Understanding openness through (in)visible platform boundaries: a topological study on MOOCs as multiplexes of spaces and times

On the EMMA interface, networked relations particularly come forward in webpages that enable interactions between learners and teachers. These interaction pages are named ‘Conversations’ (i.e., text boxes with comments and replies on lesson pages; see Fig. 1), ‘EMMA posts’ (i.e., pages where all learners and teachers can post and reply) and ‘virtual classrooms’ (i.e., similar to EMMA posts, yet only accessible for learners and teachers in the same MOOC). Furthermore, learners can interact with each other on dedicated Twitter and Facebook pages presented through hyperlinks in the majority of MOOCs. All these pages follow a particular structure: contributions of learners and teachers are delineated in comment- and reply boxes, are labelled by a name, a photo and a date. The comments are ordered chronologically, which creates a sense of linear time, yet it also emphasizes the asynchronous nature of the interactions (Bødker & Brügger, 2018). Moreover, the lines around the text boxes and the individual ‘signatures’ withhold contributions to merge into a different form like, for example, a Wiki. Thus, although the interaction pages enact shared spaces in which learners can act and communicate, the specific structure also illuminates that they act separately without being able to make adjustments to this separation (cf., Star, 2010). This space-time establishes an imaginary of learning as a practice of attaching pieces of information on a common surface, instead of attaching (i.e., relating) learners to each other. This network topology therewith presents itself as a spatiotemporal formation, a stabilizing construction so to say, in which the boundaries bring learners and teachers together in a coherent frame while remaining fragmented (Fig. 2).

Overlapping with this construction is a network topology that is enacted through release and closing dates of MOOCs and their lessons. That is, traces on the EMMA interface show a high peak of interactions within particular periods and on particular pages, namely, those of page-courses that are announced to be ‘active’ and closing soon (see Fig. 3). Initially, these announced dates help to locate on what pages their teacher and peer learners are present and in what week, in order to know where and when to ask them questions or finish collaborative assignments. Even though MOOCs are still ‘open’ in a self-paced mode after the deadline, the announced dates operate as spatiotemporal boundaries that establish a ‘current time’ or a ‘synchronous’ time that is typical for a classroom (Bødker & Brügger, 2018; Leaton Gray, 2017). This network topology is, thus, spatiotemporally forming a co-presence and bounds a common path with shared spaces-times in which learners can attach to each other rather than merely attaching to the same surface (Fig. 4).

Moving beyond the interface, networks are made up by relations between EMMA associates² that generally remain invisible on the screens of visitors’ devices. These associates mainly comprise institutions that provided MOOCs (i.e., providers), partners that cooperated in developing platform features (i.e., partners), the EC as the funding institution and the EMMA team. Because the development of the website remained contingent upon the relations among associates, the EMMA team had a daily task in maintaining these relations through recurrent contact. As some of the associates worked on the communication side, some on the educational design while others on the data processing and analysis, the EMMA team had to negotiate between different forms of knowledge. In order to do this, they worked with ‘hybrid’ knowledge that helped them integrate the ‘local languages’ (i.e., jargon of the different associates) into the general objectives of EMMA (Callon, Lascoumes, & Barthe, 2009; Star & Ruhleder, 1996). For example, team members with didactic responsibilities translated their background knowledge into ideas for digital educational tools, whereas IT staff members merely used their knowledge on coding languages to develop these tools. Subsequently, teachers were able design their MOOC according to their knowledge on the subject matter and based on their educational objectives with a ‘ready-made’ platform, not knowledgeable of its underlying code. This means that team members were continuously working on boundaries rather than working to overcome boundaries. However, these boundaries enact shared spaces in which they could work together without common knowledge yet with a common interest (Star, 2010; Star & Griesemer, 1989). This made that the formation of relations appeared as a stabilizing construction, yet in a different form than on the interface. That is, all associates were connected through a common interest, and where able to adapt the construction from their own disciplinary viewpoint (Fig. 5).

Networked relations beyond the interface are also performed through particular tools. One tool was the ‘Gantt chart’, a timeline in which ‘work packages’ (i.e., clusters of tasks) and ‘deliverables’ (i.e., planned outcomes of the work packages) are ordered. Another set of tools comprised different kinds of software, such as those for project management, file sharing, telecommunication and slide presentation. As one of the team members noted, both configurations made sure associates were ‘on the same page’. In other words, similar to the establishment of release- and closing dates on the interface, these beyond-interface practices helped to locate ‘who’ was present ‘when’ and ‘where’. In this instance, the boundaries helped to locate associates’ presences ‘in’ work packages, in tasks, and within working files. However, the two sorts of tools operate in different ways. The Gantt chart as an orderly timeline enacts an imaginary of EMMAs development as a linear process, which clarifies the collaborative workflow and can be shown to a wider public, including website visitors and the EC (Alhadeff-Jones, 2017; Latour, 1986). Unplanned and less linear work, however, was only appearing in the software that was hidden from the ‘outside world’ (Star, 1999). Hence, these overlapping topologies are performed by boundaries that separate spaces and times for those producing the platform and those who are seeing the platform (Fig. 6).

On the interface of EMMA, access topologies stretch over multiple webpages that are all connected to the homepage. Hence, the URL to the homepage can be understood to operate as an ‘outer boundary’ or, more precisely, as a first access point or an entrance (cf., Star, 1999). The homepage shows a photo of three young individuals holding digital devices within a sketch of an urban environment, combined with the motto ‘MOOC different!’ (see Fig. 7). Through the text, MOOCs on EMMA are positioned as ‘different’ from ‘others’, a position that enables to form a distinct identity (Srinivasan, 2017). The adjoining photo then relates to an imagined learning audience with particular characteristics (e.g., age, background, taste). This combination of text and image thus establishes a selective imaginary, where EMMA provides entrance to a targeted group of learners.

A second selective imaginary appears through another entrance, namely, registration. Registration, as a requirement to become a learner, orders personal information including names, email addresses, birth dates, spoken languages, geographic locations, and optionally gender and professions. Moreover, it requires consent to policies and conditions, including consent to data collection and conformation to certain behaviors. Here, registration sets preconditions for being ‘in’ EMMA and therewith selects and orders the entrance further (cf., Burridge et al., 2017). Specifically, the platform itself only opens up to the learner if and after the learner is willing to be ‘open’ as visible towards and within the platform (Decuypere, 2018). Together, these imaginaries produce access topologies that select who can access through EMMA, particularly through the performance of boundaries.

A new topology emerges as a series of ‘access points’ after registration. That is, a series of hypertextual images is presented that shows titles of MOOCs, names of providing universities and release dates, and while hovering over also presents teachers’ names and a small introductory text. By granting partial insight into MOOCs, these hypertextual images can be seen as ‘windows’ to different classrooms. Through a click, and subsequently through the ‘enroll’ button, learners can access the MOOC lessons and units that can then be followed through ‘next’ and ‘previous’ buttons. By this means, hyperlinks enact boundaries similar to walls and time schedules in a traditional school, as they guide a sequential path through an entrance, through registration, through hallways, through classrooms and through a series of lectures. The spatiotemporal formation of the relations is therefore one with multiple ordered paths, all coming back to the same initial entrance (Fig. 8).

Contrasting this school imaginary, an overlapping topology produces access through the website in a less sequential order. That is, within each lesson, the EMMA learner is constantly directed to and through different pages designated for particular activities. For example, learning materials and assignments often appear on external websites or software programs and learners need to go to a separate personal page called ‘coursebook’ in the personal learning environment (PLE) to see their notes. While hyperlinks establish connections between these pages, they also establish a separation of different spaces (Dahlberg & Bagga-Gupta, 2017). This separation is enacted through the absence of simultaneous visibility, that is, an overview of the different activities. The combination of connection and separation, here, establishes a fragmented space-time configuration typical to a digital school (Hassan, 2017; Leaton Gray, 2017). The spatiotemporal formation of learning within this topology is characterized by breaks, turns and ‘blind alleys’ which forms relations like a maze (Fig. 9).

Beyond EMMAs interface, access topologies are enacted by a wide infrastructure that generally remains hidden for website visitors. However part of this hidden infrastructure, the webserver made its central role visible in moments its operations stopped or altered (Star, 1999). First, the webserver showed itself when a MOOC with ‘live’ lectures suddenly became popular and it needed to be upgraded in order to generate sufficient space for these large synchronous online gatherings. Later, the webserver’s size was reduced again due to budget cuts, which manifested in longer page load time and it could provide access to fewer visitors. Besides staging the webserver’s operation as a host, i.e., establishing access through EMMA for website visitors, these moments show how the webserver enacts the imaginary of a mutable space as it forms EMMA in different sizes and speeds (Harvey, 2012; Ruppert, 2012). The webserver’s operations therefore do not change the relations that make up spaces and times of EMMA, yet it makes the figure of EMMA appear in a continuously stretching and shrinking form (Fig. 10).

Hidden for website visitors, yet specifically visible to teachers and other EMMA associates, is the access enacted through ‘back-end’ interfaces. EMMAs back-end allows teachers and team members to post their contributions with a personal photo and thus, to establish a presence or a co-presence on the front-end together with learners (see Network topologies). Yet the back-end also embeds a notification system that visualizes questions and comments of learners within a single pane instead of scattered over multiple pages as on the front end. This produces a new topology that not only intersects with networked co-presence, yet equally centralizes learners in order to enact access to learners. More specifically, instead of connecting learners and teachers among each other, the visualizations on the back-end establish a one-to-many relationship where teachers and team members could get access to learners through a centralized pane (Fig. 11).

On the EMMA interface, several practices apparently perform spaces-times of Europe. For example, through an image of the European flag on the bottom of the home page and next to the EMMA logo in video lectures, the EU is presented as a ‘grounding’ and closely affiliated institution of the platform. Moreover, outbound links to proclaimed ‘Related initiatives’ all lead to websites or platforms released under EU support, designating a European field of online education. In this way, EU member countries are not only bound together, but they are also separated from ‘outside’ countries (cf., Büttner, Leopold, Mau, & Posvic, 2015). Specifically, this performs an imaginary of Europe with boundaries around EU territory. Other practices, however, emphasize boundaries within the EU area. For example, the pictogram of a neutral flag combined with a text button ‘Language’ signifies a drop down menu for the selection of nine available languages on the platform (Fig. 12). The combination of this image and text reinforces the association between language and countries or regions (viz., Catalan is included as a regional language), while the selective range of languages additionally shows how the platform includes as well as excludes specific regions in the European imaginary (Opitz & Tellmann, 2015). This establishes an ambiguous spatiotemporal formation that marks a difference between what it means to be ‘in’ Europe, ‘in’ a European field of online education and ‘on’ EMMA, making it possible to be included and at the same time excluded through multiple boundary systems (Fig. 13).

European topologies are further performed on the EMMA interface by specific operations of the integrated translation system. This system operates automatically by using learners’ IP-addresses to set the default platform language to the region they are situated in. In addition, it allows learners to select languages manually through the abovementioned drop-down menus and through dropdown menus in each MOOC. However, the system does not translate learners’ comments and replies, and merely shows them in the original language they were posted in. As a consequence, interactions between learners mainly occur within a single language, generally their native language. This translation system, thus, seems to favor and aid interactions within linguistic boundaries that, as they closely connect to familiar languages and regional boundaries, enact learners’ home country. In contrast to EMMAs objective to traverse European boundaries, the particular workings of the translation system therefore seems to form these boundaries. However, as most MOOCs and some conversations are at least equally available in English, learners can access foreign content or interact through English. Still, since it is impossible to simultaneously see translations to other foreign languages, English does not work as a ‘bridge language’ that grants access to content in other local languages (Harvey, 2012). Instead, English enacts a boundary as a space in between that connects the linguistically bounded spaces and brings them closer together (Prince, 2015; see Fig. 14).

Besides the translations, the course materials themselves appear to perform European topologies on the interface. That is, each MOOC articulates the regional origin of the providing university in some specific way, for instance, through cultural examples or non-translated remainders in reading materials. Other examples of boundaries are presented in MOOC assignments where learners were invited to visit a natural setting in a specific region or a website in a regional language. Learners, however, challenged these boundaries as they showed (to be from) places ‘outside’ the addressed regions in the virtual classroom (see Fig. 15). Thus learners are able to slip through the boundaries and to bring spaces ‘inside’ EMMA that were initially excluded from the European imaginary (Burridge et al., 2017; Mol & Law, 1994). The spatiotemporal formation of the course materials is therewith characterized by delimited areas, yet with porous outside borders (Fig. 16).

European practices beyond the interface emerge out of relations between the EMMA team and its associates. While they largely overlap with network topologies, these topologies are specifically European through new relations. That is, the European nature of these topologies arises out of the relation between the EC, EMMA and other European as well as non-European platforms. The EC formed these relations through practices of competition and comparison specifically because of its ambition to develop platforms that can challenge ‘foreign’ MOOC providers (European Commission, 2013). Initially, through competition, the EMMA team was urged to focus on a distinct learning audience and to offer something different and ‘better’ than existing non-European platforms. Yet through comparison, EMMA was evaluated along similar indicators as these other platforms, inviting convergence rather than distinction (e.g., Hartong, 2018). Given these contradictory logics, the EMMA team constantly negotiated between practices that are generally associated with European education and practices of privatization and commodification more specific to North-American MOOC platforms (Van Dijck & Poell, 2015). For example, they had to balance decisions on the use of multiple European languages or only English text, and publicly funded admittance or efficient business models. Through these comparative and competitive practices, the topology forms a conflated Europeanness that merges values of non-European platforms with a typical European space (Fig. 17).

Another European topology comes forward in relations between the EMMA team, the providing universities and partners, thus excluding the EC. This topology operates around a particular structure that connects these associates over their geographical, cultural and linguistic localities and makes them act as a collective (Opitz & Tellmann, 2015). Central in the practices is the English language which, although appearing on the interface as well, has a specific operation in practices beyond the interface. That is, since the translation system relies on human editing, each provider was urged to translate their learning content particularly in English so that other associates could also develop translations to other languages, i.e., their native language. These English translations therewith enact more than just a space in between, yet rather act as a shared space constructed by and from the different local practices (cf., Prince, 2015; Star, 2010). With respect to the spatiotemporal formation, this means that English enacts a boundary where different localities can meet and interact (Fig. 18).

On the interface, data topologies especially appear to intersect with other sorts of topologies. One example of such an intersection is visible in the registration procedure which, as previously noted, selects access through EMMA. However, it equally concerns data practices as it generates personal accounts with records of learners’ clicks and translates them into personal data visualizations. For example, the PLE shows learners visualizations of vertical timelines with their latest activities, ring charts on how many units they visited and how many quiz answers were correct and text bars on the amount comments and replies. These visualizations perform an imaginary of self-tracing learners who scrutinize their own actions. Positioning their previous steps like passed achievements on these lines or charts, these visualizations equally present and perform the ‘future steps’ needed to complete a learning trajectory (Thompson & Cook, 2017). This is yet another way in which data topologies interweave with access topologies, as these visualizations sketch a map of ‘where’ and ‘when’ learners need to go. However, rather than a map that promotes co-presence, this map positions access through EMMA as an individually adjusted path.

Besides relating learners to their own actions on EMMA, data practices on the interface equally relate learners to each other in two specific ways. First, the PLE presents aggregate statistics of other learners in the virtual classroom underneath individual statistics, specifically on the same indicators (see Fig. 19). By presenting these statistics together, a topology emerges that brings learners together in a comparative relation. Even without an additional text, this establishes an imaginary of the outcomes that are valued and that one can compete in (Decuypere, 2016). More specifically, it establishes an imaginary of learners who aim to complete more lessons, to get higher quiz grades and post more comments and replies. In a second way, learners are related through data that are integrated into the network visualization presented in the PLE (Fig. 20). This visualization shows learners as nodes in a network and their interactions in the form of line. However, by leaving out other complex details of their learning experience, the data install a particular imaginary on what it means to be in relation to each other (see also Marres, 2012). Here, relations within the EMMA classroom are evaluated by the amount of interactions and the positions of the individual learners in the web of relations rather than what to communicate about and how to maintain relations. By relating learners to each other in this specific way, although counterintuitive, these visualizations establish individualizing spaces (Thompson & Cook, 2015, 2017; Williamson, 2017). That is, the visualizations make learners individuals as they present their connections with other learners as a way to mirror themselves, rather than to see themselves as integrative part of a learning audience (Fig. 21). Through these visualizations, the data operate as a fold, a material arrangement that produces and reproduces an imaginary of learners and their learning practices by doubling or mirroring them (Deleuze, 1986). This topology does not just operate on the screen of users, but is intricately interwoven with practices beyond the interface.

Data practices beyond the interface equally operate by visualizing learners, yet specifically by showing learners and their actions to the EMMA team and her associates while they are on the other side of the interface. That is, the data enact visible indicators for the EMMA team and teachers to interpret learners’ behavior on the platform. For example, data show how frequently learners visit particular MOOCs, what kinds of interactions improve their retention or how many minutes they watch a video or read a text before quitting it. This means that data materialize traceable learners whose actions are not only visible to themselves, but also to others (Ruppert, Law, & Savage, 2013). These traceable learners are composed of data that can be documented through learners’ clicks and through survey questions, while it excludes their gazes on the screen or off-line practices like note-taking, scheduling or multi-tasking. Thus again, an imaginary of learners is established through a selection of indicators, performing simplified norms regarding what is important for learning on EMMA.

This data topology, the one that makes up an imaginary of learners based on data, is of particular importance for the spatiotemporal configurations of the platform. Namely, it is this imaginary that appears to bind together the different European- network- and access topologies. This ‘binding together’ starts with an interaction between data- and European topologies, as data are initially used as a visible passageway to successively inform, transform and again inform the EMMA team and the EC on specific outcomes. They mainly show achievements on the amount of learners on the platform and their retention, aligning with the interest of the EC in expanding and intensifying open and online learning in European countries (European Commission, 2013). Yet data also showed the EMMA team what interaction pages retained the largest learning communities and what hyperlinks eased access through the platform. In turn, this elicited continuous reconfigurations of network and access practices based on what data showed about the best ‘fit’ between learners’ retention and the platform design. This has particular implications for the workings of openness and boundaries in EMMA. First, as the configuration of EMMA focuses on retaining existing learners rather than inviting new learners, boundaries around EMMA do not prioritize to ‘open up’ to learners who are not on the platform yet. Second, learners are reinforced in their behavior rather than invited to change. For example, they are not impelled to prolong their attention span when watching lectures or reflect on the content of their interactions with peers. It implies that data establish a platform that operates as a mold around learners and their behavior. While this mold is enacted through practices beyond the interface, it interacts with the fold topology which forms learners by mirroring them in data visualizations. Together, these interwoven topologies establish a spatiotemporal formation in which a form of the learner is established and stabilized in a ‘(back)ground’ that fits around her or him (Fig. 22). What is specific to data topologies is, then, that they produce the boundaries of different network, access and European topologies by continuously interconnecting (with) them. As we will equally argue in the conclusion, openness can therewith be seen as established in multiplexes of spaces-times, that is, in multiple interconnected forms. This shows how boundaries do not merely imply closures, yet also specific forms of interactions.