Content uploaded by Ben Williamson
Author content
All content in this area was uploaded by Ben Williamson on Jun 07, 2023
Content may be subject to copyright.
1
INTRODUCTION
Learning machines, digital data and
the future of education
01_Williamson_Ch-01.indd 1 7/4/2017 4:56:39 PM
2 Big Data in Education
In October 2015, over 1,000 young software developers and hackers attended
HackingEDU, a three-day educational hackathon held at the San Mateo Event
Center in San Francisco. Originally launched at the 2014 Google Summit, the
annual HackingEDU event – the ‘world’s largest educational hackathon’ – is
intended to help software developers and programmers, most of them college
students, ‘revolutionize the education industry’ while competing for over
US$100,000 in prizes (Hunckler, 2015). Featuring expert workshops, panel discus-
sions and guest speakers, HackingEDU 2015 was supported by major technology
companies including IBM, Google, Uber, PayPal and Automattic, as well as by
successful educational technology businesses such as Chegg and EdModo. It
emphasized the ways in which technologies might be used to ‘disrupt’ and ‘revo-
lutionize’ education, much as ‘Uber revolutionized the transportation industry
based on a simple concept: press a button, get a ride’, as the event’s partnership
director phrased it (Uber 2015). The technology projects produced during
HackingEDU 2015 included titles such as Learnization, CereBro, PocketHelp,
QuizPrep, BrainWars and StudyTracker, almost all of them relying on a combina-
tion of digital data and database technologies and constructed by their young
designers using a variety of programming languages, software programs and
hardware devices.
Elsewhere in San Francisco, many other fledgling edtech projects are annually
developed through the support of edtech ‘incubator’ or ‘accelerator’ programs.
Incubators typically help entrepreneurs and new startups to test and validate
ideas, while accelerators turn products into scalable businesses, often through
direct equity investment, and help provide entrepreneurs with legal, IT and
financial services along with mentorship, working space and access to educators,
entrepreneurs, business partners and potential investors (Gomes 2015). For
example, Imagine K12 is ‘a startup accelerator focused on education technology’:
Our goal is to improve your company’s chances of success. We do this through a
combination of strategic advice and mentorship, a series of speakers and seminars
designed to help founders make better decisions, value-added networks of entre-
preneurs and educators, and $100,000 of initial funding. … Companies begin
receiving support from Imagine K12 immediately upon their acceptance, including
$20k of funding. … [A]ll accepted startups are required to move to Silicon Valley for
an intensive four-month program. (Imagine K12 2015)
Edtech incubator and accelerator programs like Imagine K12 provide the space,
support and investment required for programmers to write educational tech-
nologies, and ultimately act as mechanisms that might realize the ‘revolutionary’
ambitions of entrants to competitions like HackingEDU. Notably, Imagine K12
has since merged with another accelerator program, Y Combinator, an organi-
zation established by billionaire PayPal founder Peter Thiel, a major donor
and spokesperson during Donald Trump’s US presidential campaign in 2016.
01_Williamson_Ch-01.indd 2 7/4/2017 4:56:39 PM
Introduction 3
A key educational technology advocate, Thiel has supported and funded many
companies and startups that focus on ‘revolutionizing’ education through data-
driven software applications (Levy 2016). For new startups that successfully
graduate from the incubation and acceleration stage, entrepreneurial investors
from Silicon Valley have been funding educational technology projects with
unprecedented financial enthusiasm since about 2010 (EdSurge 2016). With
webs of political support and entrepreneurial investment for educational tech-
nology growing, a new digital future for education is being imagined and
pursued in governmental and private sector settings alike, with significant con-
sequences for learning, policy and practice.
HackingEDU is an important event with which to start this book for a number
of reasons. It locates education as it currently exists as a problematically broken
system which is in need of revolutionizing. It proposes that the solution is in the
hands of software developers and hackers who can write code. It suggests that
the availability of masses of educational data can be used to gain insights into
the problems of education, and to find solutions at the same time. And it also
demonstrates how private sector technology companies have begun to fixate on
education and their own role in fixing it. Incubators and accelerators such as
Imagine K12 and Y Combinator can then step in with entrepreneurial experi-
ence to grow new products into successful startup businesses, to enable
programmers to fine-tune the code and algorithms required to make their prod-
uct run, and to gain financial investment required to push it out into practice.
The promise appears simple. Take a model like Uber, the mobile app that has
transformed taxi services by harvesting locational data from its millions of users,
and then translate that model into a template for educational reform. Fund,
incubate and accelerate it until it performs optimally. All it takes to revolutionize
education for the future is a few million lines of software code and big piles of
digital data.
Digitizing and Datafying Education
The goal of this book is to understand and detail how digital data and the code
and algorithms that constitute software are mixing with particular political
agendas, commercial interests, entrepreneurial ambitions, philanthropic goals,
forms of scientific expertise, and professional knowledge to create new ways of
understanding, imagining and intervening in education. Education is now a key
site in which big data and algorithmic techniques of data mining and analysis
performed with software are proliferating and gaining credibility.
Yet the quantitative increase in data brought about by recent developments
and the qualitative effects they are beginning to exert in education have gone
largely unnoticed amid much more high-profile concerns about the data mining
01_Williamson_Ch-01.indd 3 7/4/2017 4:56:39 PM
4 Big Data in Education
conducted by social media companies on their users, targeted online advertising
that is driven by consumer data, or the data-based forms of surveillance being
practised by governments (van Dijck 2013). A ‘new apparatus of measurement
has drastically expanded’ with the availability of digital data in diverse areas of
public and private life, ‘allied with a set of cultural changes in which the pursuit
of measurement is seen to be highly desirable’ (Beer 2016a: 3). Education, by
contrast, appears more ‘ordinary’:
Given that so much attention has already been paid to social media corporations
and governmental and security agencies, what we now need to attend to is
other, more ordinary actors, as social media data mining becomes ordinary.
(Kennedy 2016: 7)
This book takes up the challenge of investigating the digital data technologies,
organizations and practices that are increasingly becoming integrated into many
aspects of education. A vast apparatus of measurement is being developed to
underpin national educational systems, institutions and the actions of the indi-
viduals who occupy them.
While the pursuit of educational measurement has a long history stretching
back to the nineteenth century (Lawn 2013), it is being extended in scope,
enhanced in its fidelity, and accelerated in pace at the present time as new tech-
nologies of big data collection, analysis and feedback are developed and diffused
throughout the system (Beneito-Montagut 2017; Selwyn 2015). Similarly,
schools, colleges and universities have employed e-learning programs for many
years in their pedagogic and instructional processes (Selwyn 2011), but with big
data and analytics processes now increasingly augmenting them, these resources
can now adapt to their users and ‘talk back’ to educators (Mayer-Schönberger
and Cukier 2014). Software and digital data are becoming integral to the ways in
which educational institutions are managed, how educators’ practices are per-
formed, how educational policies are made, how teaching and learning are
experienced, and how educational research is conducted.
The presence of digital data and software in education is being amplified
through massive financial and political investment in educational technologies,
as well as huge growth in data collection and analysis in policymaking practices,
extension of performance measurement technologies in the management of edu-
cational institutions, and rapid expansion of digital methodologies in educational
research. To a significant extent, many of the ways in which classrooms function,
educational policy departments and leaders make decisions, and researchers
make sense of data, simply would not happen as currently intended without the
presence of software code and the digital data processing programs it enacts.
To fully appreciate how digital data are being generated and exerting material
effects in education, then, it is essential to view data and the software code and
algorithms that process it in relation to a range of other factors that frame their use.
01_Williamson_Ch-01.indd 4 7/4/2017 4:56:39 PM
Introduction 5
Political agendas relating to education policy and governance, commercial inter-
ests in the educational technology market, philanthropic and charitable goals
around supporting alternative pedagogic approaches, emerging forms of scien-
tific expertise such as that of psychology, biology and neuroscience, as well as
the practical knowledge of educator professionals, all combine with new kinds
of data practices and digital technologies. That is, the mobilization of digital
data in education happens in relation to diverse practices, ways of thinking,
ambitions, objectives and aspirations that all shape how data is put to use,
define the tasks and projects through which data is deployed, and co-determine
the results of any form of educational data analysis. The role and consequences
of digital data in education cannot be understood without appreciating their
relations with the other ordinary features of education – policies, accountability
mechanisms, commercial imperatives, charitable intentions, scientific knowl-
edge and professional practice.
In this sense, the subject of this book is the combined process of ‘datafying’
and ‘digitizing’ education. Putting it simply, ‘datafication’ refers to the transfor-
mation of different aspects of education (such as test scores, school inspection
reports, or clickstream data from an online course) into digital data. Making
information about education into digital data allows it to be inserted into data-
bases, where it can be measured, calculations can be performed on it, and
through which it can be turned into charts, tables and other forms of graphical
presentation. ‘Digitization’ refers to the translation of diverse educational prac-
tices into software code, and is most obvious in the ways that aspects of teaching
and learning are digitized as e-learning software products. If you want to build
some digital e-learning software, you have to figure out how to do that in lines
of code: to encode educational processes into software products. Diverse aspects
of education from policy, leadership, management and administration to class-
room practice, pedagogy and assessment are now increasingly subjected to
processes of digitization, as software is coded and algorithms are designed to
augment and rework everyday tasks and processes across the education sector.
Datafication and digitization support and complement one another in myr-
iad ways. For example, when a piece of e-learning software is coded in digital
form, it is often designed in such a way that it can generate information about
the ways that it is used (visible in, for example, the log files that demonstrate
how a user has interacted with the software). That information can then be
used, as analysable digital data, to help the producers of the software learn
more about the use of their product, data which can then be used to help
inform the writing of better code (a software patch, upgrade or update) or the
programming of new software products altogether. To take another example:
when millions of learners around the world all take a standard global test, the
activities they undertake ultimately contribute to the production of a massive
database of test results. Making sense of the vast reserves of data in such a
01_Williamson_Ch-01.indd 5 7/4/2017 4:56:39 PM
6 Big Data in Education
database can only be accomplished using software that has been coded to
enable particular kinds of analyses and interpretations. The software does not
have to be especially appealing – the datafication of education depends to a
significant degree on the digital coding undertaken to produce very mundane
software products like spreadsheets and statistical analysis packages – but it is
certainly becoming more seductive with the ready availability of highly graph-
ical forms of data visualization software, as well as more accessible and easier
to use. With both educational technologies and educational data, processes of
digitization and datafication support and reinforce each other.
In short, much of education today is being influenced and shaped by the pro-
duction of lines of code that make digital software function, and by the
generation of digital data that allows information about education to be col-
lected, calculated and communicated with software products. Does this matter?
Yes, it matters urgently, because the coding of software products for use in educa-
tion, or the application of coded devices that can process educational digital data,
are beginning to transform educational policies, pedagogies and other practices
in ways which have so far been the subject of very little critical attention.
As new kinds of software are developed for use in educational contexts that rely
on both software code and digital data, we are beginning to see new ways in which
schools, universities, educational leaders, teachers, students, policymakers and
parents are influenced. Schools are being turned into data-production centres,
responsible for constantly recording and auditing every aspect of their perfor-
mance (Finn 2016). Leaders are being called on to act on their data to improve the
institutions they manage (Lewis and Hardy 2016), often using ‘learning manage-
ment systems’ to assist in administrative tasks (Selwyn etal. 2017). Students are
becoming the subjects of increasingly pervasive data mining and data analytics
packages that, embedded in educational technologies and e-learning software, can
trace their every digital move, calculate their educational progress and even pre-
dict their probable outcomes (Suoto-Otero and Beneito-Montagut 2016). Students
in universities are experiencing ever-greater use of online tools to measure their
progress (Losh 2014), with their assignments being entered into massive global
plagiarism detection databases (Introna 2016). At the same time, university man-
agers are required to make use of complex performance indicator metrics and
institutional data dashboards to facilitate decision-making and planning (Wolf
etal. 2016). Even early years settings such as nurseries are increasingly required to
collect data on young children’s development so that it can be tracked against
national and international benchmarks (Roberts-Holmes 2015; Moss etal. 2016),
which is mirrored by the growing use of analytics technologies in adult education
and professional learning (Fenwick and Edwards 2016).
Beyond the spaces of learning, policymakers are increasingly exhorted to
develop data-driven or ‘evidence-based’ policies that are crafted in response to
insights derived from digital data (Sellar 2015a), including school inspection
01_Williamson_Ch-01.indd 6 7/4/2017 4:56:39 PM
Introduction 7
data presented on institutions’ ‘data dashboards’ (Ozga 2016). Parents, too, are
encouraged to become educational data analysts who use digital ‘school com-
parison’ websites to inform their choices about which schools to enrol their
children in (Piattoeva 2015). For teachers, a new industry in educational ‘talent
analytics’, or ‘labour market analytics’, has even appeared (Beneito-Montagut
2017), with fully-automated software products like TeacherMatch acting as
‘advanced education talent management’ platforms for the recruitment, assess-
ment, professional development and ‘talent investment’ of teachers, using
matching algorithms to match schools with staff just like a social media dating
service (TeacherMatch 2015).
Many commercial organizations are changing their business models and prac-
tices to engage in education, such as Google with its Google Apps for Education
suite of free-to-use cloud services for schools (Lindh and Nolin 2016). Meanwhile,
existing commercial ‘edu-businesses’ such as Pearson – a global education textbook
publisher – have moved to become prominent educational software providers and
key collectors of educational data (Hogan etal. 2015). Commercial tools for data
collection, processing and analysis are finding their way into the discipline of edu-
cational research, knowledge production and theory generation too, in ways that
are reshaping how education is known and understood (Cope and Kalantzis 2016).
And finally, an increasing number of private sector ‘data brokers’ are starting to
collect education-related data, curate and aggregate it using analytics tools, and sell
it back to education stakeholders (Beneito-Montagut 2017).
It’s not just the people and organizations of education that are affected by the
recent acceleration of data-processing software, but curriculum, pedagogy and
assessment too. The notion of a curriculum containing the content-knowledge
to be taught in schools is itself being challenged, as new kinds of ‘adaptive’
learning software are developed that can semi-automate the allocation and ‘per-
sonalization’ of content according to each learners’ individual data profile
(Bulger 2016). Pedagogy is being distributed to automated machines such as
‘teacher bots’ and ‘cognitive tutors’: computerized software agents designed to
interact with learners, conduct constant real-time analysis of their learning, and
adapt with them (Bayne 2015). And the notion of assessment as a fixed event is
being supplanted by real-time assessment analytics and computer-adaptive test-
ing, which automatically assess each learner on-the-go and adapt to their
responses in real-time (Thompson 2016). What is even meant by ‘learning’ is
being questioned with the collection of datasets so large that enthusiasts believe
they can reveal new truths about learning processes that educational researchers
working within disciplinary frameworks such as psychology, sociology and phi-
losophy have been unable to detect before (Behrens 2013).
Many of these developments and innovations with digital software and data in
education exist technically, but they are also the product of extensive claims, pro-
motional activity and imaginative marketing which centres on the idea that
01_Williamson_Ch-01.indd 7 7/4/2017 4:56:39 PM
8 Big Data in Education
technical solutions have the capacity to transform education for the future.
Businesses with products to sell, venture capital firms with return on investment
to secure, think tanks with new ideas to promote, and policymakers with problems
to solve and politicians with agendas to set have all become key advocates for data-
driven education. Of course, we need to be at the very least cautious about many
of the claims made about the transformative and revolutionary potential of many
new developments, if not downright sceptical – and, indeed, a little resistant.
But the point I pursue throughout is that what we are currently witnessing are
signs of a new way of thinking about education as a datafied and digitized social
institution. Seriously powerful organizations are at work in this space, organiza-
tions with a forceful and influential shared imagination concerning the future
of education. It is easy to be dismissive of the claims-making, hype and hubris
that surround emerging developments like learning analytics and computer-
based cognitive tutors. But it’s less easy to dismiss these developments and the
claims that support them when you can see that some of the world’s richest and
most powerful companies are dedicating extraordinary research and develop-
ment resources to them; when you can read reports advocating and sponsoring
them by influential think tanks; when you hear that politicians are backing
them; when you discover that enormous sums of venture capital and philan-
thropic funding are being invested to make them a reality.
A shared vision of the digitization and datafication of education is emerging.
Diverse ideas and actors have combined to produce collective imaginative
resources that can be used to animate research and development (R&D) prac-
tices, to persuade politicians, to generate investment, and to galvanize new
practices (Jasanoff 2015). Of course, education has long been a site of future
imagination. A ‘dominant myth of the future of education’ in recent years has
been one that ‘emerges out of an instrumental conception of education as pri-
marily concerned with serving the formal economy’ (Facer 2011: 8). Visions of
data-driven education complicate this dominant myth of the future. While eco-
nomic fantasies of human capital development persist, they are being
supplemented and extended by dreams of new forms of governance and citizen-
ship, new scientific aspirations of psychological optimization and cognitive
enhancement, and new commercial objectives to insert private sector technolo-
gies and practices into public education.
Myths and imaginative visions, moreover, can become material realities when
given technical form and inserted into social contexts. The developments traced
out in the following chapters are all parts of a new emerging imaginary of the
digital future of data-driven education that appears to be considered desirable,
and that many organizations and individuals seem to agree could and should be
attained through putting new technical developments into practice in the present.
The twin processes of digitization and datafication form the basis for the book,
but the practices of coding educational technologies of various kinds and of
01_Williamson_Ch-01.indd 8 7/4/2017 4:56:39 PM
Introduction 9
datafying education through diverse techniques are all also situated contextually
and are animated by a particularly powerful imaginative resource which envi-
sions education as a massively data-driven and software-supported social
institution. The difference that digital data make in education is the result of the
highly diverse efforts of programmers, project managers, businesses, startup
accelerator programs, policymakers and politicians, think tanks and innovation
labs, school managers, leaders, and educators themselves – the material practices
of all of them shaped by an imagined vision of a digitized and datafied future
which has become increasingly pervasive, persuasive and seemingly desirable.
Datafying Education
‘Datafication’ refers to the transformation of many aspects of education into
quantifiable information that can be inserted into databases for the purposes of
enacting different techniques of measurement and calculation. Datafication
itself has a long history, detailed more fully in Chapter 2. Recent developments
such as the establishment of data labs and data centres for educational data min-
ing and analysis, and the proliferation of specific products such as learning
analytics, adaptive learning software and computerized tutors, all rely on the
constant collection of masses of digital data. Large-scale educational data has
been available from the aggregation of test results or school census information
for decades. The key shift with big data is that it is now collected in or near real-
time directly as learners interact with software systems. That is to say, large-scale
datasets have been historically gathered primarily through assessments and data
collection events that have to be separated off from the normal rhythms of the
classroom; big data are captured from within the pedagogic machinery of the
teaching and learning process itself by being pieced together from the millions
of data points that are generated as learners click on content and links, engage
with digital educational materials, interact with others online, and post
responses to challenges. Digital course content, online courses, e-textbooks,
digital simulations, and more, provide the front-end interface for the production
of educational big data, behind which lies a sophisticated back-end infrastruc-
ture of data collection, information storage, algorithmic processing, and
analytics and data visualization capacities.
Underlying these developments is a set of powerful animating visions or
imaginaries of datafication. The authors of Learning with Big Data: The Future of
Education (Mayer-Schönberger and Cukier 2014) imagine that big data will
‘reshape learning’ through ‘datafying the learning process’ in three significant
ways: (1) through real-time feedback on online courses and e-textbooks that can
‘learn’ from how they are used and ‘talk back’ to the teacher; (2) individualiza-
tion and personalization of the educational experience through adaptive
01_Williamson_Ch-01.indd 9 7/4/2017 4:56:39 PM
10 Big Data in Education
learning systems that enable materials to be tailored to each student’s individual
needs through automated real-time analysis; and (3) probabilistic predictions
generated through data analytics that are able to harvest data from students’
actions, learn from them, and generate predictions of individual students’ prob-
able future performances. The authors imagine school as a ‘data platform’ where
the real-time datafication of the individual is becoming the ‘cornerstone of a
big-data ecosystem’, and in which ‘educational materials will be algorithmically
customized’ and ‘constantly improved’ (Mayer-Schönberger and Cukier 2014).
A significant amount of data-driven activity has been undertaken in the
higher education sector, through widespread use of learning management sys-
tems and online programs such as MOOCs (Massive Open Online Courses)
(Knox 2016). But schools are also being targeted for datafication. The US think
tank the Center for Data Innovation has produced a report advocating a vision
of a ‘data-driven education system’ for schooling. ‘U.S. schools are largely failing
to use data to transform and improve education, even though better use of data
has the potential to significantly improve how educators teach children and
how administrators manage schools’, its author claims (New 2016: 1). Instead,
the think tank argues that a data-driven education system should achieve four
main goals:
Personalization: Educators dynamically adjust instruction to accommodate students’
individual strengths and weaknesses rather than continue to utilize a mass
production-style approach.
Evidence-Based Learning: Teachers and administrators make decisions about how to
operate classrooms and schools informed by a wealth of data about individual and
aggregate student needs, from both their own students as well as those in comparable
schools across the nation … rather than by intuition, tradition, and bias.
School Efficiency: Educators and administrators use rich insight from data to explore the
relationships between student achievement, teacher performance, and administrative
decisions to more effectively allocate resources.
Continuous Innovation: Researchers, educators, parents, policymakers, tech developers,
and others can build valuable and widely available new education products and
services to uncover new insights, make more informed decisions, and continuously
improve the education system. (New 2016: 2)
These goals for data-driven education systems accurately capture the dominant
imaginary related to the collection and use of data in schools. ‘Personalization’
has become perhaps the main keyword of data-driven education, emphasizing
systems and processes that can be intelligently tailored to the individual students.
The use of evidence to perform comparisons across institutions and systems has
a long lineage in education policy, but with digitization is becoming much easier
and quicker to conduct. Achieving efficiency is paramount for schools, with per-
formance management tools now available to ensure that students, teachers and
01_Williamson_Ch-01.indd 10 7/4/2017 4:56:39 PM
Introduction 11
administrators are all producing measurable outputs. And as larger and larger
quantities of data become available – as masses of educational big data – new pat-
terns and insights are being sought to address the goals of various stakeholders,
such as the improvement agendas of policymakers and the new product develop-
ment plans of businesses. The imagined datafication of schools is to be attained
through pursing these goals of personalization, evidence-based learning, effi-
ciency and continuous innovation.
How do such goals and imaginative visions look in practice? Compelling
examples of how the datafication of schools might look in the imagined near
future of education are provided by Silicon Valley ‘startup schools’. Startup
schools are new educational institutions designed as alternatives to the main-
stream state schooling model, and they originate in the technology
entrepreneurship culture of Silicon Valley, the technofinancial heart of the
global tech industry. A prominent example is AltSchool, set up in 2013 by Max
Ventilla, a technology entrepreneur and former Google executive. It ‘prepares
students for the future through personalized learning experiences within micro-
school communities’, and its stated aim is to ‘help reinvent education from the
ground up’ (AltSchool 2015a). A recent profile of its founder claimed that ‘when
Ventilla quit Google to start AltSchool, in the spring of 2013, he had no experi-
ence as a teacher or an educational administrator. But he did have extensive
knowledge of networks, and he understood the kinds of insights that can be
gleaned from big data’ (Mead 2016). After establishing in four sites in San
Francisco as a ‘collaborative community of micro-schools’, AltSchool later
expanded to Brooklyn and Palo Alto, with further long-term plans for new
schools and partnerships across the US. It has since hired executives from
Google, Uber and other successful Silicon Valley startups, many with experience
of big data projects. The AltSchool chief technology officer, formerly the engi-
neer in charge of the Google.com homepage and search results experience, has
stated that ‘I am highly motivated to use my decade of Google experience to
enable the AltSchool platform to grow and scale’ (AltSchool 2015a). The
AltSchool ‘platform’ is described as a new ‘central operating system for educa-
tion’, one designed according to ‘technology-enabled models’ that are
transforming other industries and institutions, such as Uber and Airbnb
(AltSchool 2015b).
The models it refers to are those of the datafication of other sectors. Airbnb
represents the datafication of accommodation letting. Uber has thoroughly data-
fied taxi services. AltSchool has been programmed to run on the same basic
model, or operating system, as these datafied sectors. Thus, it depends on a
sophisticated data analytics platform. Its suite of digital tools is intended to
‘make personalized education a reality’, which it seeks to accomplish by support-
ing teachers to ‘develop Personalized Learning Plans and to capture student
progress toward them’:
01_Williamson_Ch-01.indd 11 7/4/2017 4:56:39 PM
12 Big Data in Education
We also create platforms for efficient classroom administration so teachers have
more quality face-to-face time with their students. … To ensure we are always learn-
ing from what happens outside the classroom, we build digital tools to support
collaboration between teachers, parents and students. … Our project-based educa-
tion approach truly comes alive when supported by carefully curated learning tools.
We mentor each student in the use of technology for learning and help them skil-
fully navigate today’s information terrain. (AltSchool 2016)
The data platform driving AltSchool is not just a technical system: it has been
constructed to support a particular cultural vision of education as being ‘person-
alized’ around each individual. Personalization is its dominant ideal, and it is
personalization that has been achieved successfully within the commercial
social media activities of many Silicon Valley companies. For instance, Google
search results are automatically personalized to each user based on their web
search history. The Facebook timeline is personalized around the friends graph
it constructs about each user’s social network connections. The logic of person-
alization drives the ways in which social media platforms make recommendations
for people to follow, consumer goods to buy, memes to share and so on. The
culture and techniques of personalization from the commercial social media
sphere are inserted into schooling through spaces such as AltSchool, and built in
to its data platforms as a technical back-end complement to the front-facing
cultural vision of education it projects. AltSchool ultimately balances and assem-
bles a range of resources that appear unproblematically to crisscross the traverse
between technological ideals and educational concepts.
Beyond technical and cultural similarities with the datafying priorities of the
tech industry, the startup school also enjoys the financial benefits of Silicon
Valley startup culture. On its establishment, AltSchool originally raised US$33
million in venture capital funding, with another US$100 million investment in
2015, including donations from Facebook’s Mark Zuckerberg and the venture
capital firm Andreeson Horowitz (AltSchool 2015b). AltSchool is, then, thor-
oughly governed, managed and financed through the discourses and material
practices of Silicon Valley startup culture. Its operating system is modelled on
social media data analytics. Its funding is almost exclusively generated through
venture capital and tech philanthropy. Its engineering and design team are
applying their social media expertise in data dashboards, algorithmic playlist-
ing, adaptive recommender systems and app development to the development
of new personalized edtech devices and platforms. The datafication of educa-
tion prototyped by AltSchool, and other startup school models, is not just a
technical accomplishment but the product of a financial investment model for
Silicon Valley startups that has been trialled in other sectors, transplanted into
education, and appears to be on the cusp of being scaled-up as a competitive
market solution to the problem of mainstream schooling. Technology visionar-
ies and imaginative entrepreneurs like Max Ventilla are becoming high-status
01_Williamson_Ch-01.indd 12 7/4/2017 4:56:40 PM
Introduction 13
education reformers, using their technical expertise in software development
and data analytics, combined with the entrepreneurial business expertise
required to generate investment, as powerful resources to attract others to their
educational visions.
As the AltSchool example demonstrates, big data is not just technical. It is,
rather, the ‘manifestation of a complex sociotechnical phenomenon that rests
on an interplay of technological, scientific, and cultural factors’:
While the technological dimension alludes to advances not only in hardware, soft-
ware, but also infrastructure and the scientific dimension comprises both mining
techniques and analytical skills, the cultural dimension refers to (a) the pervasive use
of ICTs in contemporary society and (b) the growing significance and authority of
quantified information in many areas of everyday life. (Rieder and Simon 2016: 2,
italics in original)
Throughout the chapters that follow, the datafication of education is treated as
the contingent materialization of future visions, technologies and skilled scien-
tific techniques, as well as of political, commercial and philanthropic ambitions,
all of which are combining into hybrid sociotechnical systems for data-driven
measurement and management.
Digitizing Education
Datafication of education requires learning environments to be highly instru-
mented to collect information (Cope and Kalantzis 2015). This means the
learning environment needs to be increasingly digitally-mediated, or digitized,
as AltSchool’s technical ‘operating system’ demonstrates. The use of the term
‘digitization’ refers to ‘the process of converting information from analog into
discrete units of data that can be more easily moved around, grouped together,
and analysed’ (Gregory etal. 2017: xviii) using computer technologies. With the
digitization of education into information that can be processed by a computer,
software and the code that enacts it becomes a significant influence in how edu-
cation is organized. Software code has become a system for regulating many of
the practices and processes of education, teaching and learning.
Described in more detail in Chapter 3, it is important from the outset to
acknowledge that code is both a product – the end-result of the work of program-
mers, working in real material conditions, with their own professional cultures
and values, and whose coding practices are shaped by business plans and
objectives – and as a productive force in the world (Kitchin and Dodge 2011). By
describing code as ‘productive’ registers the ways in which code is programmed
to perform tasks that it then enacts (or, to use the specific computational term,
‘executes’). Code instructs a software program to ‘do something’ on a computer,
01_Williamson_Ch-01.indd 13 7/4/2017 4:56:40 PM
14 Big Data in Education
and in that basic sense it can be seen as productive. But it is also productive
because writing code to execute a particular kind of task also fundamentally alters
the nature of the task it is being instructed to perform (Mackenzie 2006).
In this book I focus on the ways that turning educational things into code
then loops back to change education. However, to think of code just in technical
terms, as a script for instructing software written in specific programming lan-
guages, would be misleading. It is certainly the case that e-learning software,
policy databases and school management programs depend on lines of code for
their functioning. But that code has itself to be written, or produced, as noted
earlier. Programmers have to craft it, using specific kinds of programming lan-
guages and code repositories. Those programmers work according to the business
plans, project management schedules and objectives of their employers. Those
business plans are the operational manifestation of powerful future visions. The
code produced to make software programs function is also dependent on finan-
cial investment, funding programmes and economic priorities. This goes beyond
the straightforward allocation of programmers’ salaries and includes the work of
entrepreneurs in securing venture capital for software startups, of politicians
providing tax incentives for technology companies, and of philanthropists mak-
ing donations to finance new technical innovations. The software programs that
enact much of education today, in other words, are also the product of imagina-
tive business and political programmes.
An illustrative example of how digital imaginaries, software, finance and
politics are interwoven in the contemporary transformation of education is
provided by Edtech UK. This organization is ‘a new strategic body set up to help
accelerate the growth of the UK’s education technology sector in Britain and
globally’:
the new body is a ‘front door’ for industry, investment and government and a conven-
ing voice for all of the education and learning technology sector including educators,
startups, scale up and high growth companies, large corporations, investors, regula-
tors and policy makers. The focus of Edtech UK is to help support, showcase and
develop the sector, with a focus on creating more jobs, developing new skills, under-
standing what works and driving economic growth. Its focus will be global from the
outset with an ambitious programme of work to take the Best of British edtech com-
panies to the world and be a launchpad for the world’s best education and learning
organisations to base themselves and grow in the UK. (Edtech UK 2015)
Edtech UK has been established by the Education Foundation, which describes
itself as ‘the UK’s first independent, cross sector, education think tank’ and is
‘focused on three priorities: education reform, technology and innovation’.
Since 2011 it has led an ‘edtech incubator’ for new educational technology com-
panies; worked with Facebook on a guide for educators; sought to influence
policy development at a national level including running Britain’s first Education
01_Williamson_Ch-01.indd 14 7/4/2017 4:56:40 PM
Introduction 15
Reform Summit in partnership with the Department for Education and the
Secretary of State for Education; developed a corporate partners network with
Facebook, IBM, Pearson, HP, Randstad Education, Cambridge University Press,
McKinsey, Skype, Sony, Google and Samsung; and delivered policy roundtables,
conferences, summits, and media events around educational technology in both
the UK and USA. Itself an ‘incubated’ project of the Education Foundation,
Edtech UK was launched by Boris Johnson, then Mayor of London, with the
support of the UK government departments of Business, Innovation and Skills,
and of Trade and Industry, as well as by a private sector coalition of organiza-
tions from the technology sector.
Political aspirations and financial capacity, as well as technical expertise and
a vision of the future of educational technology, are all combined in the activi-
ties of Edtech UK. It has powerful political support, it is modelled on financial
lobbying and accelerator organizations, and it mobilizes a hybrid discourse of
investment, venture capital, startup and scale-up, and economic growth. Its cor-
porate brochure for attracting new edtech startups to London promises
extraordinary benefits. It references a ‘large and profitable market’ for educa-
tional technology; the benefits of ‘flexible procurement’ regulation which allows
schools autonomy in their choice of technology suppliers; proximity to global
edtech companies like Pearson and Knewton and the presence of ‘talent, venture
capital, co-working space, government support, seed funding and events’ in
London; seed enterprise investment, tax breaks and ‘entrepreneurial relief’ for
early-stage companies; plus, it claims, the incentives of ‘global education tech-
nology sector spending at $67.8bn in 2015 and a global “e-learning” market
worth $165bn, which is poised to reach $243.8bn by 2022’ (Education
Foundation 2015). It is only amid the political, financial and commercial activi-
ties of Edtech UK that the work of programmers in producing educational
technologies can take place.
Edtech UK is a compelling example of how the digitization of education –
through support for new edtech startup companies – relies on the financial flows
that make up the lines of information in a bank account, as well as on establish-
ing political lines of linkage, as much as on the lines of code that actually make
the software work. As Lynch (2015) conceptualizes it in The Hidden Role of
Software in Education, a new kind of ‘software space’ made of code, algorithms
and data produced by commercial actors, programmers and analysts is nowadays
working alongside both the ‘economic space’ of investment, funding and
finance and the ‘political space’ of educational policymaking and governance,
then exerting its influence on the ‘practice space’ of teaching and learning.
Edtech UK is emblematic of how imaginative future visions, software, economics
and politics combine and interrelate with one another to impact on the practice
spaces of education. The digitization of education is not simply about the trans-
lation of educational practices into software products, but about the manifold
01_Williamson_Ch-01.indd 15 7/4/2017 4:56:40 PM
16 Big Data in Education
ways in which code comes into being, according to particular values, priorities
and objectives, and in accordance with specific kinds of aspirations for the
future of education.
The Digital Imagination and Materiality of Education
The examples of AltSchool, Edtech UK, Center for Data Innovation and
HackingEDU we have encountered so far provide us with some sense of the
imagined possibilities of datafication and digitization being associated with edu-
cation. The aim of this book is neither to uncritically celebrate these developments,
nor to debunk them. Instead, my intention is to consider how the twin processes
of datafication and digitization are emerging from, and simultaneously reinforc-
ing, a particular kind of reimagining of the future of education. Some sense of
this reimagining is apparent from AltSchool’s emphasis on personalized learning
supported by data analytics platforms, and from Edtech UK’s involvement in
seeking to grow a future edtech market through both business and political net-
works. How to make sense of the work of imagination that underpins these
diverse and emerging approaches?
In order to do this kind of analysis, I make use of the concept of ‘sociotechni-
cal imaginaries’ from the field of science and technology studies (STS). By
sociotechnical imaginaries, what is meant are ‘collectively held, institutionally
stabilized, and publicly performed visions of desirable futures, animated by
shared understandings of forms of social life and social order attainable through,
and supportive of, advances in science and technology’ (Jasanoff 2015: 4).
Sociotechnical imaginaries are not just science fiction fantasies: they constitute
the visions and values that catalyse the design of technological projects. The
dreamscapes of the future that are dreamt up in science laboratories, technical
R&D departments, software companies and entrepreneurs’ offices sometimes,
through collective efforts, become stable and shared objectives that are used in
the design and production of actual technologies and scientific innovations –
developments that then incrementally produce or materialize the desired future.
Through sociotechnical imaginaries, transformative scientific ideas, technological
objects and social norms become fused in practice and help to sustain social
arrangements or create new rearrangements in cultures, institutions and routines.
Sociotechnical imaginaries are therefore the product of specifically political acts
of imagination, because they act as powerful aspirational and normative visions
of preferred forms of social order.
The concept of sociotechnical imaginaries has been taken up to understand the
visions and values that underpin digital developments such as social media and
search engines. The capacity to imagine the future is becoming a powerful consti-
tutive element in social and political life, particularly as it infuses the technological
01_Williamson_Ch-01.indd 16 7/4/2017 4:56:40 PM
Introduction 17
visions and projects of global media companies (Mager 2016). Organizations such
as Google and Facebook, Apple and Amazon can be understood as dominant pro-
ducers of sociotechnical imaginaries, whose aspirations are therefore becoming
part of how collectively and publicly shared visions of the future are accepted,
implemented and taken up in daily life. As a variation on the term ‘sociotechnical
imaginary’, Mager (2015: 56) describes ‘algorithmic imaginaries’ that emerge from
‘a very specific economic and innovative culture’ associated with Silicon Valley
technology companies, and which privilege their originators’ ‘techno-euphoric
interpretations of Internet technologies as driving forces for economic and social
progress’.
The production of such desirable imaginary futures is both social and technical,
which is why they are referred to as ‘sociotechnical’. That is to say, such futures are
produced by particular social groups within specific social contexts, and they are
also projected through the design of particular kinds of technologies – or express a
view of particular futures in which those kinds of technologies are imagined to be
integral, embedded parts. Unpacking sociotechnical imaginaries requires research
that focuses on ‘the means by which imaginaries frame and represent alternative
futures, link past and future times, enable or restrict actions in space, and natural-
ize ways of thinking about possible worlds’ (Jasanoff 2015: 24). In slightly different
terms, the imagining of a ‘digital future’ projects a kind of ‘mythology’ (a set of
ideas and ideals) that animates, motivates and drives forward technical develop-
ment but is always much more contested and messily realized, and never as simple,
straightforward or idealized as it is imagined to be (Dourish and Bell 2011).
Imaginaries in this sense act as models or diagrams to which certain actors
hope to make reality conform, serving as ‘distillations of practices’ for the shap-
ing of behaviours and technologies for visualizing and governing particular
ways of life and forms of social order (Huxley 2007: 194). Sociotechnical
imaginaries animate technical projects and social organization, and provide
models for ways in which certain spaces and places might be designed and
arranged. The organization of societies in this sense depends on shared imagina-
tive resources, language and practical techniques that combine in the materiality
of ‘fabricated spaces’ – that is, spaces that have been ‘realized’ in the form in
which they have been imagined (Rose 1999a: 33). In other words, sociotechnical
imaginaries are often enacted and materialized through linguistic and concrete
practices in ways that weave the underlying vision into the fabric of society.
Thus, while sociotechnical imaginaries ‘can originate in the visions of single
individuals or small collectives’, they can gather impetus ‘through blatant exer-
cises of power or sustained acts of coalition building’ to enter into ‘the
assemblages of materiality, meaning and morality that constitute robust forms
of social life’ (Jasanoff 2015: 4). Fabricated spaces, then, are the result of imag-
inaries that have been realized and materialized through particular technical,
discursive and practical acts.
01_Williamson_Ch-01.indd 17 7/4/2017 4:56:40 PM
18 Big Data in Education
We can understand new educational projects and places such as AltSchool as
the fabricated material product of a specific sociotechnical imaginary of educa-
tion. It has been brought into existence as a new fabricated space of education
through discursive and material means as ways of realizing a future that is seen by
its advocates and sponsors as desirable and possible to attain. In other words,
AltSchool itself acts as an imaginary model for the future spaces of schooling that
it is seeking to fabricate in reality through operationalizing its technical platforms,
and which it is supporting discursively through reference to specific kinds of pro-
gressive educational thinking. Moreover, we can think of AltSchool as an extension
of Silicon Valley, translating its particular culture and spaces of innovation to the
education sector. AltSchool represents the sociotechnical imaginary of Silicon
Valley relocated to the materiality of the classroom. Given AltSchool’s aspirations
to scale its model to other sites, we can appreciate how AltSchool functions as the
material product of a sociotechnical imaginary which defines how education in
the future might be, could be, or perhaps even should be, and that might shape
and delimit the everyday practices of all those who inhabit it. In this sense, the
current sociotechnical imaginaries and mythologies of education, in which digiti-
zation and datafication will play a significant role, are already becoming the lived
reality of education – with all of the mess and potential contestation that entails –
and need to be critically examined for the material effects they might exert.
Researching Digitization and Datafication In Education
If imaginary spaces become material zones to inhabit, they can therefore exert
real consequences on those who experience them. To tease open the material
consequences of emerging sociotechnical imaginaries of education, it is impor-
tant to look closely at the software that will make such spaces operational.
Researching the digitization and datafication of education therefore requires
some novel methodological and conceptual approaches. Although the science,
technology and society (STS) concept of sociotechnical imaginaries can help to
understand the future visions that are animating and catalysing recent and
ongoing technical development, we also need methods and concepts to grasp
their (actual or potential) material consequences and effects. The emerging field
of digital sociology has begun to address how digital technologies, software and
data are being embedded into all kinds of social and cultural activities, institu-
tions, relations and processes (Orton-Johnson and Prior 2013):
For some theorists, the very idea of ‘culture’ or ‘society’ cannot now be fully under-
stood without the recognition that computer software and hardware devices not
only underpin but actively constitute selfhood, embodiment, social life, social rela-
tions and social institutions. (Lupton 2015a: 2)
01_Williamson_Ch-01.indd 18 7/4/2017 4:56:40 PM
Introduction 19
For digital sociologists, digitization has important implications for our ways of
knowing, studying and understanding the social world, which demand interdis-
ciplinary approaches drawing from a longer history of internet studies, media
and cultural studies, science and technology studies, surveillance studies and
computational social science (Daniels etal. 2016; Halford etal. 2013).
Digital sociology, then, confronts the ways in which ‘new digital media, the data
they produce and the actors involved in the collection, interpretation and analysis
of these data’ now increasingly structure and shape the social world (Lupton 2015a:
17–18). It seeks to understand, for example, how people’s everyday lives are increas-
ingly mediated through routine digital transactions with governments, commercial
organizations and public institutions; how space is experienced through mobile
devices; how social media has become part of social networks; and how we learn
about the world through new digital media forms. Many of the central preoccupa-
tions of sociologists, such as identity, power relations and inequalities, social
networks, structures and social institutions, now need to be considered from the
perspective of the ongoing digitization and datafication of many aspects of society.
‘Software studies’ has emerged as an interdisciplinary orientation to the study
of software, and includes research from the arts, philosophy, humanities, geog-
raphy, cultural studies and the social sciences. Studies of software tend to share
two key emphases. They focus on the software, programs and social cultures that
produce effects in social life from a critical social scientific and cultural perspec-
tive, and on the social and material work that contributes to its production.
Software studies seek to engage with the ‘stuff of software’ and:
to see behind the screen, through the many layers of software, logic, visualization,
and ordering, right down to the electrons bugging out in the microcircuitry, and
on, into the political, cultural and conceptual formations of their software, and out
again, down the wires into the world, where software migrates into and modifies
everything it touches. (Fuller 2008: 1)
This is clearly a tall methodological order, requiring expertise in the technicali-
ties of software, the political and cultural processes involved in its production,
and the social consequences that occur as it then spreads into highly diverse
practices of work, leisure, politics, culture, economics, social relations and so on.
In order to establish a set of methodological parameters for such research, Kitchin
and Dodge (2011: 246) have usefully defined a ‘manifesto for software studies’:
Rather than focus purely on the technical, it fuses the technical with the philo-
sophical to raise questions about what software is, how it comes to be, … how it
does work in the world, how the world does work on it, why it makes a difference
to everyday life, the ethics of its work, and its supporting discourses. Software stud-
ies then tries to prise open the black boxes of algorithms, executable files, [database]
structures, and information protocols to understand software as a new media that
augments and automates society.
01_Williamson_Ch-01.indd 19 7/4/2017 4:56:40 PM
20 Big Data in Education
Their manifesto particularly highlights the need for critical research on the ways
in which code emerges, how it performs, and how it seduces and disciplines. In
terms of how code emerges, they urge for greater attention to the knowledge,
practices, materials and marketplaces that are involved in the production of
code, and the political, economic and cultural contexts that frame its produc-
tion. They suggest performing detailed ethnographic studies of how developers
produce code, and the life of software projects, to understand how software is
created and how it is put to work in specific contexts.
Kitchin and Dodge then suggest that software studies might attend to the
ways in which code performs. By this they mean analysing in detail ‘the contex-
tual ways in which code reshapes practices with respect to industry, transportation,
consumption, governance, education, entertainment and health’, as well as
‘knowledge production, creative practice, and processes of innovation’, and
studying how code ‘makes a difference’ to those spaces and contexts through
imbuing them with the capacity to do new types of work (Kitchin and Dodge
2011: 249). They also argue that code seduces and disciplines, largely because it
offers people real benefits in terms of convenience, efficiency, productivity and
creativity, whilst also enforcing more pervasive forms of surveillance and man-
agement. In particular, Kitchin and Dodge note how software is supported by
powerful and consistent discourses, such as those of safety, security, empower-
ment, productivity, reliability, economic advantage, which persuade people to
willingly and voluntarily embrace it. As such, software and code are amenable to
forms of documentary and discourse analysis.
‘Critical data studies’ is another emerging body of interdisciplinary research that
engages with the datafication of many aspects of society. A special issue on the topic
of critical data studies introduced the field as a ‘formal attempt at naming the types
of research that interrogate all forms of potentially depoliticized data science and to
track the ways in which data are generated, curated, and how they permeate and
exert power on all manner of forms of life’ (Iliadis and Russo 2016: 2). Iliadis and
Russo (2016: 5) further highlight the identification of social data problems and the
design of critical frameworks for addressing them. As a set of approaches to the
critical examination of various forms of digital data – including big data, open data
and data infrastructures – as well as the diverse practices of data science as a social,
professional and technical discipline, critical data studies has found purchase with
geographers, sociologists, philosophers and researchers of education.
In one of the first publications detailing critical data studies, the geographers
Dalton and Thatcher (2014) set out seven defining commitments: (1) situate
data regimes in temporal and spatial context; (2) reveal data as inherently polit-
ical and expose whose interests they serve; (3) unpack the complex,
non-deterministic relationship between data and society; (4) illustrate the ways
in which data are never raw but always intentionally generated; (5) expose the
fallacies that data can speak for themselves and that exhaustive big data will
01_Williamson_Ch-01.indd 20 7/4/2017 4:56:40 PM
Introduction 21
replace smaller-scale sampled data; (6) explore how new data regimes can be
used in socially progressive ways; and (7) examine how academia engages with
new data regimes and the opportunities of such engagement.
In another article outlining concepts and methods for critical data studies,
Kitchin and Lauriault (2014) seek to provoke researchers to unpack the complex
‘assemblages’ that produce, circulate, share/sell and utilize data in diverse ways.
Data assemblages, as they define them, consist of technical systems of data col-
lection, processing and analysis, but also the diverse social, economic, cultural
and political apparatuses that frame how they work. In this broad sense, a data
assemblage includes: (1) particular modes of thinking, theories and ideologies;
(2) forms of knowledge such as manuals and textbooks; (3) financial aspects such
as business models, investment and philanthropy; (4) the political economy of
government policy; (5) the materiality of computers, networks, databases and
analytics software packages; (6) specific skilled practices, techniques and behav-
iours of data scientists; (7) organizations and institutions that collect, broker or
use data; (8) particular sites, locations and spaces; and (9) marketplaces for data,
its derivative products, its analysts and its software.
Approaching critical data studies in terms of sociotechnical data assemblages
is productive for research into the production and use of educational data. This
book provides a series of explorations of big data as it is entering into the com-
plexities of education and reworking teaching, learning, assessment, governance
and educational research itself. For the field of education research, big data is a
new and emerging phenomenon about which there remains limited knowledge
(Beneito-Montagut 2017). In the following chapters, I combine the focus on
sociotechnical imaginaries with digital sociology, software studies and critical
data studies approaches as a methodological strategy to perform a series of
critical analyses of the ways in which assemblages involving software code,
algorithms and digital data are making a difference in education.
This is not to suggest that existing approaches to educational research,
description and explanation are irrelevant. Rather, part of my aim is to demon-
strate that educational research can be productively extended by engaging with
software and data from a critical perspective. Studies of educational policy, for
example, have already begun to engage with the software packages and data
infrastructures that enable policy information to be collected, and that also
allow policies to penetrate into institutional practices. In the following chapters
I seek to understand how some of the software technologies penetrating educa-
tion today have come into existence and inquire into the imaginaries that
animate them; to explore the forms of expertise and knowledge they work in
relation with; to examine how they are being put to work in specific contexts
and spaces and how they are shaping particular practices; and to explore how
they are promoted and supported by certain discourses emanating from diverse
public, private and philanthropic sectors.
01_Williamson_Ch-01.indd 21 7/4/2017 4:56:40 PM
22 Big Data in Education
Learning Machines
By working with concepts of sociotechnical imaginaries and critical approaches
to software and data, I aim to show how powerful future visions are fast being
turned into the ordinary artefacts that are enabling digitization and datafication
in education. A useful term to capture these artefacts of educational digitization
and datafication is ‘learning machines’. This is a term I borrow from Michel
Foucault. In his highly influential work on regimes of discipline, Foucault (1991)
traced some of the ways in which schools function to supervise and discipline
pupils, particularly through techniques like timetabling, sitting them in rows in
classrooms, and organizing them in ranks according to age, performance, behav-
iour, knowledge and ability. Together, Foucault (1991: 147) argued, these
techniques ‘made the educational space function like a learning machine, but
also as a machine for supervising, hierarchizing, rewarding … according to the
pupils’ progress, worth, character, application, cleanliness and parents’ fortune.’
He detailed how classrooms functioned by placing pupils in categories, classifica-
tions and rankings based on constant assessments of their qualities, age,
development, performance and behaviour. Through techniques of ordering and
ranking pupils according to diverse categories, Foucault argued, ‘the classroom
would form a single great table, with many different entries’, and he noted that
classrooms are ‘mixed spaces’ – ‘real’ insofar as they consist of buildings, rooms
and furniture, but ‘also ideal, because they are projected over this arrangement
of characterizations, assessments, hierarchies’ (1991: 148).
The categorization and tabularization of educational institutions, spaces, pro-
cesses and individuals is perhaps the ideal aim – or dominant imaginary – of big
data in education. In this sense, what Foucault designated learning machines
takes on new resonance in the era of big educational data. The learning
machines being imagined and built today consist of computational technologies
that can capture and process data about learning; that can intervene in learning
practices, processes and institutions; that can ‘learn’ from the data they process;
and that can be understood as techniques of power, instruments for the control
of activity, behaviours and bodies, and processes of knowledge generation. They
are smart learning machines, the material and operational form of the socio-
technical imaginary of big data in education. Through big data, schools,
colleges, universities and other informal learning contexts are becoming
‘machine[s] for learning, in which each pupil, each level and each moment, if
correctly combined’, are becoming ‘permanently utilized in the general process
of teaching’ by a ‘precise system of command’ which operates ‘according to a
more or less artificial, prearranged code’ (Foucault 1991: 165–6). The smart
learning machines associated with digitization and datafication in education are
the product of lines of code, in the technical sense, that also enforce particular
codes of conduct. Digital software allows institutions, practices and people to be
01_Williamson_Ch-01.indd 22 7/4/2017 4:56:40 PM
Introduction 23
constantly observed and recorded as data; those data can then be utilized by
learning machines to generate insights, produce ‘actionable’ intelligence, or
even prescribe recommendations for active intervention. The ideal sociotechni-
cal imaginary of big data in education is now being materialized and
operationalized through smart learning machines, made of software code and
data, which might inhabit real educational spaces.
01_Williamson_Ch-01.indd 23 7/4/2017 4:56:40 PM
