What changed
- Daily practice is 20 questions, about twenty minutes. Your child can choose 10 instead on a short day.
- Bootcamp’s weekly set is 50 questions, a longer look back over what the week taught. Your child can choose 20 instead. There is no timer, and they can stop and carry on later in the week.
How we tested it
We can’t try a change on real children and wait a term to see if it helped, so we test it first in a simulation: a computer model of how children answer questions and learn. We ran the same month of practice at different lengths, and compared the results.
The tables show the difference, not raw scores. “25% more” means a quarter more than the shorter session managed; “About the same” means the difference was too small to matter.
Twenty a day compared with ten
| After a month | With twenty a day |
|---|---|
| Real mistakes spottedMistakes the child really makes, named correctly by the app. | 25% more |
| Real mistakes fixedMistakes the child really made and had stopped making by the end of the month. | 29% more |
| Questions at the right levelHard enough to stretch the child, not so hard they are lost. | About the same+2% |
| Progress madeHow much the child improved over the month. | About twice as much+103% |
At twenty a day, the app spotted 25% more of the mistakes children really made, and fixed 29% more. The questions stayed at the right level for each child, even with twice as many of them.
Progress roughly doubled too. That is mostly because the simulation is built so that practice brings progress, so twice the practice gives about twice the progress. We read that row as a check that nothing went wrong, not as a discovery.
Fifty a day compared with twenty
| After a month | With fifty a day |
|---|---|
| Real mistakes spotted | 20% more |
| Real mistakes fixed | 35% more |
| Questions at the right level | No change |
| Questions answered correctly | About the same+1% |
| Progress made | About two and a half times as much+150% |
A longer session fixes more mistakes again (35% more), and just as many questions stay at the right level. Progress rises with practice for the same reason as before.
Why twenty a day, and fifty a week
A simulation can’t see tiredness, boredom, or a child who stops wanting to come back, and those matter more than any table. We think twenty minutes is something a ten-year-old can keep up every day, and keeping it up is what counts. So twenty is the default, and ten is always there.
The weekly set is where a longer session pays off. It has no timer, it can be split across the week, and it is where the week’s learning gets checked.
Keeping a longer session balanced
A short session only has room for a few kinds of question. A long one can tip over: too many checks on old mistakes at once, the same weak topic again and again, or one mistake chased all session. Kudos keeps the mix in proportion however long the session is: going back over mistakes, working on weak spots, and moving on to new topics. That is why the tables show questions staying at the right level as sessions get longer.
How we chose the way Kudos picks questions in the first place is in how we chose the question algorithm.
What this does not tell us
This is a simulation. Whether real children gain as much from longer sessions is what the app will show us over time, as it checks whether mistakes stay fixed, and as families tell us how the exam went.