The Inefficiency Is the Education

Building a blended learning physics course has left me with serious doubts about “blended learning” as an educational category. That may sound contradictory. I am building a course that deliberately uses video analysis, simulations, computational models, shared datasets, wireless communication technologies, software-defined radio, and eventually amateur-radio work connected to the possibility of a contact with the International Space Station (ARISS). Yet the further I go into the design, the less convinced I am that the important question is how much technology can be incorporated into instruction. Instead I wonder, What kind of intellectual work should students be doing, and what kind of technology really fits the task?

Contemporary educational technology too often begins from the opposite premise. The device or delivery system is treated as the innovation, and curriculum is subsequently reorganized around it. Lessons are modules. Discussion is replaced with posts. And participation is reduced to a record of logins, submissions, and digital traces. The language of education remains, but the substance thins.

Physics makes this problem difficult to ignore.

Mechanics is learned by confronting the material problem of motion: measuring position, arguing about what counts as a reference point, producing imperfect data, comparing evidence, discovering the failure of an intuitive explanation, and constructing a better model. A graph is not the motion. A simulation is not the physical system. A computational model is not nature. Each is a representation produced under particular assumptions.

Students must continually distinguish what they measured, what the computer calculated, what the model assumed, and what they themselves inferred.

Technology is enormously useful when it makes something inaccessible available for investigation. A simulation can isolate a variable not easily isolated experimentally. A spreadsheet or Jupyter notebook allows students to investigate many cases quickly and expose a mathematical relationship. But these affordances require surrendering the lesson to the machine.

One of the most important decisions I made while designing the course was to reject the assumption that a blended course must rely on digital mediation everywhere. When students point at a measurement, challenge another group's interpretation, return to the apparatus, revise a graph, and argue about whether the evidence actually supports a claim, moving that interaction onto a commercial platform is more of a reduction than an improvement on the educational experience.

This is why I developed a physical Class Evidence Wall rather than automatically using an online discussion board. Groups post graphs, calculations, observations, claims, diagrams, and unresolved questions. Other students physically move through the room, question the people who produced the evidence, annotate it, disagree with it, and return to their own work. The common object is not “content,” but part of a social process of inquiry.

A discussion board can record communication. It cannot manufacture community.

The problem, then, is not simply that schools use too much technology. The more consequential question is what happens when digitalization becomes a way of reorganizing the production of education itself.

Education is expensive largely because meaningful education is labor intensive. A teacher must spend time with students, interpret what they understand, notice what they misunderstand, design experiences, respond to unexpected questions, mediate disagreements, maintain a classroom community, evaluate work, and continually alter instruction in response to what is happening in front of them. Laboratories require equipment, space, preparation, maintenance, and enough adults and time for students to work safely and seriously. The job of education does not scale particularly well. A teacher can record a lecture once and distribute it to a thousand students, but a teacher cannot attend carefully to a thousand students at once. The limitation is not a technological defect. It has to do with the human character of education.

From the standpoint of efficiency, however, this is precisely the problem.

Digitalization offers the possibility of separating educational content from the teacher who produces and interprets it. Curriculum gets chopped into discrete units, recorded, standardized, stored, replicated, purchased, licensed, and delivered through platforms. And as student activity is translated into data, courses are organized as sequences of modules rather than sustained intellectual relationships among teachers, students, disciplines, and material environments.

What was previously an integrated form of labor is consequently unbundled. One person or team designs the curriculum. That company owns the platform. Another administers assessments. A less expensive employee supervises students, moving them along through the material. The teacher increasingly becomes a monitor, troubleshooter, or “learning coach” while the intellectual organization of instruction is transferred to software.

This possibility was recognized remarkably early by David Noble in his critique of what he called “digital diploma mills.” Noble argued that the automation of instruction is not merely a technological development but a labor process in which instructional knowledge is transferred from teachers into technological systems, increasing managerial control while making it possible to substitute cheaper labor for professional labor. This process is increasingly recognizable in K–12 schooling.

And in fact, digitalization means the de-resourcing of public education. A school deprived adequate staff and resources is presented with educational technology as a remedy for the conditions produced by its deprivation. The shortage becomes the justification for the platform. There are not enough teachers, so students can take an online course. There is not enough time for individual instruction, so adaptive software personalizes learning. There is not enough laboratory equipment, so students use a simulation. There are not enough counselors, tutors, or specialists, so automated systems provide guidance and intervention.

The technological solution appears to answer scarcity while leaving the scarcity itself politically unquestioned.

What a vicious circular logic. De-resourcing creates the conditions that make digital substitution appear necessary; digital substitution then becomes the justification for further de-resourcing. Once a district demonstrates that one teacher can supervise students working through centrally produced digital lessons, the labor-intensive alternative is labelled 'inefficient' rather than educationally valuable.

This does not mean that digital education is cheaper in any straightforward accounting sense. Technology requires devices, infrastructure, maintenance, licensing, technical support, upgrades, and increasingly recurring payments to private vendors. Noble himself warned against assuming that computer-based instruction simply reduces educational costs. The more significant change concerns where resources go and who controls the educational process. Money that might support stable professional staffing and publicly controlled institutional capacity instead is flowing toward platforms, software licenses, consultants, analytics systems, curriculum providers, and other forms of educational contracting.

This transformation also creates a commodity that can be reproduced at enormous scale. Stephen Ball describes a policy environment in which curriculum, pedagogy, assessment, school services, and even policy expertise increasingly are things that private organizations package, circulate, and sell. Educational problems are reconceived as markets for solutions. Digitalization dramatically expands that possibility by breaking down educational activity into standardized products capable of being distributed across schools and jurisdictions.

The unit of education changes accordingly.

Instead of beginning with a group of young people investigating a problem with a knowledgeable adult, the system begins with the module: a bounded instructional object containing content, an activity, an assessment, and a measurable outcome. The module is inserted into a course; the course inserted into a platform; the platform generates records showing completion and performance. Learning is reduced to things the system can count.

This is one reason that datafication and platformization belong to the same transformation. Ben Williamson has shown how commercial platforms, by organizing communication, behavior, assessment, and data collection, platforms can reshape the practices and relationships of schooling itself. Recent research on school platformization similarly describes apparent organizational efficiency mixed with intensified monitoring, surveillance, digital exclusion, and effects on teachers' working lives.

The danger is not that students spend too much time looking at screens. It is that a particular model of education becomes embedded in the architecture of the screen.

The language of “personalization” deserves particular scrutiny. Genuine personalization is extraordinarily labor intensive. It means knowing a student well enough to recognize what she understands, what interests her, when she needs encouragement, when she needs contradiction, and when an apparently incorrect answer reveals an interesting line of reasoning. Digital personalization means something quite different: algorithmically selecting the next standardized item on the basis of previously recorded responses. The first expands a human relationship. The second individualizes consumption.

A student learning mechanics does not merely need the correct sequence of information presented at the appropriate difficulty level. The student needs something to happen that requires explanation. Someone claims that a heavier object should fall faster. Another student disagrees. The class devises a way to test the claim.

Those moments are inefficient.

They take time. Apparatus must be distributed and sometimes repaired. Measurements fail. Students misunderstand instructions. Groups argue. The teacher moves from table to table asking different questions. A lesson may take a direction that was not encoded in the lesson plan.

But that inefficiency is not waste surrounding the education.

It is the education.

This is why the technologies in this physics course make sense to me only when they enlarge that human and material process rather than replace it. Video analysis allows students to interrogate motion they cannot easily measure by eye. For example, satellite tracking places classroom mathematics in relation to an actual object moving hundreds of kilometers above the Earth.

Technology expands what students and teachers can investigate together.

That is fundamentally different from using technology to make the teacher, laboratory, classroom, or intellectual community less necessary.

The distinction I am arriving at is therefore not between traditional and technological education, nor even between face-to-face and blended learning. It is between technology that augments human educational capacity and technology that substitutes for it. One begins with students, teachers, materials, questions, and the world and asks what tools might extend their powers of investigation. The other begins with the problem of delivering education more efficiently and asks which portions of human labor can be standardized, automated, monitored, outsourced, or eliminated.

Those approaches may use some of the same devices.

They represent very different ideas of what a school is for.