Is it possible that both Nick Gibb and Lucy Powell are right?

Gibb, the Highlander of school ministers, has written a blistering piece in the Telegraph, taking issue with the new Education Secretary’s claims that the school system built by Michael Gove and Gibb from 2010 onwards led to the present crisis of high and growing young adult unemployment. Others have entered the field with rich insights well-worth reading, with which, for what it’s worth, my own previous work is largely sympathetic.

But, like others on “my” side of the fence, I am also torn – because while I still believe in what I wrote in 2018, I am deeply troubled by the fact that our education system is still not achieving the outcomes we all desire. Whether progressive or traditionalist, no one ought to be happy with a system in which so many young people are unable to find valuable and rewarding employment, and so many report profound dissatisfaction with their experience of schooling.

So, I agree with Nick Gibb, but I also think Lucy Powell has a point. And I worry that the sound of them so fiercely disagreeing will drown out the cries for help from the system we actually need to listen to.

This is not some glib Third Way-ism, where I am going to suggest that deep down everyone agrees and if only they could sit by the fire and sing Educational Kumbayah, all would be well. There are real differences about policy that really matter here. But there is also an issue that people are, to some extent, talking at cross-purposes – not about totally different things, but about different aspects of the system. And the fact that education is a system is part of the problem.

To explain what I mean, I want to draw a distinction between what is merely complicated and what is genuinely complex.

A complicated thing – a clock, say, or a car engine – is a set of moving parts fitted together to produce a specific goal. Complicated things may not be simple, but they are relatively easily explicable and replicable. You can deconstruct a clock to literally find out what makes it tick, removing pieces one at a time until the tick stops and being able to conclude “it’s that bit”. With enough patience, you can put it back together so that it ticks as it did before, or adjust the clockwork so it ticks faster, or add aspects so it chimes the hour. Fundamentally, even in beautifully complicated clocks, you have something predictable – you can know through analysis what its parts do, what role they play in achieving the desired outcome, and how to adjust them.

A complex system is something else.

Complex systems are abstractly deconstructable, but practically indivisible. You can think through how a system operates and attempt to draw conclusions about its parts, but you cannot take it apart and test your assumptions, because the parts adapt – they respond to each other and to the interventions you make, so the system you are trying to improve is never quite the same system you started with; in a sense, that system broke the moment you changed something and it is therefore difficult to draw clear conclusions about which bit does what. A clock does not rearrange its gears in response to being wound, but in complex systems, the elements work together to generate outcomes in a way so interdependent that removing any single piece changes the whole without revealing reliably what that piece did and how it worked.

To see what I mean, consider the standard primary school experiment of growing cress.

Every child is charged with looking after a small box of dirt in which, the teacher assures you, there are cress seeds. But you do not have full autonomy as to how you treat the cress – one pot will be set upon the windowsill in full sunlight and diligently watered every day; another will sit beside it but receive no water; two more will find themselves living in a closed cupboard, opened occasionally so that just one of them can be watered. After several days, the sun-kissed and well-watered cress has grown, and everything else is dead.

If my experience is anything to go by, the major thing you learn is that cress is very dull, and that it is generally best to be on the team that locks plants in the dark and doesn’t water them because that’s a lot easier to manage.

But generally, students are meant to conclude that both water and light are necessary for a plant to grow.

This is not a conclusion that could have been reached by dissection. The individual plant was not the object of study – instead it was the growth system, the interaction between light and water that enables growth. Changing either variable killed the plant, so the cause of growth is not water or light, but something synthesised between them. The system had to be studied as a whole. And arguably, there is still more to do – had the teacher also created groups with seeds but no soil, the children would learn that there is at least one further necessary but not sufficient contributor.

Now, cress-and-cupboards is a simplification – children learn that multiple factors may need to be present to generate successful outcomes, but a teacher kicking the activity off can be very confident of what will happen to each pot. The failed growth may be multi-causal but the interaction (or not) of the variables is confidently predictable – but in complex systems, the elements can and do interact in ways which result in the system as a whole behaving in ways that no study of its individual parts would predict.

There is little chance the cress will resent being put in the cupboard, organise the other pots in protest at lack of water, or bloody-mindedly evolve to grow without soil halfway through day three, but equivalents of such unplanned outcomes happen in complex systems all the time.

I set this out because education is a complex system, not a complicated mechanism. It is not possible to break down into tiny pieces the mind of a child and build it howsoever you wish, identifying a precise cause and effect of every interaction. The mind is a complex system, and so too is the framework we build to help shape it.

This matters because it explains, somewhat, why Powell and Gibb are furiously disagreeing.

The new Labour government have looked at the system of education and noted that many students do not seem to enjoy it very much, do not appear able to routinely access non-university-based higher learning, and do not appear to be filling roles in the economy conducive to improving its health. They think this is a problem, and they have concluded that the curriculum is the cause – but this is like running the cress experiment and concluding that water is irrelevant to plant growth because some of the watered plants still died.

On the other hand, defenders of the current system are pointing at the flourishing cress on the windowsill and proclaiming the whole system works because it grew, even though all the other pots are bare.

This is not a relativistic argument that says both sides have equal merit. Some knowledge just is more powerful than other knowledge: mathematics is the universal language, physics explains the interactions of everything from suns to sticks, being able to read is the gateway to agency over your future knowledge acquisition. I don’t think the choice is either/or, but in a crowded curriculum, I would want more time given to groundings in science and the disciplinary thinking that explains why the applied world works as it does.

That said, the victory of a knowledge-rich curriculum has clearly not been sufficient to give all young people the agency over their lives which is their right. That does not mean I agree with progressives that we should privilege generic skills over disciplinary thinking – I see clear evidence, now and historically, that this was even less successful.

But it does mean that those of us who believe in a knowledge-rich curriculum need to consider what else is necessary but not sufficient for the achievement of that agency. International evidence on curriculum coherence – the work of Tim Oates and others on the multiple control factors that determine whether an education system actually delivers on its intentions – suggests that getting the curriculum right is essential but never enough on its own. You have to attend to assessment, accountability, teacher quality, institutional autonomy, and a dozen other factors simultaneously, because they interact as a system. Optimising one while neglecting the others is self-defeating.

Gibb was certainly attentive to the implications of Oates’ work – his policies did address multiple factors: curriculum, yes, but simultaneously assessment, accountability, teacher training, school structures. It is one reason the reforms achieved genuine gains.

But attending to multiple factors is not the same as attending to all of them, or to the interactions between them, nor does it deal with issues which arise when the system you want to affect is lodged in and being heavily influenced by a bigger, supra-system.

The educational control factors most conspicuously under-examined in the past fifteen years have been the ones hardest for a single department to control: the welfare system’s interaction with student motivation, the housing and transport costs that shape whether young people can access education at all, the employer behaviour that determines whether qualifications convert into good work, and the local institutional ecosystem that determines whether there is a coherent pathway from school to further learning to employment in any given place. These are not educational variables in the narrow sense – they are the soil in the cress experiment, the factor that nobody thought to test because it did not seem to be part of the same system.

But it is, and its absence kills the plant just as surely as darkness does.

It often feels to policymakers deeply unfair that their intentions so rarely translate into their desired outcome, but complex systems have a way of generating their own purposes, different from those who established them or any actor within them, but made up of the synthesis of all of them.

The British management cybernetician Stafford Beer’s most enduring contribution is often rendered as “the purpose of a system is what it does” – and if that is so, the purpose of the current English schooling system is not the agency and wellbeing of all its students. But it is a mistake to conflate the purpose the interdependent system develops with what its creators intended. Powell suggests Gove and Gibb wanted a world in which a million young people were unemployed – that simply is not true. But equally, Powell and her boss, Andy Burnham, do not want a world in which young people are blocked from the powerful knowledge that their more privileged peers consider their birthright.

They are all wrestling with a system so complex that the organic growth and interaction of its parts has bred something no one intended and no one is happy with.

The lesson of systems thinking is not that the problem is insoluble – it is that focussing on individual elements, even when that achieves significant success, rarely drives system improvement without proper attention to holistic effects, unintended consequences, and perverse incentives. Whatever successes curriculum change has resulted in, we still have to account for the pots that didn’t grow – and that means taking seriously the possibility that the factors which were not reformed are undermining the factors that were.

On the other hand, few changes ever yield only positive outcomes, and even fewer do so at exactly the time necessary to clinch a political argument. We have to learn to determine both our risk-tolerance for failure and the timescale we think reasonable for things to turn around, and to build the accountability frameworks that will help us learn how the system works and how it can work better. Ripping up the curriculum to fix a problem caused by welfare policy, housing costs, and employer behaviour risks substituting activity for impact, yielding worse outcomes overall.

After all, if everyone who took the lessons of the primary plant experiment to heart went out and bought a peace lily, they’d swiftly find that the watering which gave them a beautiful crop of cress rots the roots of their new acquisition.

Systems are necessarily complex. Our policy discussions to improve them must be so too.