Too Loose, Too Tight, Not Quite Right

So I came across an OECD blog post recently arguing for something fairly intuitive.

Teachers should have professional autonomy, but within shared frameworks and clear goals, so that students get something like a fair, consistent education regardless of which classroom they happen to land in.

It’s the ultimate “No shit, Sherlock” thing Schleicher could say. Self‑evident, tautologous almost. But I suppose the real interest lies in how things are actually being designed for autonomy and accountability, especially at a time when we keep saying we want more agency for students and teachers.

When autonomy floats free

In my fee‑paying private school, teachers tend to have a lot of freedom. They design units, choose resources, set their own routines. On paper it looks like trust, and in some ways it is. But underneath, that autonomy can sit on surprisingly weak structures. Curriculum guidance is loose. Shared planning depends on who has time and energy that week. Classroom observation is rare, or very surface‑level when it does happen.

I mean, there are policies and processes. They tend to show up as meetings, emails about parent complaints, and “we’ll keep an eye on this” conversations. Many concerns get technically addressed without being genuinely followed through. It feels like accountability without consequences, and when the school’s reputation is cushioned by fees and branding, the pressure to act decisively on weak practice is also, honestly, quite low.

From a student’s point of view, this can quietly turn learning into a lottery. Two classes in the same year group can feel like entirely different schools. Families who can afford to buffer that variation with tutoring or by moving schools do. The system lives with it because it can.

But if we are serious about designing education around agency then a system where so much still depends on which teacher you happened to get is already falling short before the lesson even begins. How much of student experience in a given school still comes down to luck? And is anyone actually tracking that?

When accountability narrows the work

And then there is the version I recognise from higher‑stakes public systems (which I have also spent time in). Here, the centre sets a clear curriculum, aligns assessments, and watches results closely. External exams matter for students’ futures and for schools’ reputations. Teachers still have autonomy in how they teach, but it sits inside common syllabuses, shared assessments, and regular evaluation.

Research drawing on PISA data from 27 OECD countries suggests that pairing school‑level autonomy over curriculum content with external exit exams can improve both student performance and equity, particularly for lower‑income students. So at the system level, there is a real upside to this model.

Inside that kind of system, the experience is genuinely different. There is a stronger sense of collective responsibility. Expectations are clearer. Students are less at the mercy of classroom‑to‑classroom variation.

But. When exam results dominate, teaching tends to narrow. Autonomy gets channelled into hitting existing targets more efficiently: better coverage, cleaner exam technique, safer choices. Work that actually matters for character, relationships, student voice, the work that builds agency in young people, struggles for time because it does not feed the main accountability indicators.

I mean, autonomy is still there, technically. But it mostly shapes how well you perform on a fixed scorecard, not whether the scorecard is asking the right questions or serving the right people. And what are the things a school says it values that do not show up anywhere in how it actually reviews or responds to practice? That gap is usually telling.

So where does that leave us?

Honestly, I do not arrive anywhere clean on this one. And I must caveat that my view is just one slice of a much larger picture.

I am not convinced by calls to simply trust teachers more. When autonomy sits on weak accountability, students get exposed to luck, and the families who cannot absorb that have no good options. But I am equally wary of calls for more accountability that really mean doubling down on exams and data. The OECD’s own research is more specific than that debate usually gets: not all forms of autonomy work the same way, and not all accountability is equity‑enhancing. Autonomy over curriculum content appears to help equity; full autonomy in hiring, without broader accountability structures, appears to reduce it. That kind of nuancing tends to get lost.

So maybe we need to shift the question itself. Rather than “how much autonomy?” or “how much accountability?”, something more like:

Autonomy for what? Accountability for what?

These feel like design questions to me. If education is being redesigned around agency, for students, and also for the teachers trying to build it, then getting this right matters. Teachers whose professional space is either totally unmoored or narrowed to test optimisation are probably not in a good position to help students develop theirs. I mean, its not hard to imagine that the conditions teachers work in tend to reproduce themselves in the conditions students learn in.

A system serious about agency would probably look different in practice. Teacher autonomy would be real, but anchored in shared expectations for curriculum and how students are treated. It is not a ceiling, but a common ground that makes genuine professional judgment possible. Accountability would have breadth: not just scores and complaint logs, but also what students say about safety, respect, and whether their school is actually stretching them. And the honest question for any leader (myself most definitely) is where the current setup makes it easier to look away from what matters, and harder to protect the things that do not show up in the data.

At this point, I am aware that saying “it’s a design problem” risks becoming its own kind of high‑level hand‑wave (the ultimate “No shit, Sherlock” directed to me). It is very easy to declare that we need better processes for autonomy and accountability, and much harder to build them in real schools with real people and real politics. So I am not going to pretend I have a neat framework tucked away somewhere. I absolutely don’t

For now, I’m going to high-level hand-wave this off as a field note rather than a solution. I am still working out the nitty gritty of what that would actually look like. Would love to hear your thoughts on what this balance ought to be.