The problem: I want to re-imagine secondary schools in England. My proposal: keep the building, the 9-to-4 day, the age bands and the exit credentials (GCSEs, A-levels). Move all academic learning into one supervised, personalised hour a day on Zenjin, an adaptive app that trains distinctions one same/different judgement per card, at each child's own pace. Give the rest of the day to collective, unassessed group time: sport, drama, choir, instruments, building things. Teachers become coaches of it. Everyone has special needs, and age is a poor metric. Zenjin's live content today is GCSE and IGCSE maths and science. How might we make this work?

Cold trial of the Seed v1 (job #849): pasted into a fresh session with no files, no web search and no knowledge of CE.

2026-11-02: part 4 re-split under the Seed's cold-first rule, Seed v1.1. The opening now starts with a step anyone with an AI can take now; the run's original asks of real people are kept, in its own words, under Waits on a person. Later the same day the cold step was re-cut to use only public material: it had leaned on Zenjin's rung graph, which is private, so the graph now waits on a person too.

Re-imagining secondary schools in England with a daily Zenjin hour

My read: the shape is right, and the physics doesn't stand in its way. What decides whether this works is one question nobody has answered yet: does one supervised hour a day of distinction training produce exam grades? That answer is what lets every other part of the plan follow, so it has to come first. The second gap is coverage. Zenjin is live only for maths and science today. So don't switch a whole school over at once. Run the new route beside the old one, start where people are already queuing, and let the results from the first group of pupils earn each next step.

1. Purpose and the present arrangement

Purpose: every child leaves with credentials that open doors, real fluency at their own pace, and years spent making things with other people. No cohort ends up worse off.

The present arrangement:

About five hours of whole-class teaching a day, paced to the middle of the class.
Maths and physics teaching are in chronic shortage, and specialist posts often go unfilled.
Persistent absence is still far above pre-pandemic levels.
Demand for special needs plans rises every year, and budgets are cracking under it.
Arts entries and music departments have been shrinking for a decade.
Accountability rests on Progress 8, which counts eight GCSE-type results, and on Ofsted.
Most secondaries are academies. They don't have to follow the National Curriculum, only a "broad and balanced" one. Much of what feels like law here is a set of habits.

The physics, stated plainly:

Running a personalised hour costs almost nothing. A laptop drawing about 20 watts for an hour uses about 20 watt-hours, well under a penny of electricity.
The content is a pattern, and patterns travel light. Each extra child costs close to zero.
Devices are cheap at school scale. Staggered across six slots, a 1,000-pupil school needs about 170 devices. That is tens of thousands of pounds of hardware against a budget of several million.
Heat goes where it goes now: through the building's walls and roof. Heating the hall is the real energy bill, and the plan doesn't change it.
No physical wall blocks this plan. Two of the five walls do touch it:
Rates: a cohort takes five years and adolescence happens once.
What was never recorded is gone: crafts, music and practical skills die when the last teacher of them leaves.

The guardrail here is a cohort's doors. A child who misses grade 4 in maths and English at 16 loses options that are slow and painful to win back. That is the irreversible loss to design around. Everything else can be tried.

2. Four different arrangements

A. Whole-school switch: your proposal, all at once

Opens: the largest possible change, a clean story, and teachers freed from preparing five lessons a day.
Risks: it bets a whole cohort's GCSEs on an untested claim about transfer, while English, humanities and languages have no Zenjin content yet. It hits Progress 8 in year one and gives every critic a target.
Makes next easier: only if it works.
Label: imagined.

B. A school within a school

One volunteer Year 7 or Year 10 stream, with families who opt in, runs the full model.
The rest of the school runs as normal, and the stream's mocks are compared with matched peers.
Opens: real evidence, staff who chose it, a fallback inside the same building, and full consent.
Loses nothing for good, because pupils can move back to the normal timetable.
Label: designed.

C. Start where people are queuing, with the subjects that exist Put the supervised Zenjin hour wherever a maths or science gap already hurts:

Further education colleges running compulsory GCSE maths resits. Most resit students don't get grade 4, and it's the same specification Zenjin already covers.
Schools with no physics teacher.
Alternative provision and the persistently absent. Good alternative provision already looks like your model: a short burst of personalised academic work, then making things.

Opens: it pays on day one, it produces the transfer data A and B both need, and it costs schools nothing they will miss. Label: the engine is prototyped, since it is live. GCSE outcome data is not in yet.

D. The frame-breaker: the afternoon belongs to the town, and the exam belongs to the child

The town's commons. Collective time is opened to the community. Retired engineers, choir leaders, joiners, mechanics, grandparents with languages and crafts come in and coach alongside the teachers. That idle capacity starts moving again.
The library of skills. The school films and records every craft it teaches. Over time it becomes a lending library of skills against wall 4.
Exams by readiness. On the academic side, credentials stop being tied to age. A child sits the exam when the rung is held, early or late, like a driving test.
Opens: the most of any option, for the most people, including adults outside the building.
Label: the community part is observed working in other settings. Exams by readiness is imagined, because it waits on Ofqual and the exam boards.

3. Where things pull against each other, and the wider frame

"One hour can't cover ten GCSEs."

This is true for production: essays, proofs, spoken language, practicals.
It is plausible for discrimination: concepts, vocabulary, method choice.
The wider frame: Zenjin trains knowing the difference, and the collective day trains doing.
Drama is English literature performed and written.
Choir is languages and music.
Building things is physics, design and technology, and maths applied.
A school newspaper or radio station is English language.
The two halves aren't competing for time. Each one is half of what learning needs.
Treat the single hour as a design target to measure. If the data says older years need more before exams, add a second pre-exam hour. Defend the shape of the day, not the number.

"Unassessed" versus credentials that need coursework.

Science has required practicals, and art, design and technology, drama and music GCSEs have non-exam assessment.
The wider frame: the school never grades the collective time, but the work is real. The concert, the play and the built object face a real audience, and that is the test.
A child who wants the credential enters the portfolio they have already made. The required practicals happen in building time and get logged.

Extended writing has no same/different card.

It has a sibling card type: which of these two is better, and why?
Pairwise comparative judgement is already observed working for marking writing reliably.
Pupils making those judgements are training their taste. That is the bridge into English, history and the essay subjects.

Teachers' subject identity versus coaching. Split the role in two:

Distinction coaches read the live map of misses and run small clinics for children stuck on a rung. Every miss is a diagnosis, so this is deeper subject work than delivering lessons, not less.
Ensemble coaches run the collective time.

Planning and marking for five lessons a day mostly disappears. That workload is the main reason teachers leave, so the change pays teachers on day one. Keep it opt-in, and keep contracts and directed time unchanged.

Screens versus phone-free schools.

Frame it honestly: one supervised hour on a locked school device with no notifications, and a phone-free rest of the day.
That probably means less screen exposure than now, not more.

Age bands versus pace.

Keep age bands for the social day, where they work.
Let the academic hour ignore age.
One rule to watch: performance tables count a pupil's first entry in some subjects, so early entry needs a deliberate policy.

Progress 8 and Ofsted.

These are regulators and incentives, not physics.
The newer report-card inspection weighs attendance and personal development. The collective day should score well there.
Answer the headline measure with the data from Arrangement C.

"Everyone has special needs."

Adaptive pacing makes differentiation the default rather than a bolt-on.
Some children need other channels into the cards, such as audio for visually impaired pupils or different motor access. Build those in from the start, not as exceptions.

4. The pick and the grafts

Pick C, then B, then A, with D's commons grafted on from the start.

C opens the most because it answers the key unknown while paying the people who host it. B turns that answer into a whole-day proof with families who chose it. A becomes a decision you make on evidence rather than a bet.

Grafts:

From D: community coaches and the skills library, starting now, since they cost little and guard what is slow to rebuild.
From the tensions above: the comparative-judgement card type as the route into the essay subjects.
Mocks as the external check, in every group of pupils, every term.

What to return

1. Changed arrangements

a) The supervised Zenjin hour as a service where maths and science gaps already hurt

What changes: further education resit groups, schools without a physics teacher, and alternative provision units each get one supervised adaptive hour a day, with a distinction coach reading the live miss map.
Physical account: grid electricity of about 20 Wh per child-hour. What loops: the content, copied free, and the children's miss data, which sharpens the card ladder. What leaks: attention and motivation, and devices at end of life, which can go back into the loop through refurbishment. Heat leaves through the building as now.
Opens: grade 4 for students repeatedly failing it, cover for unfilled posts, and the transfer evidence everything else waits on.
Label: prototyped.

b) The split day, as one volunteer stream in one school

What changes: the Zenjin hour in the morning and a collective day after it, run by coaches, with mocks compared against matched peers.
Physical account: the same building, the same heating and the same staff hours, redistributed. What loops: the performance cycle of rehearse, show and rehearse. What leaks: pupil and staff goodwill if it is run badly.
Opens: a full-day proof, a template other schools can copy, and a fallback in the same building.
Label: designed.

c) The afternoon as a town commons, plus a library of skills

What changes: community members coach alongside staff after safeguarding checks, and each craft taught is filmed and recorded.
Physical account: human time that is currently idle. What loops: skills, and the recorded pattern. What it guards against: crafts dying with their last practitioner.
Opens: purpose for retired people, mentors for children, and a record against wall 4.
Label: observed working elsewhere in other forms; designed here.

d) Comparative-judgement cards for the writing subjects

What changes: "Which of these two is stronger?" sibling cards for English, history and religious studies, calibrated by teachers' pairwise judgements.
Physical account: the same as (a). This one is all pattern and no atoms.
Opens: Zenjin's route beyond maths and science, and pupils' taste trained the same way as their knowledge.
Label: the method is observed working for assessment; as training cards, it is imagined.

2. One step now

Approach one further education college this week, or one school with a physics vacancy. Offer a supervised Zenjin hour for one maths resit or Year 11 group, free for this term. In return, ask for permission to compare each pupil's held rungs against their January mocks.

The college gains practice for a group it is struggling to move and relief for a stretched department.
Zenjin gains the first transfer data, early enough to calibrate before the summer 2027 exams.

3. One long route

Exams taken when ready, on screen.

GCSE-equivalent exams would be sat when a rung ladder is held, not at a fixed age. They would draw on banks of exam questions and be taken under secure supervision. Meanwhile Zenjin's content grows across English, humanities and languages through comparative-judgement and listening cards.

What it waits on:
Ofqual accepting secure, adaptive on-screen exams. - Exam boards building calibrated question banks. - Evidence that held rungs predict exam performance. - Card types for production skills: writing, speaking and proof.
Possible backers:
An exam board looking for the next exam format. - A science-education foundation such as Gatsby. - The Education Endowment Foundation, to evaluate independently. - The Department for Education, once the resit data exists, because resits are its most expensive failure.

Expect five to ten years.

4. One opening for the next person

Build the transfer map for GCSE maths.

For each specification point in a published GCSE maths specification (AQA, Edexcel and OCR all publish theirs free), write out the chain of distinctions a pupil must hold to answer it, each as one same/different judgement a card could ask. This can be done now from the specification and Zenjin's public description alone, without a single student and without seeing inside the app. Where a specification point cannot be reduced to a chain of single judgements, the map already shows a missing multi-step production card. The map is the thing any rung graph then gets checked against.

Waits on a person

Zenjin's own rung graph, to lay the map over it and list which rungs feed each specification point. The graph is not published, so this needs someone at Zenjin.
Test, with a few hundred students on past-paper questions written on paper, what level of holding a rung predicts a correct written answer. Two outcomes are possible:
- The rungs predict well. One hour a day has its evidence, and the map becomes the progress report parents and Ofsted read.
- They don't. The map shows exactly which multi-step production cards are missing.

5. A changed premise

Proposed addition, Claude (Opus), 27 October 2026.

Premise 13: Attention is the human flux. In human systems the binding resource is rarely energy. It is attention: a child's, a teacher's, a parent's. Like sunlight, it is a daily income, not a stock. It replenishes overnight and is wasted if it isn't spent well. The design question is what arrangement spends it where it compounds.