Designing in between Local Government and the Public – Using Institutional Analysis in Interventions on Civic Infrastructures

In ‘infrastructuring’ participation, Le Dantec and DiSalvo (2013) introduce the concept of ‘attachments’ as a set of relations and commitments individuals make to a shared matter of concern. These attachments are voluntary, for the most part, informal and flexible. Dewey defined those ‘civics’ as groups ‘indirectly and seriously affected for good or evil’ who become ‘distinctive enough to require recognition and name’ (Dewey 1927). In the formation of civics in relation to shared concerns, Le Dantec and DiSalvo (2013) see ‘infrastructuring’ as a set of practices of enablement, self-recognition, appraisal of resources (software, information) and people available towards a revision of the present situation. Here, infrastructuring is described as distinct from participatory design, as it focuses more on a possible realisable future with the means available ‘in place’, where available (digital) technology is used as part of an intervention. Innovation and what is defined as ‘new’ is then very much included, embedded and explored in the everyday practices and, thus, what is ‘new’ goes beyond technology-focused definitions of newness. Le Dantec’s argument reminds us of earlier iterations attending to everyday practices as interesting sources of designs (Star 1999; Suchman 2005). Perhaps it might also be exemplified in past endeavours in which participatory design activities were applied over a timeframe of eight years for the design of a media space (Dalsgaard and Eriksson 2013). As a result, established methods of co-design took on a novel quality as they became embedded within and taken for granted by stakeholders involved in the construction project.

In practice, the distribution of ‘ownership’ in civic infrastructures and the work of ‘infrastructuring’ itself, across many different actors with formal and informal influence, quickly become a political matter. For example, in the development of a community network, researchers noted the irony that ‘ordinary people’ whom the project sought to empower were absent from early design discussions (Carroll 2005). In redesigning a semi-public space using media technologies, boycott by or turnover or resistance of key stakeholders was observed (Saad-Sulonen 2010). This may particularly be the case if there are very different expectations towards use of space. Furthermore, as developing and embedding any civic infrastructure is costly, it requires private investment and specialist skills, and will thus almost always be mediated through experts (Rattray 2006). Thus, perhaps the degree to which technology deployments can truly be driven ‘bottom-up’ by citizens may be limited (Townsend 2013). Often, intermediaries play a role by ‘creat[ing] connections between [...] different tools’ and the infrastructures they create are ‘both technical building blocks and artful integrations’ (Saad-Sulonen 2012, p. 23).

In the use of civic technologies, geographic space has greater influence in the politics of the design context (Dourish and Bell 2007) and is important in designing for everyday ‘situated engagements’ of public expression (Bohøj et al. 2011). For example, a review of 40 community websites around Amsterdam found that most websites are focused on informing, helping and asking for help, and connecting with neighbours; however, the review also identified resource sharing, organising activities and crowdfunding of community projects (Niederer and Priester 2016). Thus, the kinds of civics relevant to this article overlap with the material context, from which they draw power and legitimise demands for change in interaction with local government (compare Natarajan 2015). For example, the development of a street archive and description of practices of information collection heavily overlapped with the material contexts of the road, creating new interaction dynamics among social actors living there (Taylor et al. 2015). A resident took on the informal role of ‘archivist’, raising her ‘power’. Subsequently, some residents chose to give their data only to the researchers involved who were perceived as neutral intermediaries. In public services, geographic proximity may also enable civic groups to form on matters of concern, as documented in a bridge repair project that helped groups of residents and a team of academics to combine to achieve a shared goal (Le Dantec and DiSalvo 2013). In the urban context, the material context thus remains of importance as a source for objects of contention, mutual interest and a sense of community.

If the goal is to change, for example, the established (political) institutions and digital infrastructures, as some demand, interventions that put community at their heart, working at a small geographical scale or as a single group, can struggle to make such ‘vertical impact’ (Taylor et al. 2015). In fact, research on local authorities’ planning processes found indications of a strong separation between information technology used at different levels of political administration (Weise 2016). Thus, community informatics, the study and embedding of software with local civic groups, perhaps disregarding local authorities’ setup, ‘is a necessary but not a sufficient condition’ for new forms of participation (Staffans and Horelli 2014) due to the difficulty of scaling such ‘bottom-up’ initiatives to meet the workings of local government. Perhaps a better strategy is to give attention to the design of suitable technical interfaces and transparency within local government. Civic technologies offer opportunities to form ecologies of people, practices and ‘data forms for generating, viewing and possibly analysing data’ in matters of civic concern (Taylor et al. 2015). The question is how such practices can be better supported, for example by the process-orientated work of local government that is more directly driven by national laws and requirements. Where are the processes – not to mention the political action – that will enable such civic infrastructures to be formed and changed over time?

Over the years, research on collaborative work has involved all levels of government. As a military psychologist, Hutchins (1995) worked for and with national government agencies studying coordination and cooperation on aircraft carriers through detailed ethnographic work. Engeström and Escalante’s (1996) detailed longitudinal study of the ‘postal buddy’ self-service kiosk for the US postal service involved work with postal agencies broadly delivering an important public service. Bonnie Nardi’s work related strongly to the role of public libraries as centres of communal learning (Nardi and O’Day 1999). Other studies charted the deployments of early local computing networks and public question-and-answer software as in Santa Monica’s PEN project (Rogers et al. 1994). This is complemented by the rich work on participatory public geographic information systems, which developed in the 1990s alongside and often out of sight of the usual literature on computer-supported collaborative work. For example, Rattray (2006) discusses the deployment of an open-access web-based geographic information system across the whole of the involved US public-sector bodies. In this line of work, some academics also experimented with participatory approaches to let unskilled residents customise ‘expert’ software (e.g. a geographic mapping application) to help create civic maps for things like perceived public safety across a neighbourhood (Leitner et al. 2002).

While user involvement in software deployments has been a pertinent concern throughout the literature on participatory design, arguably there has been a change in the quality and intensity of research that has regained an interest in the boundary of the public sector (government) and the public at large (Bødker and Zander 2015). As shown by Handler and Conill (2016), socio-technical developments now make it likely to encounter large datasets online. The literature on open data and associated hackathons has been testament to this. Local government now grapples with being accountable to publics who have increasingly become apparent online (Olivier and Wright 2015). In a take on participatory design in everyday civic life, the concept of ‘infrastructuring’ recognises that user involvement can also relate to the creative linking of a range of software particularly in low-resource and perhaps low-skill social contexts in civic groups (Le Dantec and DiSalvo 2013). Dismissing any comprehensive technical ‘solutions’, some useful approaches were made in studying organisational settings through contextual design, combing artefact study with prototype design (Beyer and Holtzblatt 1999) or the ‘locales framework’ emphasising the value of everyday practice as inspiration for design (Fitzpatrick et al. 1998). Added to this, hybrid methodologies have been proposed combining life logging and automated trend spotting to establish how issue-based publics form (Ludwig et al. 2016). In meta-design (Fischer et al. 2004), the wider socio-technical and socio-political contexts of institutional processes are the outcomes of design and, in turn, help to explain constrains to design choices.

In terms of participatory approaches with local government and civic actors, UrbanSim (US) and the Aarhus Media Space (Denmark) provide examples for prolonged engagement with political forces in a design project. Initially, the UrbanSim project aimed to provide a civic technology for developing multiple perspectives on quality of life with the ambition of letting anybody come up with their own assessment schema (Friedman et al. 2008). They recognise the impracticability of participatory methods that claim to involve everybody. They thus involved a set of civic groups in a ‘targeted co-design process’ by drawing heavily on forms, data schemas and assessment methods that those groups had developed already. Furthermore, the co-design of the Aarhus Media Space, a new library for the city of Aarhus (Denmark), counts as another example of longstanding, intensive engagement of the public, over a timeframe of eight years (Dalsgaard and Eriksson 2013). Through the long timespan and recurring participatory action, participation became seen as a normal, expected thing, part of the ‘infrastructure’ of the project if you will. Time, resources, committed leadership and understanding of the value of participatory activities were mentioned as key issues. The cases demonstrate (1) how software development tapped into existing datasets from local organisations (Friedman et al. 2008); and (2) how long-term projects relate to stakeholders and their processes (Dalsgaard and Eriksson 2013).

Short-lived design projects are more likely to discount the role of established practice and the diversity of audiences’ different agendas. For a research strategy, the study by Dalsgaard and Eriksson (2013) pointed to the importance of developing a structured scheme through which to capture results from activities with various stakeholders, so to ‘infrastructure’ means to capture learning. This is a particular concern for the strategy outlined in this article also. Prior methodological approaches have perhaps failed to exploit these broader insights for guiding and shaping the design of their technical interventions. The argument of our paper is the underdevelopment of suitable ethnographic approaches for the design of civic infrastructures based on awareness of political structures that influence the deployment of digital technologies and where user groups may not initially be readily identifiable. This underlies the need for social ‘sustainability’ of civic infrastructure, in the sense that any collaborative software is only effective if it speaks to the needs and concerns of the audiences it engages (Chilana et al. 2015) on a voluntary basis.

Essentially, this article suggests revisiting institutional concepts in systems design through a focus on regular (i.e. patterns of) interactions and the recognition of those patterns in design. This approach is particularly useful for work that takes the organisation of local government as a starting point, for example by relating to available public data or administrative processes issued by these organisations. The emphasis here is on looking at the interactions between local government and civic groups (Bødker and Zander 2015). Based on the available literature, we have attempted to group requirements towards an analysis strategy along three important characteristics of the civic infrastructure, the set of software involved in establishing, mediating and sustaining civic participation in local government. It is likely not an exhaustive list (see Table 1).

In response to these requirements, the rest of this article outlines working within local government contexts to generate insights on potential future interventions. The strategy relies on the observation of regular practices that may be influenced by existing regulatory requirements and that, in turn, may influence the designs of civic technologies that are made available to the public at large. When looking at the various ways in which research projects have approached political participation on local matters, the institutional approach outlined may thus lend itself to the kinds of methods that aim to work in collaboration with local government (Korn and Voida 2015). In participatory action originating with interests that diverge from local government actors, ‘disruption’ occurs where there is an orientation that does not fit the established political process and where political action is very much part of special moments. Korn and Voida (2015) note ‘activist technologies often support immediate, short-lived campaigns and events that result in a number of individual protest actions’. Increasingly, this stream of work begins to recognise the importance of recurrence, raising the need for methods aimed at the institutional aspect of civic action. While this is an issue in the literature, what is essential in longer-term participatory design practice are plans for ‘how to capture insights from participatory activities in a structured way’ (Dalsgaard and Eriksson 2013).

In the object-centred democracy that Le Dantec and DiSalvo (2013) describe, various civic issues become action situations to be supported by civic infrastructure. For example, in the Tenison Road project documented by Taylor et al. (2015), the group of residents who self-organised a data archive and analysis of the traffic movements through their street is part of a wider network of actors including, for example, the traffic planners at the municipality. Through the flows of traffic, their interests may be linked, and yet both groups act in different arenas associated with different rules, interests and incentives, facing the problem of traffic management from different viewpoints (compare Natarajan 2015). In a democracy focused on civics arising from the mutual recognition of matters of shared interest (‘object-centered democracy’), Le Dantec and DiSalvo (2013) suggest that software and related data artefacts involve practices of ‘infrastructuring’ to support civic participation. Civic infrastructuring practices arise in recurring, often voluntary, participation in the linking, use and consumption of sets of software and data.

The institutional analysis and development (IAD) unpicks the interactions that make up these infrastructure choices. Ostrom (2005) breaks complex settings into several bounded action situations (e.g. a series of related events for which similar rules may apply) with the rationale that ‘what is a whole system at one level is a part of a system at another level’ (Ostrom 2005). For a methodological strategy, this implies studying the interaction between a range of actors across local government and civic actors across ‘various locations on a micro-macro continuum’ (Mjøset 2009). In this way, civic informalities and the process-orientated actions of local government may be put in conversation through the participation practices observed across local government and various interest groups. Imagine the example of interactions between civic actors of the Tenison Road project and relevant city council officials (Taylor et al. 2015). Thus, action situations occur on multiple levels. Levels have different agency and the rules they create may either perform as authoritative enforcing links (e.g. rules negotiated by local government affecting a civic group) or sequential links (one consultation building up on results from an earlier one) (Figure 1).

The institutional approach’s strength lies in the awareness of how and where the design context originated, which define the social interactions we observe. Both local government and civic groups make ‘operational’ and ‘collective’ choices for how they like to organise, concepts that can be related to Fischer’s idea of meta-design (Fischer et al. 2004). Meta-design aims at those contexts (rules, procedures, software) collaborators agree to temporarily fix to establish a series of collaborative moments. At the ‘operational level’, practices affect the content of participation moments; for example, through the submission, evaluation and manipulation of information artefacts (e.g. files, documents, data). Conversely, on the ‘collective choice’ level, action situations relate to setting contexts for participation; for example, by configuring the sets of software enabling participation moments. They include, for example, the participation in modifications to software that would have implications for all users. Opportunities for intervention exist on each level. If a local government tries an engagement campaign with new rules of participation, it may choose to formalise its use through agreements at the collective action level.

Interactions are analysed through (1) formal rules; (2) observable practices of actors; and (3) narratives (Lowndes and Roberts 2013). In addition, for civic tech, we also notice where the action is (both in a geographical sense as well as in the sense of social contexts). Ostrom (2005) suggests seven rules when looking for patterns of interaction for individual action situations. According to Ostrom, ‘rules form a part of the structure of the situation rather than a solution to the [interactions] at that level’ (Ostrom 2005). Thus, they become objects of study. The goal is to avoid dogmatism in relation to ‘rules’ and rather approach them as descriptors of any patterns traced by asking relevant questions (see Figure 2). These categories of rules help to describe different roles and associated capacities to access information and participate in choices. From the perspective of collaborative systems, they influence how articulation and coordination work involved in interactions occurs.

In civic infrastructure, positions are the potential roles or personas that participants may take on within an action arena. From there on, it is possible to focus in detail on the practices of the actors in the various (e.g. how information access differs across several roles: admin, normal user, institutional user) in more general terms, by abstracting what information rules were in place. With an ethnographic approach, rules-in-use can be probed to understand the acceptable practices, habitual actions or actions that have been proven to derive beneficial outcomes between various social actors across social contexts and with various ICTs. A rootedness in everyday practices and procedures (Galloway 2004) implies the deployment of rules as sensitising devices in actor interviews. Thus, they are useful for discerning observed or reported patterns of interaction among various actors and ICTs within a specific context to speculate about potential interventions that provide a benefit, such as a faster process, greater satisfaction by the various participants and, lastly, voluntary participation.

Digital mediation of civic interactions and the need for local government practices to adapt provide ample context for design. This is current in a wide range of domains and technologies, including, for example, mapping and crowdsourcing apps that seek to help publics gather by ways of commenting on places and spaces (Bohøj et al. 2011; Saad-Sulonen 2012). In the UK, for example, lower costs of online self-service (£8.62 for face-to-face vs. £2.83 for telephone vs. £0.15 for web) are among the factors driving local government transformation (Local Government Association 2015). Here, we look for projects at the boundary of local government and the public at large for example around a specific public service. This may be the case where services are redesigned to manage demand in different ways. For example, in the UK, local authorities consider which services are not required of them that could be colonised by other actors instead. In the UK, ‘neighbourhood planning’ serves as an example where legislation enabled this to happen (Parker et al. 2014). This might imply asking how civics could take on ‘self-governance’ of their key concerns (Boonstra and Boelens 2011). Extending the argument of ‘multiple participations’ (Saad-Sulonen 2014), the roles of local government officials shift as capacities and responsibilities are shared out to other civic actors in the process. Thus, these changes call for development-orientated processes, both to design new rules of interaction but also technologies and systems that fit ‘in’. This, we argue, is at the heart of the method presented here.

Whereas computer-supported collaborative work may set its focus on collaboration within and across defined organisations, civic infrastructures involve local government and various public groups trying to influence the design context. Thus, best cases are found by focusing on the interactions across and between multiple actors on choices affecting the outcomes for the wider public. For the case discussed in this article, the authors had several initial conversations with two planning officers regarding the key challenges in engaging with external audiences. Additional focus for a research priority arises from conversation with public representatives. However, what is mentioned here as challenges cannot be taken at face value. In the context of local government, they may well surface in public expressions made by individual participants voicing frustration, contestation of or dissatisfaction with existing practices. These public views can be easily corroborated but may well be contested by local government representatives referring to a divergent set of values, resources, capabilities and legal requirements. For example, we encountered the following comments when reviewing the documentation of a public consultation. These two comments were part of a set of official statements expressing confusion and sometimes anger towards the events they took part in (see Figure 3), and thus indicated a problem space to address:

In the formation and use of civic infrastructures, one of the crucial challenges to designers is the diversity of actors, material and social settings involved. This has been recognised in the participatory design literature. For example, in a review of relevant studies, Halskov and Hansen (2015) differentiated between ‘implicit’ and ‘explicit’ involvement of ‘users’ that are taking and steering the design process. Increasingly, users may partake implicitly, for example, through the selective involvement of key audience representatives and the use of audience research from secondary sources, such as industry reports, but crucially also through feedback already provided to, for example, local government from official consultations (such as that the author encountered in their own work: see Box 1). However, beyond mere collation of such resources, work is required to draw links ideally temporally and geographically.

To situate social interactions, this methodology contends that material contexts are a key filter for actors who become involved in civic infrastructure projects. In the formation of public choices, material spaces serve as areas for sociality, interaction and political contestation. Alluded to in various examples, such as that of a street-level data archive for Tenison Road, design approaches for civic infrastructure could give greater recognition of the material situatedness of political expression (Taylor et al. 2015). Taylor et al. (2015, p. 2871) recommend ‘think[ing] of structures that support some kind of representation of data’s active presence in place. These might express how data travels geographically and between people, and when, where and with whom it gathers significance (traversing through the contours and across the boundaries of a social geography)’. Material structures and physical space perform co-constitutive roles for social exclusion, and material contexts have been rediscovered to play greater roles in policy making (Murdoch 2006). Applications of computing to the urban context rely on networks of people, technologies and the materiality of their places (Foth et al. 2011). These points give substance to the key argument to study context first through awareness of places, people and technologies involved.

Approached from the standpoint of public services, this calls for time to be spent on ‘data work’, collating the substantial corpus of relevant documents, including the various documents and media involved in the process under study. These policy documents, media, event attendee registers and records of public representations made from both public archive information and direct enquiry with organisations involved map onto the artefacts in circulation. They may include structured evidence, maps, action plans and process documentations. In our work, the materials of potential use are listed below (see Figure 4).

For the establishment of context, the key technique is then the identification and mapping of key entities within this dataset and, therefore, the establishment of person-place-material linkages across the data corpus. This relates to the advice to plan for capturing insights of later participatory events in long engagements (Dalsgaard and Eriksson 2013). For the establishment of the material contexts, a range of parsing techniques can be deployed that are increasingly automated. As mentioned, groups involved in civic infrastructure often exhibit a strong attachment to place (Bilandzic and Venable 2011); their material contexts present a source of power and legitimacy (compare Natarajan 2015). For example, for an analysis of civic participation in urban planning (overview provided in Figure 8), we parsed statements from public consultations to achieve a geographical view of key matters statements from public consultations to achieve a geographical view of key matters of concern. Based on public records, we could establish the patterns of interactions of citizens across a district, shaping the contours of activity underlying civic participation. The ability to view engagement with a public service in this way enabled us to recognise different forms of civic participation across individuals and established differing levels of participation across towns and hamlets of the district. This technique helped to identify potential participants for contextual interviews, carried out at a later stage. It has become ever easier to apply such methodological strategy to data sources online (Hecht and Gergle 2011). Most online datasets (including social media data) contain geospatial references facilitating the description of entities, places, localities and concerns in space (Hahmann and Burghardt 2013). This makes it extremely handy to establish understanding of the socio-spatial context to specific civic concerns within the overall landscape of engagement campaigns run by political organisations.

Often, the focus on a particular public service and its engagement with external audiences enables us to recognise patterns of interaction; for example, in the formally defined ways to become involved and obtain influence in and through this service. Usually, there are recurring events and meetings of various sorts that take place as part of government operations that we would have picked up through an earlier document review (see Section 4.2). The preparation of strategies of action and the associated consultations offer a myriad of rhythms, often expressed through processes structured into stages at which engagement moments become particularly well defined (see Figure 5). In the search for patterns, the gaps between recurring actions (e.g. sets of public engagements) are often the most interesting, as it is here where certain decisions are made. In the further exploration of the socio-technical-spatial context, the approach now looks explicitly for documentation of process, including digital technologies involved. In this step, it is key to encounter lack of documentation around technologies, artefacts and processes in use. Methods such as process tracing (Langley 2009; Mjøset 2009; Pettigrew 1990; Langley and Tsoukas 2010) are of assistance. Here, we notice any sets of software and sources of data storage (which may well be non-digital) employed in mediating interactions across audiences. For designers, these artefacts of civic infrastructures become the central ‘objects of contention’ (Suchman 2005). Through ethnographic techniques, including mapping and tracing of actors’ different roles involving the infrastructure, we can establish ‘users’ and ‘providers’ of ICT facilities and the ‘producers’ and ‘consumers’ of information artefacts.

An institutional approach employing a structured frame, such as the adapted IAD based on Ostrom (2005), helps in situating political action associated with civic infrastructure within a socio-technical context. It serves as a guide for the analysis and requires updating and customisation to the specific case at hand. Across a range of action situations, the analyst identifies the information artefacts, software and actors involved and determines principles that guide interaction. At times, abstract mapping, including sketches, may be appropriate to documented social interactions (Leitner et al. 2002). Recognising the social life of infrastructures, being in flux and co-constituted and instantiated through various everyday actions (Amin 2014), the analysis documents and maps out past events with knowledge of actors, ICT facilities and information artefacts involved and creates initial assumptions about the relations within the design context. For example, alongside the mapping of participants in the preceding section, event chronologies can be helpful in differentiating between online and a range of offline events, recurring events (such as monthly public meetings), and abstract patterns of interaction across the multiple social settings and sites over time. Through the longitudinal view, stable categories and rigid infrastructural arrangements appear more changeable and the ‘logic’ of the present setup is appreciated. In the context of local government, study of archival data is of importance. For example, in the authors’ case (see Section 5), cross-linking data from official participation records across different engagement campaigns (action situations) enables us to establish who the contributors are, how many there are, how much they have contributed (both when and the number of comments) and what type of stakeholders they are. Grouping is then possible by organisation membership (government, charity, commercial organisation) and location.

The methodology calls for a focus on a select set of key action situations where contestations and interactions between local government and various civic actors may occur. Following the strategy suggested here, interviews are employed to provide insight into the sociality around and uses of various technical artefacts and data components encountered in the design space. After establishing actors, technologies, and key media and communications (referred to as ‘constituents’) from documents, databases and conversations with key actors, a range of stakeholders (officials, community representatives, technologists) are interviewed to investigate patterns of interaction across a range of action situations. Here is where the institutional framework helps to create a frame for the analysis, where it is both productive to (1) focus on actions and activities as carried out in contextual enquiries while also (2) looking for statements that indicate how decisions on particular events were taken. Using in-depth interviews (e.g. contextual interviews: Beyer and Holtzblatt 1999), we can reconstruct the patterns of interaction across a range of action situations (such as remarkable events) by looking for ‘rules’ introduced in Figure 2. In interacting with interviewees, we think about (1) practices, (2) stories and (3) formal rules that let us draw conclusions about the institutional setup (Lowndes and Roberts 2013).

Discussing the roles of various participants in ‘data work’ and relating the conversation to wide types of data from published records of comments made by the public to systematic statistical data is not necessarily obvious to participants. The interview excerpt below (Figure 6) portrays a situation the authors encountered when enquiring about the role of civic groups that played a prominent role in a case of civic infrastructure use. The excerpt demonstrates the starting point of the institutional frame for different ‘roles’ that organisations and individuals within these social constructs may play. In this excerpt, several roles are mentioned, including that of co-organiser, active contributors (and even local residents). In the next step, the interview discusses the rules applying to individuals contributing from different positions. As a result, the analysis perceives roles and associated responsibilities of various actors across different social settings.

A key part of the institutional approach is to understand dependencies, indicated by the differentiation between ‘collective choices’ (applying to all civics) or ‘operational choices’ (or only one group). In so doing, the analysis can consider instances where narrative accounts of participants going back months (see Pettigrew 1990) indicate constraints to the actions of a civic, and trace through where these constraints originated. Here, analysts can employ event chronologies, images of workshops and information on statements made in online consultations (with related information on the time and mode of submission) to help recall details of their interaction with other actors, technologies and information artefacts in hindsight. Russell and Chi (2014) suggested that adherence to a walkthrough from the distant past to the near present, face-to-face modes of interviewing and avoidance of value-laden questions were helpful for retrospective reflection.

In the data synthesis, we then combine both data from the analysis of context and the detailed contextual interviews by development of a coding scheme with grammatical, process and rule codes (Saldaña 2012). The rule codes provide institutional categories and are inductive concepts (arising from the framework; see Figure 2). On the other hand, a markup of the chronology of actions taken by participants and official decision-makers over time establish a chronology of action (Baskerville and Myers 2014) arising from the dataset. Analysts could, for example, employ what Saldaña (2012) termed process coding to identify recurring interactions across a range of participants and action situations. He states that recurrence is ‘both natural and deliberate – natural because there are mostly repetitive patterns of action and consistencies in human affairs, and deliberate because one of the coder’s primary goals is to find these repetitive patterns of action and consistencies in human affairs as documented in the data’ (Saldaña 2012). In terms of rules relating to recursive action situations, a series of related events with very similar rules of interaction may become apparent through this analysis.

Below is a suggested list of key coding classes that we developed from our own work and that can be applied to both archival data as well as contextual interview responses (Figure 7). Our final coding schema included seven code classes to establish linkages across the research data. In the application of the institutional frame, the ‘rule’ codes match the questions we introduced in Figure 2. The material relevant to rule codes had different qualities, depending on whether these were applied to data from the contextual interviews or documents publicly released by, for example, the local government officers. Especially if applied to interviews, this clarified patterns in practices. For example, regarding internal officer meetings, it became clear that there are some generally accepted assumptions that guided interactions in those meetings as to how public comments might be flagged and discussed in the team. Those interactions were not otherwise explicitly stated, but may be clarified through the artefact-focused contextual interviews. In terms of interventions in the public sector, the analyst notes that boundaries apply to who may join different events and inclusion/‌exclusion criteria are negotiated.

Studies of civic infrastructure amass a large amount of heterogeneous data (e.g. qualitative and quantitative, geographically referenced, temporally placed) about the individuals, places, ICTs and information artefacts across a range of social settings. Throughout the projects, the generation of a database helped to map the various actors across action situations; so, as insights become available, they can be added so that its data schema becomes a template for a future intervention. For the multi-site, multi-institution research context of technical interventions in the urban context, databases consolidate data from different social settings, sites and technologies. Integrating information from a range of archival data facilitates a cross-linking and slicing of the data along different dimensions of interest (e.g. by events, participants, interactions). For example, to elucidate a socio-spatial context, information in archive materials and public data can often be geo-coded to link to a spatial analysis package establishing relations and linkages between people and places. A database where relevant data fields can be flexibly generated, changed and linked can emerge as an imperfect, interactive representation of the patterns of interaction and the information space itself, capturing important properties, statistics for actors and ICT facilities, as well as information artefacts. This is an iterative process and helps in combination with the development of design probes and prototypes to deploy in the research setting (Baskerville and Myers 2014). Thus, structured databases may present a ‘structure’ for a technical prototype and thus inform the technical intervention.

The approach we outlined here, using institutional concepts and thinking in terms of ethnographic exploration of civic infrastructures, is centred on the premise of recognising the ‘patterns’ that form public choices. Democratic processes depend on formal codified rules, and even more so in informal, unspoken and perhaps undocumented agreements. An example of this is the life cycle many public choices in local government go through, often involving recurring moments of public engagement or political debate in set meetings, each often with recurring patterns of participation. Korn and Voida (2015) presented a well-formed discussion of the different bases on which moments of political expression may stand and, consequently, how designers may craft technology or systems supporting them. They suggest that political expression in everyday life may be aligned with local government (as a form of situated engagements, for example, using government-provided apps) or may be unaligned and thus provide various forms of ‘friction’ that speak to critique established local government processes. In the various ways in which political expression may happen in the everyday, institutional approaches, one might then argue that designers would still benefit from recognising and describing the formation of patterns of engagement within these projects. We consider that even the most informally organised civic group may have many unwritten understandings of how agreements are made that concern the group and, increasingly, many run websites and other hyperlocal media platforms to make their voice heard (Niederer and Priester 2016). During the initial research and requirements stages, the levelling of action situations (as recurring participatory patterns) as well as their influence over outcomes (e.g. across local government and civic groups) helps to differentiate design requirements for different audiences.

In respect of information systems research, Goldkuhl (2012) suggested that studies of technology can take a predominant interpretivist or pragmatist orientation. Ethnographic approaches in the design of technologies are usually focused on designing interventions. Such approaches, focused on action and change, are associated with ‘pragmatist’ ends. Pragmatism is a wide-ranging field, but has been hugely influential in design research (Dalsgaard 2014). However, other approaches may be more aligned with interpretivist traditions and are, perhaps, the kinds of studies that provide researchers and practitioners with food for thought; for example, by reworking and revisiting the successes and failures of previous deployments (for a good example see Engeström and Escalante 1996). Goldkuhl (2012) suggested that approaches from a pragmatist orientation and an interpretivist orientation can be combined, with each surfacing with a different dominance: he advised that ‘either interpretivism is seen as instrumental for a pragmatist study or pragmatism is seen as instrumental for an interpretive study. This means that each paradigm can be the base paradigm allowing elements from the other paradigm to be used in an instrumental and supportive fashion’ (p. 144). Thus, in terms of the desired ends of a study, we choose a direction, and it is advisable to choose a pragmatist approach if the research outcomes are believed to create ‘constructive knowledge’ that may be useful in ‘action’. Principally, if the study is purely aimed at generating knowledge that is interesting from a theoretical standpoint (Goldkuhl 2012), an in-depth retrospective study is worthwhile (an example for this is Engeström and Escalante’s (1996) study of the postal buddy system in the US, a talking self-help kiosk developed with good intentions for the US postal service, but which failed to be adopted; see Engeström and Escalante 1996).

An interventional use can be coupled with action research intervention in the study context (Dalsgaard 2014; Mjøset 2009). For instance, work by Saad-Sulonen (2012) was an example of a follow-along (interventional) study. For several years, she tracked the use of an online platform for planning. In such contexts, it is usually of interest to consider not only direct end-users, but also the needs of those sponsoring the technology, maintenance staff and operators. Design of ethnographic approaches include ‘potential rationing’; in other words, a future design possibility as well as a co-design (Baskerville and Myers 2014). Here, the interactive template of the IAD helps to ask essential institutional questions necessary to embed the technical intervention across various actor groups.

The assertions of user empowerment found in the literature on participatory design is complementary with the approach described but, in addition, this institutional methodology emphasises the political context of the various social interactions involving civic infrastructure. Lowndes and Roberts (2013) contend that institutions can be studied as detached from the observed or reported practices and rules, but from the standpoint of user-centred design, it seems more desirable to become actively involved through document study and interact with participants; for example, through contextual interviews. Participants in the analysis become conversational partners and potential project champions that facilitate the success of an interventional approach that seeks to change existing patterns of interaction through technical support.

As these social settings often involve a large number of people, even co-design approaches have begun to take a nuanced stance to appreciate the value of representation of different user audiences through proxies and/or even archival data. For example, in the understanding of ‘participation’, Halskov and Hansen (2015) differentiate between ‘implicit’ and ‘explicit’ involvement of ‘users’ that are taking and steering the design process, and the idea of a ‘mutual learning between users and designers’ which may indicate a sort of middle ground of intensity in which users have an influence on the design process (Halskov and Hansen 2015). The ‘macro-HCI’ approach, by which specific patterns of interaction are derived from archival data and existing ICT facilities, provides a template to identify participants who partake more explicitly.

The underpinning of the institutional approach – its assumption of small iterative changes in practice – is suitable for everyday actions affecting civic infrastructure (Galloway 2004; Amin 2014). At its core, it is strongly concerned with the tracing of actions and rules. The philosophy lends itself to design research approaches, which feature a bias towards action and change (Dalsgaard 2014). In application, capacity building was involved and the underlying assumption was to distribute ownership over the technical intervention across institutional and community actors. Generating comprehensive relational data archives alongside the analysis helps in multiple ways. It helped us to learn about important challenges to technical interventions (and thus has practical relevance), it elucidates the aforementioned actors’ needs (and therefore has relevance for the intervention), and it has scientific relevance. It does so since explanations for success or failure are found by tracing patterns of interaction. Ostrom’s principles of good governance for sustainable common pool resources are one example (Hess and Ostrom 2006). Given the experience of the authors, it may be easier to understand ‘success’ and ‘failure’ by looking at the history of the design context under study, since the concerns of local authorities and those of their respective areas can differ quite substantially.

Lowndes and Roberts (2013, p. 10) caution against ‘overstretching’ the use of what is ‘institutional’. They state that an ‘institutional explanation puts political institutions first’. Applied to the study of information systems, such an institutional perspective, based and explored through reported practices of individuals, their reflections on those practices and meta-data from ICTs, implies that institution(s) are probed through quantitative and qualitative sources and that the political is understood here in relation to the use of technical interventions on civic infrastructures. Given the political nature of civic infrastructure, the institutional dimension is useful for unearthing authoritative links, but it is important to view those as rather more informal arrangements, especially when civic groups are concerned. There is no use in trying to overly formalise an informal context, but being able to frame what describes the choice-making in civic groups helps to ensure all relevant stakeholders are properly considered in a project.