3 Literature Review
3.2 After Design: Infrastructuring, Appropriation And Tailorability
3.2.4 Meta-Design and End User Development
The proposal that end user development may be a key route to sustainability, and that it might be seen in the context of 'infrastructure' is also explored at length in a discussion related to PD under the banner of meta-design. Meta-design as a whole is a diverse field that looks for methodologies for wider collaboration and the opening up of new forms of creativity. Metadesign looks to expanding the “scope of design” and design practice to “cope with the complexity of natural human interaction made tangible by technology” (Giaccardi 2005 p.344).
Literature in meta-design shows strong connections to the early days of interactive arts in the 1980s with its utopian ambitions for creative user involvement and continues to encompass theorising forms of user or audience participation in the cultural and creative field. Meanwhile, within the field of computer science, the banner of meta-design has been used by some writers, notably Gerhard Fischer, as a way of further highlighting the need for tailorability after design.
Meta-design, like PD, emphasises the need for co-creation to enable the production of effective tools. Within the meta-design perspective, co-creation is wholly facilitated through the design of technological tools for tailorability. It is through such tools that users input creatively and the blurring of users' and designers' roles take place. Fisher says:
“A necessary, although not sufficient condition for meta-design is that software systems include advanced features permitting users in creating complex customizations and extensions”.
Meta design is therefore a more technologically led approach than PD although there is also a recognition that various forms of human centred learning is required to fully enable users to customise. Meta-design also lacks the concern for ethics and values found in PD. However, the strength of the meta-design perspective is its focus on sustaining of “emergence, evolution and adaptation” (Giaccardi 2005 p.13). Echoing the distinction by Ehn and others between 'design after design' or 'design in use', Fischer distinguishes 'use time' from 'design time'. For Fischer, it is empowering the user during use time that is most important (Fischer 2003). During design time, rather than attempting to use participation to get as close as possible to a finished design, the emphasis of effort should be on creating incomplete designs. These incomplete designs should be complemented by the development of tools that are explicitly created to allow the user to engage creatively with the system and perform necessary steps for completion and ongoing adaptation to changing needs. (Giaccardi & Fischer 2008)
The work by Wulf and Pipek and also Fisher, is based on the extremely important observation that adaptation is always required throughout the use of a system, and that fact should not be resisted or seen as a flaw in initial design. While both note the importance of tailorability and the potential value of appropriation, meta-design puts these at the centre of the design project. Both also observe that the concept of infrastructuring, as multi layered and consisting of both social and technical elements is crucial to support ongoing adaptations to changing
circumstances and needs.
Comparing the infrastructuring perspective to PD, I do not see that meta-design “transcends” PD as suggested by Giaccardi and Fischer (2003 p.2), Rather that it has a different emphasis and can be seen as complementary. With its observation of the inevitable incompleteness of design solutions, meta-design pinpoints an important but uncomfortable truth about software
development. Meta-design contributes by both alerting us to the widespread nature of this problem of the need for changes after design and encouraging us to embrace it as an opportunity. The need for tailorability is also increasingly recognised within PD especially though its embrace of the notion of infrastructuring and its interest in bottom-up empowerment. The concept of infrastructuring that emerges from PD is one that is broader in scope, and more critically grounded than meta-design. Its recognition of the diversity of infrastructure, its social elements, located contingencies and power relationships make its message about tailorability less straightforwardly put but fully grounded in practice and the observed complexity of
situations in which IT systems are deployed. Within a community setting, this stronger attention to politics is significant so the infrastructuring perspective is closer to describing the complexity of deploying and maintaining an IT system in this context, and the broad range of ever-
changing dependencies – most of them non-technical - that are an ongoing challenge.
Despite their differences, both meta-design and the study of tailorability within PD lead to the same next step of proposing forms of 'end user development' (EUD). EUD is focussed on addressing exactly this issue raised by both meta-design and the infrastructuring perspective of PD; how best to create tools that allow people who are not developers to take on at least some development tasks so as to be able to tailor their own systems. EUD is an established
perspective in computer science which studies and facilitates a broad range of methods by which users who are not developers may take on programming-like tasks. The field is surveyed
in 2006 in the book End User Development (Lieberman et al. 2006), which defined EUD as follows:
“a set of methods, techniques, and tools that allow users of software systems, who are acting as non-professional software developers, at some point to create, modify, or extend a software artifact”.
(Lieberman et al. 2006 p.18)
The most basic example of EUD is the spreadsheet in which the ability of users to write formulas that process data has been well studied as form of EUD (Nardi 1993). Another example given in the literature is the writing of custom email filters. These, as configuration files, can be seen as an activity in between programming and customisation. Some see EUD encompassing visual programming environments such as Scratch that are designed for children to learn programming concepts while building fun and creative games from 'blocks' of pre- written code, manipulated via an easy drag-and-drop interface. Aside from being notoriously difficult to design for, the major critique of EUD is that it leads to messy, undocumented and potentially insecure or inconsistent applications or data (Lieberman et al. 2006). Put in other words, the worry is that without the rigour of software engineers, too much freedom for users can be a dangerous thing.
However on the positive side, meta-design and the infrastructuring in PD connect with EUD by suggesting that EUD can be used by people to shape and refine their tools of work, and that this, in turn, is linked to sustainability. Tailorability, with appropriation, give users a stronger sense of ownership over and connection to their tools, and enables systems to adapt to changes in the external operating environment. Perhaps there is a balance of risk in EUD to be worked out. On the one hand a system without EUD may quickly cease to be relevant enough to work practice to be usable or be too expensive to adapt to new needs. On the other, there are the risks of bad practice, as detailed above, should the parameters of customisation be set too widely.
PD is focussed on the ethics and personal/social development that might come from 'design in use', whereas meta-design is more focussed on expanding creativity. EUD takes a practical approach that is linked to effective work and sustainability through flexibility. However, all seem to be missing a key element. This element is learning; how the capacity of these end users is increased to encompass the job of end-user tailoring. In discussions about EUD a question inevitably arises; how to locate or perhaps even create these local users with the expertise to perform local customisations who take such a key role. These expert users must be individuals with both domain knowledge and the right kind of technical knowledge or aptitude to tailor. Nardi and Gantt discuss this in relation to CAD users in their study from 1992. They identify a role in the workplace for those who can both understand the domain and tailor applications for themselves and others. They call this role “gardeners”, However, they highlight how difficult is is to find people who can do this, saying:
“A person who is technically skilled but uninterested in intensive interpersonal interaction may not have much of a green thumb when it comes to helping other users”
The idea of a special kind of user has gained some traction in studies of innovation. Some writers, for instance Von Hippel, emphasise the importance of such users to the development of new products, proposing that 'lead users' should be identified and explicitly brought into the design process (Hippel 1988). Others are more interested in how 'power users' emerge informally from user groups and then become key bridges between developers and users (for example Baskerville et al. 2000) . Meanwhile, the 'gardeners' described in Nardi's study are valued for their combination of technical and interpersonal skills (Gantt & Nardi 1992).
Within PD an ongoing and concrete example of how this kind of bridging role can be well defined and benefit both local use and wider development is in the work of 'implementation mediators' within the Health Information Systems Programme (HISP) project. The HISP project is a long running and well-documented PD project that works in partnership to develop a health information system that is responsive to the needs of communities and healthcare workers in the global South. The project is long running and geographically distributed but must fit in to a wide range of different local contexts (Braa et al. 2012) . To help address some of the challenges in both design and use, HISP makes use of “implementation mediators”. Implementation mediators work in the field in different countries; they have technical skills and are domain experts. They make sure HISP is working well in the local context and is tailored to local needs. But by also feeding information and requests to developers, their role also connects the local use context to the global production effort, as well as connecting technical and domain experts (Shidende & Mörtberg 2014) . Within a PD context, as Shidende and Mortberg argue, the implementation mediators must also have skills in facilitating wider participation in the communities they work in (2014).
It is certainly the experience documented in this project that the emergence of requirements and the need for tailorability after design is always present. This need is also something I have observed in work outside of this project that I have done as a web developer in other
commercial and not-for-profit contexts. In undertaking this project, the team did not explicitly plan for EUD, though we knew that our FLOSS-based system could create opportunities for it. We had no way of knowing if the community organisation would have the capacity or the interest to tailor for themselves. In the Hublink project, the desire for it emerged, and to some extent, it was possible to enable it because of the framework we were using. In the project description, we see that evidence points to both the desire and capacity for end-user development arising via the mutual learning outcomes of PD.
The project description provides evidence that tailorability has been key to the sustainability of this project, especially due to the changing external circumstances that led to new requirements well after the design phase. Without the ability to tailor, the application would soon have become unusable. The significance of these environmental changes are illustrated in the last part of the project description in which I present empirical evidence that shows the pattern of different kinds of tasks. Evidence shows that, out of all the team members, those who had participated in the design process were the most keen and most able to make use of these tailoring capabilities.
Infrastructuring, therefore, allows us to link tailorability, FLOSS and the mutual learning outcomes of PD as parts of the human/social side of this infrastructure, and secondly, that open source software can provide a key technical element. Meanwhile, we will see later in this review how the specific characteristics of FLOSS practices and products make them especially suitable for deployment in an infrastructuring context.