Managing the complexity of the classroom

complexitySystematic pedagogy is not defeated by the complexity of the classroom: it is the solution

In my previous post, Pedagogical romanticism, the ninth in my series on educational purpose, I proposed that our current model of education provision, which relies heavily on the intuition of autonomous teachers, was failing to manage the scale of modern education. But the alternative to the intuition of the human teacher is some sort of systematic pedagogy, which is commonly thought to be defeated by the complexity and unpredictability of the classroom. In this part I respond.

In response to my previous post, Pedagogical romanticism, I was pleased to receive a tweet making the argument that I knew needed to be answered next.


Procrustes was an Athenian bandit who invited travellers to stay the night, only to murder them in the small hours, fitting their bodies to their bed by stretching them on the rack if they were too short or amputating any protruding limbs if they were too large. The charge is that systematic forms of education tend to fit the student to the system instead of the system to the student, failing to take into account the natural variability of the classroom.

It is an argument that needs to be answered in two parts: one with respect to what we teach children (the ends of education) and one that relates to how we teach them (the means). This post is about the second of these: managing variability and complexity in our pedagogy.

In his 2007 keynote to the Association for Learning Technology, Dylan Wiliam argues that systematic approaches to education are defeated by the complexity of the classroom.

There was a craze in America a few years ago for perfect teaching, where they would give these teachers scripts, you know, designed by experts on how to teach really well. And they were really scripts, there were things like “Now, walk around the classroom” [laughter]. And the point is, they were useless because classrooms are effectively chaotic places. Actually, even well behaved classrooms are chaotic places, in that the difference between one course of action and another course of action is so small that it’s effectively only described well by chaos theory.

Chaos theory is commonly pictured as the process by which the flap of a butterfly’s wings might cause a hurricane weeks later and thousands of miles away. The metaphor represents causal processes that create a massive divergence of end states based on infinitely small differences in initial conditions. Yet if you were to look for meteorological metaphors for the classroom, the best fit would more often be the doldrums than the hurricane. In terms of “state”, most classrooms are frustratingly stable: one lesson looks very like another and three quarters of the students still do not understand what you taught them three months ago.

The unpredictability of chaotic systems is also dependent on a lack of purposeful monitoring and intervention. Dylan Wiliam is best known for his research on formative assessment, arguing persuasively for precisely this kind of “cybernetic” approach to teaching. Systems that are intentionally regulated by cybernetic feedback are not chaotic.

Even without such corrective control mechanisms, it is no longer thought that students learn in radically different or unpredictable ways. Attacks on Howard Gardner’s theory of learning styles are now so common that it would be superfluous to repeat them here.

In arguing for systematic pedagogy, Brian Simon recognised that this approach had fallen out of fashion because of teachers’ belief in the extreme variability of their students (p.42):

If each child is unique, and each requires a specific pedagogical approach appropriate to him or her and no other, the construction of an all-embracing pedagogy, or general principles of teaching becomes an impossibility.

Such child-centred pedagogies have recently come under sustained attack. The source of variability on which Dylan Wiliam bases his model of the chaotic classroom is quite different (p.4):

what gets learned is actually very, very difficult to predict…the SESM projects found that typically in teaching one third knew the content at the beginning, one third didn’t know it at the end, so only one third learned the content, and half of these had forgotten the content six weeks later [laughter].

This variability divides the class into only two parts: those who have “got it” and those who haven’t. It is hardly a picture of chaotic unpredictability.

It is true that those who have not “got it” are likely to have fallen into a variety of different conceptual errors, so the class is likely to become more fragmented than suggested by a “got-it/not-got-it” model; but the number of likely conceptual errors is generally small and their nature is predictable. For example, when comparing fractions, Wiliam notes that:

a lot of the kids have the naïve strategy that the biggest bottom makes the smallest fraction.

The classroom is not hyper-divergent after all. The nature of its fragmentation is modest and relatively predictable. We may not have systematically mapped the different misconceptions that students commonly adopt—but there is no reason why this should not be done, once we have specified our objectives.

The fact that the classroom is relatively predictable does not mean that it is not complex: it means that its complexity is of a different sort.

I call complexity “diagnostic” when it involves the need to match a relatively small number of cases with solutions that are numerous or poorly defined. This is the challenge facing the fictional detective struggling to solve a seemingly impossible crime or a psychologist trying to reveal the deep, repressed trauma of a disturbed patient. It might be the challenge facing a teacher who needs to respond to a highly original piece of student work.

I call complexity “logistical” when one must match a relatively large number of cases to solutions that are relatively few and well defined.


In most cases, the complexity of the classroom is of the second kind. The teacher in a mass education system is faced by large numbers of students who require constant feedback, mostly of a predictable kind. Sometimes, it is true, students require original, “hand-crafted” feedback and that is one reason we need highly skilled teachers, and these are the challenges that will keep such teachers motivated and usefully employed. But for most of the time, the feedback required is mechanical, predictable, relentless.

Dylan Wiliam has himself been a relentless advocate for formative assessment (I see a picture posted by @LearningSpy on Twitter of him presenting yesterday at ResearchEd in Washington, promoting the same message that he was promoting 20 years ago).


I believe in the potential of formative assessment as much as Professor Wiliam. I accept that its potential is backed by research evidence. But one cannot help noticing that it has not been successfully deployed at scale—and not through want of effort or financial support.

I suggest that the failure of formative assessment is attributable to our current, decentralized model of educational delivery: the romantic view of teacher-as-autonomous-craftsman, which Wiliam himself supports. Our failure to describe educational process in systematic terms means that we consistently fail to meet the logistical demands made by formative assessment.

Digital systems (that are available to help us manage systematic processes) are good at handling logistical complexity, poor when faced with diagnostic uncertainty. Teachers have been bamboozled by false theories of child-centred education into thinking that their problem was diagnostic uncertainty when in fact it was logistical complexity. They mistook complexity for uncertainty, supposing that what was unmanageable must therefore be intrinsically chaotic.

The Procrustean challenge conceives of technology as a script, a straight-jacket, even an implement of torture. It is supposed to de-professionalize the teacher and constrain the student to follow standard procedures. Yet with today’s technology, this is the opposite of the truth. It was a hundred years ago that at the Battle of the Somme, British and Commonwealth soldiers under human officers marched towards their enemy in unwavering, inflexible ranks. Non-digital classrooms still march in what Americans have for at least 50 years called a “lockstep”. It is not humans operating on their own initiative, but modern digital systems based on explicit descriptions of systematic process that are operating at scale to personalize almost every part of our lives. Every part, that is, except education, where serious forms of digital management have never been attempted.

In my next post I shall consider the challenge of Procrustes from the perspective of how we set our educational objectives.

2 thoughts on “Managing the complexity of the classroom

  1. Interesting! I think the “logistical challenge” diagram accurately summarises the reality in a typical classroom. Digital systems undoubtedly have great potential, as you describe. Sadly, however, I have been unimpressed with most of the examples I have encountered so far. That is not to say that some form of logistical “scripting” could be useful (
    To my mind, most teachers follow the approach of Wellington rather than Napoleon: their battleplans are made of little bits of string, so that if one piece breaks, they simply knot two other ends together. That said, if anyone can put together even a partial system that saves time and effort, I think the teaching world will beat a path to their door!

    • Many thanks.

      I completely agree with you that the examples of digital tech that are currently available are not generally up to much. Towards the end of this series, I will propose a reason for that and a way that the problem can be solved.

      I like the knotted string analogy. In some ways, I don’t think this is a bad thing. Large and inflexible courses do tend to be rather Procrustean and the way that I envisage a new edtech market working is for the software industry to provide teachers with flexible and disaggregated activities, which teachers can apply, aggregate and share as they see fit, plugging into replaceable management systems for assignment and tracking. This not only avoids top-down prescription, it also ensures a dynamic and flexible market, avoiding commercial lock-in. The key enabler will be interoperable data standards – a bit obtuse & techie from the point of view of the educator but important for these reasons.

      I also agree with you that, given appropriate labour-saving tech, teachers will bite the industry’s hand off. And that also, I suggest, is why this approach ought to be highly attractive to politicians, whose only approach at the moment seems to be to impose tougher exams or more rigorous inspections, whipping the dead donkey, as it were. Initiating a market for the sort of professional “tools of the trade” that teachers need and want, seems to me to be a win-win, if only politicians understood how to make it happen.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s