Notes for a talk at the online webinar ‘Going Online During and After the Pandemic’ on 6 May 2020. Thanks to Mariya Ivancheva and Brian Garvey for the invitation.
Over the last five years or so, data analytics and automated processes have become increasingly prevalent in higher education. Many functions, practices and tasks of HE are being made ‘machine-readable’ by digital technologies, as teaching, learning and administration are transformed by ‘datafication’. Datafication is also advancing automation, as the processing of digital information by increasingly capable or ‘smart’ technologies makes it possible to automate many tasks. As universities have now begun to consider a future disrupted by Covid-19 that seems likely to be characterized by distance learning, datafication and automation may expand and intensify across the sector.
Writing in The Guardian this week on online teaching and the future of HE during and after the Covid-19 pandemic, former universities minister Chris Skidmore argued ‘universities need to track student progress remotely’ in order to ensure they are not at risk of disengaging or dropping out, and singled out examples of institutions that have already developed high levels of capacity for on-campus student tracking. The current crisis is already escalating high-level demands for universities to monitor and track students more intensively.
This work-in-progress post offers some initial, provisional observations about how such technologies and practices of datafication and automation are being mobilized in HE in the context of Covid-19 (following up some previous posts focused mainly on the schools sector). The aim is to encourage HE educators, researchers, administrators and students to think critically about the existing extent of datafication and automation and its potential reach and consequences in coming months and years.
Datafication of HE
The datafication and automation of HE is powered by three main forces. First, at a broad societal and cultural level, there is now widespread desire for quantification and measurement made possible by digital data processing systems. We can see this, for example, in consumer acceptance of wearable biometric technologies (FitBits etc), in daily displays of epidemiological data by politicians and science experts, and public discourses about Covid-19 case counts. Although deeply contested, digitally-generated numbers and visualizations have a powerful allure. This is the case in HE too, as historically evident in rankings and league tables as ‘authoritative’ sources of numerical knowledge, and increasingly evident today in widespread uptake of data dashboards, platforms, and analytics.
Second, higher education policy has become increasingly focused on performance monitoring, competition and marketization over the past decade. The Office for Students’ emphasis on data-led market regulation in England is a key manifestation of this political desire for control through quantification of the sector. QAA Scotland has led an Enhancement Theme exploring the use of learning analytics and other digital forms of data for enhancement of the student experience and outcomes. Organizations including the government departments of education and business, sector agencies HESA, Jisc and QAA, the think tanks PolicyConnect, Nesta and HEPI, the consultancies KPMG, McKinsey’s and Deloitte, plus a range of business actors have all promoted the use of data analytic technologies for enhanced performance monitoring and improvement in HE.
And third, a ‘global education industry’ of data solutions services, infrastructures, platforms and app providers has sought to make markets for their products in HE—covering everything from recruitment and admissions, through learning management, student information and library resource systems, cheat detection and ‘contract cheating’ identification, online assessment and exam ‘proctoring’, to employment matching, graduate tracing and talent analytics. This is a highly lucrative space in which for-profit organizations may generate venture capital investment and capture institutional customers looking to boost reputational advantage, improve recruitment, and enhance measurable outcomes such as student experience surveys, grades, and graduate employability.
Together, widespread desire for quantification, HE marketization, and the global education industry have resulted in a proliferation of data analytic technologies across the sector, which we see now intensifying and expanding in response to the coronavirus pandemic.
Digital infrastructures of teaching and learning
Two particular technologies of datafication and automation have come into sharper view during the Covid-19 pandemic—learning management systems and online degree programs. These are not ‘spectacular’ technologies of datafication and automation–compared to hyperbole about artificial intelligence–but their mundane status as parts of universities’ infrastructures for teaching and learning positions them to become especially consequential in any recovery or reorganization of institutions around distance education.
Learning management systems are the backbone of HE courses across the planet and the companies running them have gathered extensive global education data sets—in some cases, data about millions of students combined. With hundreds of providers—key ones in the UK being Blackboard, Moodle and Canvas—and a total LMS market said to be valued in the tens of billions, providers differentiate themselves from their competition through restless upgrades and new feature designs.
A key aspect of LMSs is the capacity for the data to be subjected to learning analytics, often through in-built analytics or third party integrations. Solutionpath is one of the leading learning analytics providers. Its Student Retention, Engagement, Attainment and Monitoring platform (StREAM) automatically generates a near real-time ‘engagement score’ for each individual enrolled at participating institutions. It does this by collecting data from a range of ‘electronic proxies’ that represent students’ participation in their course, including LMS data, building access card swipes, software logins, library loans, attendance, and assignment submissions. These data are combined then automatically analysed and presented on dashboards. If patterns in student behaviour change over time, alerts are triggered within the StREAM platform to facilitate staff intervention
Recent interest has developed in using LMS stores of big student data for automated recommender services. The company Instructure behind the LMS Canvas, for example, was just acquired by the private equity firm Thoma Brava for $2bn, based partly on its proposal to use its global student datasets for personalized learning recommendations—a program it initially code-named DIG. As Instructure’s chief executive announced in an investor meeting last year,
We have the most comprehensive database on the educational experience on the globe. So given that information that we have, no one else has those data assets at their fingertips to be able to develop those algorithms and predictive models. … [W]e can take that information, correlate it across all sorts of universities, curricula, etc, and we can start making recommendations and suggestions to the student or instructor in how they can be more successful. … Our DIG initiative, it is first and foremost a platform for ML and AI, and we will deliver and monetize it by offering different functional domains of predictive algorithms and insights.
Competition in the LMS field, and the level of interest from investors and customers alike, is catalysing this shift to algorithmic forms of insight, prediction, and recommendation—a case of automation creeping into the pedagogic encounter between educators and students.
The development of LMSs to adopt recommender and relevancy algorithms is reflected in the newer category of Learning Experience Platform (LXP). The key feature of an LXP is that it is designed to automate the ‘intelligent discovery’ and ‘recommendation’ of relevant learning content. Whereas conventional LMSs are based on searchable course catalogues, an LXP is organized more like YouTube or Netflix as a content management platform with in-built recommendation technologies. An LXP collects continuous data from learners’ behaviour, learning and performance in order to perform these analytics processes.
The AULA LXP, for example, used by multiple HE providers across the UK, presents itself as a ‘digital campus’ platform that partners with academics ‘to design high quality learning experiences’, which it terms ‘Dream Courses’, and to ‘scaffold the shift to blended and fully online’ teaching. Its ‘LMS Data Importer’ automates the migration of all content and information from other legacy systems, and the platform features an Engagement API for monitoring real-time student engagement which offers personalized, targeted recommendations to educators to improve it. AULA has partnered with WonkHE, the HE news, opinion and analysis organization, on an online conference about higher education responses to COVID closures.
So LMS providers are no longer just digital backbone or infrastructure to university courses, but increasingly active partners in pedagogic processes. They act as providers of online learning scaffolding, as ‘recommendation engines’ for AI-enhanced ‘personalized learning’, and ‘digital campus’ or ‘dream course’ developers, utilizing their extensive and continuously updated data sets for teaching innovation, institutional outcomes enhancement, and measurable performance improvement. LMS platforms are likely to become even more central to HE teaching with the shift to online degree provision.
Online program management (OPM) refers to infrastructure services provided by vendors to enable universities to deliver online and distance education courses. Currently growing rapidly in the US and UK, OPM service providers also provide extensive data analytics in their platforms, offering convenient ways to automate student tracking and monitoring. OPM companies include 2U, Noodle Partners and Academic Partnerships, big education publishers, including Wiley and Pearson, as well as MOOC providers that have diversified into the OPM market (Coursera, FutureLearn).
One of the most successful providers, 2U, provides the OPM platform 2UOS (2U Operating System). 2UOS consists of an online teaching and learning platform, a suite of data analytics for generating information about students, technical support, and targeted, program-specific digital marketing campaigns using machine learning and AI. It has carefully presented itself as a key technology for universities to transition to online teaching during the Covid-19 crisis. OPM providers have positioned themselves to support institutions’ internationalization strategies, as universities seek out a share of the international student market, but now find themselves in position to support the transition online for both international and domestic students too.
A key aspect of the success of OPMs is that the companies usually cover the up-front costs of setting up an online degree program, and provide the technical infrastructure for university partners to build their courses on. This model saves universities having to front the costs or building the technical platform. The companies then take 50-60% of the student fees as a return on their up-front investment.
As with LXPs partnering with institutions on ‘Digital Campus’ or ‘Dream Course’ development, then, OPM providers present themselves as private pedagogic platform intermediaries. They are, increasingly, situated between the enrolled student and academic staff on a given course. This, of course, may become especially the case as students almost exclusively study online or through hybrid models.
The rush to remote education and online degrees is now a significant market event. OPMs, claims the education market intelligence consultancy HolonIQ, constitute part of a $7billion ‘Global OPM and Academic Public Private Partnership Market’ that ‘COVID-19 will substantially accelerate’ to a $15b market by 2025:
More so today (COVID-19) than ever, Universities around the world are increasingly seeking private partners to rapidly build capability, to boost and differentiate their offerings, accelerate growth and achieve long-term sustainability. As such, Private Equity and Capital Markets are watching the Academic PPP segment closely.
Moreover, OPMs are a key growth technology in a much larger Global Online Higher Education market valued by HolonIQ at $36billion in 2019 and projected to rise to $74b by 2025, opening up new ‘opportunities’ for market providers:
These changes to market dynamics are likely to accelerate with COVID-19, and while the biggest online players are gaining market share on the strength of their national reach and brands, this is the opportunity for predominantly offline providers to amplify their current online offerings with existing and new learners.
Online program management platforms, along with learning management systems, are key parts of an education technology sector that in the first three months of 2020 alone, according to the HolonIQ, ‘delivered $3billion of Global Edtech Venture Capital’.
These private platforms, powered by venture capital, private equity and student fees, are now doubly empowered in the multitrillion dollar global education market as universities rapidly transition to online teaching. In the language of finance, Q1 of 2020 produced a remarkable boom in edtech pandemic markets.
But this quarter of financial activity may have long-lasting consequences and implications for higher education much further into the future. As a provisional list, these include:
- Deferment of expertise to platform companies and the increasingly capable systems they are inserting into higher education programs and practices. Datafication and automation have been legitimized by policy actors, consultancies, edu-businesses and think tanks as ways to improve HE. These developments encourage the delegation of judgement to automated systems, as decisions normally taken by workers are deferred to advanced analytics and automation. They also reshape pedagogies and curriculum design to fit the digital templates and forms of teaching made possible by the platforms, with providers themselves increasingly involved in designing ‘Dream Courses’ and online degree programs.
- Fusion of higher education to the model and business of platforms. Platforms depends on the extraction and analysis of data as a route to profit. In the ‘platform university‘, student data are not just used for performance measurement of the university, but as a source of valuation for private edtech companies. Private edtech is now thriving on the platform ‘network effects’ of increasing numbers of institutions emulating one another to adopt public-private partnership arrangements with platform vendors. In Platform Capitalism, Nick Srnicek argues digital platforms have become key infrastructures of society; we might add that edtech platforms have now become infrastructural to higher education in the Covid context, with potential for long-term lock-in effects.
- Austerity for universities and profitable market-making for platform companies. The shift to online education through platforms such as LMS integrations and OPMs raises risk of further worker precarity in universities in contrast to soaring private investment and customer spending on new platforms. Laura Czierniewicz compellingly notes that ‘underfunding and financial cuts which drive up the risks of sectoral fragmentation and breakdown’ are now paralleled by marketisation and ‘the increasingly unfettered infiltration of big corporate forces substantially reshaping higher education’.
- New analytic engines of real-time anxiety. If rankings and league tables, as Wendy Espeland and Michael Sauder have argued, are ‘engines of anxiety’ driving ‘reactive’ behaviours in HE—where people perform to satisfy a given measurement rather than out of collective mission or shared values—then new forms of datafication and automation may also generate new anxieties. Liz Morrish has argued that demands have increased on the academic workforce over concern about university rankings and league tables, creating ‘a culture of workplace surveillance’ in universities. Digitally-enabled datafication could exacerbate these pressures as it potentially introduces ‘real-time’ performance measurement into working spaces including university offices and classrooms, with increased surveillance of both students and staff a very concerning possibility after the pandemic.
- New forms of algorithmic governance in HE. Algorithmic governance, as Christian Katzenbach and Lena Ulbricht define it, refers to ‘algorithms as a form of government purposefully employed to regulate social contexts and alter the behaviour of individuals, for example in the treatment of citizens or the management of workers’. Algorithmic governance involves the creation of detailed profiles about individuals (‘data doubles’) leading to potential for ‘social sorting, discrimination, state oppression and the manipulation of consumers and citizens’. As algorithmic and analytic processes become increasingly automated and ‘out of control’, they lead to forms of ‘automated management’ where humans have diminishing oversight over the kinds of decisions made by them.
Together, these developments suggest the expansion of the model of the datafied university that is increasingly governed by algorithmic, data analytic, and automated systems and in which the roles of staff and students may be redefined. Staff and students in HE will, in coming months, need to scrutinize the technologies of the datafied, automated university further, and come up with collective responses to help (re)build the HE institutions of the future.