Notes on Williamson, B., Gulson, K. N., Perrotta, C., & Witzenberger, K. (2022). Amazon and the new global connective architectures of education governance. Harvard Educational Review, 92(2), 231-256.
This path-breaking paper opens up a new agenda for critical studies of digital education which shakes off the comforting bromides which too often characterise critical approaches in order to designate the underlying challenge with a heretofore unprecedented analytical specificity: the infrastructural dominance which large tech conglomerates are close to achieving in post-pandemic education. This is something I talked about with one of the authors in an online seminar earlier this year, as well as my collaborators Janja Komljenovic and Susan Robertson with whom I spent the years before the pandemic exploring the idea of the platform university. This is the podcast from the event:
Infrastructural dominance. It’s an incredibly helpful way of designating an emerging characteristic of the university as an institution, in which the campus has been decentred and former constraints upon ‘digital innovation’ have been worn away by the frantic pivots of the pandemic. The paper has a much wider focus on “the emergence of a vast range of ‘data-driven’ technologies for use in educational policy and practice settings” and how they intersect with the political economy generated by “an increasing emphasis on educational performance data over the ensuing decades [which] has involved the production of ever-more complex information systems”. The technical capacities of these systems “afford novel capacities for measuring, monitoring and managing education systems, institutions, staff and students” but the byproduct of these capacities is that corporate operators take on a newfound role within the operations of educational systems at primary, secondary and tertiary level. They locate this as part of a wider process of networked governance in which businesses, philanthropies and international organisations take on a more prominent role within education. The authors suggest four implications of this:
- “increasing private sector power to define the future of education and its public values”
- “a realignment of education to match the purported demands of the digital economy”
- “algorithmic control of pedagogic and curricular processes”
- “delegation of political authority in education to automated decision-making
They suggest the inflection point for institutional change rests in the growing dependence of public institutions on private technology provides, from user-facing digital platforms (Coursera, Google Classroom, Class Dojo etc) through to cloud computing infrastructures (Azure, Google Cloud, AWS etc) which service the backend operations of educational institutions, as well as those of the digital platforms themselves. These digital platforms and cloud computing infrastructures intersect higher up the stack through APIs such as Google single-sign on which incorporates a vast range of services into an ecosystem controlled by the giant by making interoperability and data sharing close to seamless for the end user and possible for the developer. These elements combine in what they term connective architectures which knit together what might otherwise have been discrete socio-technical elements into an overarching infrastructure which is much more than the sum of its parts.
The thrust of the paper analyses what connective architectures mean for educational governance i.e. what does it mean for the “authoritative processes and institutions that are responsible for decision-making within formal structures of education systems”? The indispensability of platform architectures to data-driven education governance means that decisions are made across a wider range of settings incorporating a wider range of actors, including those who still not be seen as obviously educational in their orientation.
The question this raises for me is what is the relationship between the infrastructural dominance which characterises the growing dependence upon emerging connective architectures within education systems and the capacity for agency of those working and studying within them? They explore these questions through a case study of Amazon Web Services (AWS) in which they identify five “operations of governance” enacted by AWS within education :
- Inscribing particular business strategies in the education sector
- Habituating educational users to AWS technologies, particularly with regards to programming
- Establishing interfaces between educational institutions and AWS cloud architecture through APIs
- Providing a platform for third party developers to create new online services
- Displacing the existing information infrastructures within education through dependence upon AWS
They argue that this delegation of public functions to private actors can be seen far beyond education, in a process described by Fourcade and Gordon as ‘cyberdelegation’ in which public institutions lacking the internal capacities for digitalised operations outsource these operations in whole or part to the private sector. The cloud systems of Microsoft, Google and Amazon can be seen throughout public sectors across the world with the ‘operations of governance’ described above seeming from my perspective to be primarily lock-in mechanisms which increase the opportunity costs of reversing infrastructural dominance within an institution. The unfortunate flip side to this is that infrastructural dominance makes things easier within the confines of the connective architecture, ranging from outreach to development communities, provision of technical support and the ecosystem of services which dependence makes possible.
I thought it was interesting that I read these mechanisms so differently while entirely agreeing with their characterisation. I suspect this reflects a divergence in how we draw upon platform studies, with the authors explaining how they see platform studies as working “to reveal their constitutive components; to surface the social, economic and technical processes involved in their production; and to trace their consequent effects”. It is a decompositional approach to platforms which breaks them down into their social and technical elements in order to understand the relations between them and the effects they are capable of producing as a consequence of those relations.
However there is another tradition within platform studies (associated with people like Jean Burgess, Nancy Baym, Joshua Green and to a certain extent the earlier work of Jose Van Dijck) which stresses how platforms are co-produced with their users through the evolution of user cultures which are a crucial feature of the business environment which firms have to respond to, at the level of marketing & communications but also the development of the platform over time. This is less a feature of back-end architectures (though consider the vast scale of developer engagement from international conferences through to asynchronous forms of engagement and local meet ups*) but it is crucial to user-facing platforms, involving specialised and business sensitive relationships with a diverse range of user groups.
This is the agent-orientated reality which I think tends to get lost within platform studies, as well as with governance approaches focused on powerful actors, emerging networking and the diffusion of strategies. Each of these analytical approaches risks falling into seeing non-powerful actors as objects of change rather than potential subjects of change, even if this capacity is often only realisable at best in collective action. We should take immense conceptual, methodological and empirical care to not inflate the influence which lay actors have over the social system while recognising the capacity for action which remains, even if its outgrowths might be limited in scope and scale. Furthermore it’s important we recognise how platform infrastructures lock in certain modes of engagement, in ways as mundane as personal habits during a commute or how one organises one’s own information ecosystem (or fails to).
My work on social media use by academics has been preoccupied by the question of how we establish user cultures which maximise our capacity to develop scholarly cultures which collectively support our use of social platforms for shared purposes. Or to put it another way how to minimise the extent to which we are locked in as individual users by the attentional logics which characterise social media leading us to use them in non-purposive ways. Since I decided to move into Education and make this the primary focus of my research, I’ve been trying to expand the scope of this question to ask how we can support the purposive use of technology more broadly and how the infrastructural dominance of corporate providers poses significant challenges to this undertaking. I see this as a matter of individual practice, networked collaboration, the public pedagogies which emerge on platforms, institutional policy making and collective action through bodies such as trade unions and learned societies.
These are all things I’ve been involved in over the last ten years and the focus of my work now is on how to theorise their potential outgrowths against the background of infrastructural dominance which this enormously thought-provoking and conceptually robust paper expertly lays out. Even if I have some theoretical disagreements with the focus of the questions, I’m entirely persuaded by the answers given. I’m simply reiterating the perennial question (what is to be done?) as well as suggesting the role we assign to agency at the level of social ontology impacts upon the extent to which our diagnoses can facilitate courses of action which might variously circumvent, mitigating and resist the tendencies which we all agree are immensely concerning. I’ve spent a lot of time thinking recently about the internal dynamics within universities with regards to the infrastructural shifts underway in the post-pandemic university:
Valuable information and knowledge formerly contained by in-house IT systems vanishes inside external, black-boxed cloud systems. Institutions’ abilities to audit or implement privacy, ensure data protection compliance, or obtain meaningful informed consent for data collection and use are limited and constrained when they employ cloud operators. Valuable information and knowledge formerly contained by in-house IT systems vanishes inside external, black-boxed cloud systems. Institutions’ abilities to audit or implement privacy, ensure data protection compliance, or obtain meaningful informed consent for data collection and use are limited and constrained when the employ cloud operators.
Given how politically contentious the role of technology in pedagogy has become within the post-pandemic environment, the exercise of influence over their remote working and online learning capacities is more important than ever to universities. This cuts across different levels of operation from university management adapting to a volatile political environment, preparation for future pandemics and climatic events which will bring future ‘online pivots’, the everyday pedagogical interaction between teachers and students, the control which academics have over their intellectual property when teaching is recorded by default etc.
There is so much at stake here for groups across the university, in the sense of being potential matters of concern due to their expansive material implications. I’m interested in how we might realise that potential contentiousness and highlight the capacities these groups have to “throw their grain of sand into the well-oiled machinery of resigned complicities” (as Bourdieu once put it) as well as to take more organised action, including the building of networks and coalitions from below which have yet to exist, through the machinery available to us. The duality of platforms is that they also provide affordances through which this collaborative and coalitional work can take place, at much more varied scales and speeds than legacy modes of collaboration enabled.
In this sense I’m interested in how we translate analysis into action and the role that ontological questions concerning agency play in such an undertaking by influencing what we emphasise and fail to in the ensuing accounts. Work in critical digital education is clearly diagnostic in its orientation and we need to think methodologically and politically about what that means for every aspect of our practice.
*This is covered on pg 10 of the paper but it reduces the social complexity of international interaction between hundreds of thousands of people into the primary strategic goal which Amazon has for organising AWS Education. This is the point where the microsociologist in me objects because there’s clearly so much more going on here which is salient for the analysis of AWS’s influence as well as social life more broadly.