Everyone who builds learning software eventually discovers the same humiliating fact: the product works beautifully in the demo and is never opened at home. The usual explanations are all about the artefact. The content is too easy, or too hard. The feedback loop is too slow. The interface is not delightful enough. Somewhere a roadmap gets rewritten, a difficulty curve gets retuned, and six months later the numbers look exactly the same.
I want to propose a duller explanation, and I think it is closer to true. In the middle years of primary school — roughly eight to ten — what sustains a learning habit is not the quality of the tool. It is whether a slot in the day exists, held open by someone else, with another human being physically present in it. The tool determines what happens in that time. The slot determines whether the time happens at all. These two can fail independently, and evaluation almost always measures only the first.
A natural experiment nobody designs on purpose
Look closely at the learning routines around any household with children of that age and you tend to find two arrangements running side by side. One is a fixed afternoon hour, an adult sitting nearby, a book, no metrics of any kind, no goals worth the name. The other is a good adaptive platform — genuinely good, with weekly targets, a focus mode, and a parent dashboard that would not embarrass a B2B analytics product.
The unglamorous one runs for months without being nagged. The instrumented one needs to be asked, every time, and produces a strange gap between minutes logged and minutes spent. On paper the second arrangement is better on every dimension a product team would name. It loses anyway, and it loses on a dimension nobody ships: the first arrangement had a container, and the second required the child to build one from scratch, daily, out of their own executive function.
Three things follow, and all three are unpleasant for people in my line of work.
Swapping tools does not swap containers. If platform A and platform B both require the child to start the session alone, moving from one to the other changes the content and leaves the bottleneck exactly where it was. Which is why the trial period is such a liar. Novelty and the feeling of being watched both produce a real improvement in behaviour, and both decay on a timescale slightly longer than a pilot.
Removing a container is a structural event, larger than any tool change. The person who sat nearby moves away. The schedule shifts. A sports season starts. What has been removed is not motivation, it is the slot in which learning occurred. And if that happens to coincide with the introduction of a new tool that runs on the child’s own initiative, the two events will hide each other perfectly: the tool’s failure gets attributed to the child not being self-disciplined, when the actual cause is that the hour no longer exists.
Rebuilding the container comes before adding work. Adding tasks when the container is gone is pouring water into a bucket with no bottom, with the added bonus of manufacturing a debt the child now feels bad about.
This is not an argument against autonomy
It sounds like one. A fixed hour looks like exactly the sort of external control that self-determination theory spent forty years warning us about. I do not think there is a conflict, because the two operate on different layers.
The container is structure: when, where, who is present. It can be supplied from outside, and supplying it costs the child nothing in autonomy. The content is choice: what to do, in what order, for how long. That layer belongs to the child. Collapsing the two produces the two failure modes you can watch in the wild — parents who conclude that respecting autonomy means providing no structure, so the container quietly disappears and learning stops happening; and parents who conclude that providing structure means owning the content too, so autonomy is confiscated and motivation stops happening. Both are the same mistake, run in opposite directions.
What this predicts
I like propositions that can lose. This one can.
If a container is removed and not rebuilt, platform-side data should degrade frequency first and session length second — whole days with nothing, rather than every day slightly shorter. If you see the opposite shape, the problem really is in the content, and I am wrong. And the vacated slot will be filled, fast, by something the child chooses, which then sets like concrete. Whoever fills it owns it, and that determines how hard rebuilding will be.
For adults there is a small operational consequence, which is that in this business presence is the active ingredient and instruction is not. The minimal viable move is to show up in the old slot, do your own work, ask nothing, check nothing, start nothing on the child’s behalf, and leave after ten minutes. What they choose to do in that time is both the test of whether the container still stands and the only honest input you have for designing what goes in it next.
We spend a great deal of effort making the ten minutes better. It might be worth spending some on whether the ten minutes exist.