microhouse type<\/em> remains largely at the control of a closed team mostly coinciding with Wikihouse founders. The microhouse<\/em> repository at github[4]<\/a> has 42 commits from only 3 unique contributors. The initial commit was made on the 25th<\/sup> of August 2016 and thereafter most of the commits refer to supplementary documentation rather than design files. In other words, the ‘source code’ was uploaded to repository when the project was mature and most of the core files were already prepared by the founders. Since then, only minor edits have occurred. The above-mentioned characteristics are very common to the early stages of OSS development, before migrating to an open distributed and global mode of production. On the other hand, Wikihouse is a seven years old effort that could not be considered as an immature initiative. Consequently, a published contribution guide along with editable CAD files, BOM and assembly manual may not be enough to promote accessibility<\/em> and the freedom to collaboratively modify<\/em> an existing design.<\/p>\n The role of modularity in ‘voluntary large scale parallel processing<\/em>‘ (Weber, 2004: 88) of a project is already described in the literature review as one dimension of its influence among others (Baldwin and Clark, 2002). If CAD files with the supporting documentation constitute the ‘source code’ of OSH, then the knowledge contained in CAD files should be organized and built in a modularized way. GNU\/Linux, Wikipedia and other known Commons could not have been developed to that magnitude, if they were not synthesized as an assemblage (DeLanda, 2006) of interconnected and semi-independent modules. On the contrary, CAD platforms are shaped to fit a much different mode of production, ranging from the sole designer to medium-sized dispersed but closed interdisciplinary teams. The first mode usually corresponds to an unstructured aggregation of 2-dimensional geometric data, while the second to a hierarchical relational model made out of preset 3-dimensional entities, which is known as Building Information Modelling<\/em>. Whilst the uploaded design files of microhouse<\/em> belong to the former data organizational structure, both are incompatible with an open source modular mode of development.<\/p>\n Beyond the collaborative role of modularity, another dimension is its tolerance of uncertainty. The case of Wikihouse, more than the other cases, highlights the need for adjustable design according to local or individual needs, faced in a globalized DGML methodology. It is oversimplistic to consider as realistic case the widespread application of OSH standard housing types<\/em> all over the world, regardless how the number of types<\/em>. Consequently, it is required to incorporate differentiation parameters in type<\/em> design that will make easier the engagement of the community and the application of each type<\/em> to different site, climate and materials among others. At this framework, it seems less arduous to design parametric housing types<\/em>, than creating a new solid type<\/em> for almost each one local instance. In the former case is required to design the change instead of the final object\u2019s properties. The global gestion leads on to a relational or procedural object definition rather than a categorical finished and unique artifact. It could not be overlooked that Wikipedia entries as well as OSS modules are defined in relation to other entries or modules resembling a distributed flexible network organization. Moreover, the design and execution of any OSS is procedural in its essence or in other words algorithmic. To summarize microhouse<\/em> despite the nearly full rating in the openness scale, it is qualitatively less collaborative and customizable than expected in the placed framework, which can be attributed to the absence of design modularity.<\/p>\n WREN<\/em> and OWL<\/em> technologies<\/em> score 6\/8 (Table 3), as besides non-editable assembly instructions they lack license allowing commercial usage. Despite fundamentally disagreeing with rating commercial usability as an openness parameter, it is noted that microhouse<\/em> is released under CC BY-SA 3.0 while WREN<\/em> system under the copyleft MPL 2.0. But, WREN<\/em> and microhouse<\/em> share a more fundamental difference. WREN <\/em>system is not designed in conventional CAD platforms, but it is created in a visual programming environment (Figure 6) which functions as an extension of a proprietary design platform. Visual programming as well as classic programming (scripting) is in-line with a relational or procedural definition of the design, being potentially more modular than conventional CAD if used as core of the design methodology. Consequently, CAD files are not necessarily the only design files, which was an axiomatic admission for open-o-meter evaluation framework.<\/p>\n Furthermore, it is interesting that both type<\/em> and technologies<\/em> have been accomplished and released under a proprietary and commercial design platform. It is generally accepted that OSS design platforms are not so common, and they have not approached yet the functionality of commercial platforms, possibly with the exception of Blender (Velkova, 2016) which is addressed more to visualization rather than design industry. Nonetheless, this is a highly restrictive fact regarding contribution potential, as it automatically excludes a great amount of the community. While design platform\u2019s openness is not part of open-o-meter evaluation, it certainly and severely affects the ability of a project to be characterized as open. To reformulate open-o-meter criteria regarding CAD files, they need not only to be in native file format but in open file format as well.<\/p>\nINTRODUCTION<\/h2>\n
\n
LITERATURE REVIEW<\/h2>\n
The multiple dimensions of OD<\/h3>\n
From OSS to OSH definition: the case of open-o-meter<\/h3>\n
OSS\u2019s structural properties: the parameter of modularity<\/h3>\n
\n
Digital fabrication from a design perspective<\/h3>\n
RESEARCH DESIGN<\/h2>\n
Research Questions<\/h3>\n
Research Approach and Methodology<\/h3>\n
\n
ASSESSMENT OF PRACTICES<\/h2>\n
Opendesk<\/h3>\n
Openstructures<\/h3>\n
\n
\n
Wikihouse<\/h3>\n