What we know about learning and evidence, and what we built from it.
Each principle below is a finding from research on how people learn or how evidence works, the design on Uweza that follows from it, and what that design costs us.
Learning
Curiosity is strongest when a person can feel the gap between what they know and what they want to know, and weak when there is no gap to feel.
On Uweza: Every course starts from a real question, such as why a harvest rots before it reaches the market, and takes in each subject the answer needs.
What it costs us: A question-led course is harder to write than a syllabus, and slower to cover ground.
People learn most when a task is hard enough to need their full attention and not so hard that they give up.
On Uweza: Studio reads the first-try pass rate of every check. Checks between 30% and 85% are in the band; outside it, the author is told to look again.
What it costs us: Authors rewrite checks that already work for the strongest learners.
Remembering is strengthened by having to retrieve something after time has passed, not by seeing it again.
On Uweza: Weeks after a topic is shown, Uweza asks again on a date the learner did not choose, with a question written fresh.
What it costs us: A Retained entry takes weeks to earn; most of a new learner’s record is not yet the strongest claim.
Doing well during a lesson and still knowing it later are different, and conditions that make a lesson feel smooth can leave less behind.
On Uweza: The Passport shows Demonstrated and Retained as separate claims, and a reader can always tell which is which.
What it costs us: We cannot print a chart of lessons completed and call it learning.
Explaining an idea in your own words exposes what you do not yet understand, where choosing an answer hides it.
On Uweza: Checks ask for the reasoning in the learner’s own words, typed or spoken; a spoken answer counts the same.
What it costs us: Marking free answers needs care, and some answers need a teacher.
Using an idea in a new field rarely happens by itself; it needs the principle stated apart from any one example.
On Uweza: Before a capstone, the learner states the principle detached from both subjects it appeared in. Carried across is earned when they later apply it, unprompted, somewhere it was never taught.
What it costs us: It is the slowest evidence to accumulate; most learners have none for months.
People understand more deeply what they build and test than what they are told.
On Uweza: Understanding is shown at a workbench: a model whose behaviour the server re-runs before it counts, so describing it is not enough.
What it costs us: Benches take longer to build than quizzes.
Nearly every learner reaches mastery with enough time and the right help at the point they are stuck.
On Uweza: Amara does not move a learner on until they can explain it back, and appears after two attempts at a check.
What it costs us: Some learners take much longer, and the platform waits for them.
Rewards for doing something can replace the interest a person had in doing it.
On Uweza: There are no points, streaks, badges or leaderboards on Uweza Education.
What it costs us: We give up the quickest way to raise daily use.
Evidence
A number that people are rewarded for stops measuring what it was meant to measure.
On Uweza: Checks are hard to game: a lesson ends when a question is answered, not when a video finishes, and nothing is counted because it was clicked.
What it costs us: Our numbers grow more slowly than they would if we counted the easy things.
When a seller knows more than a buyer and cannot prove it, honest sellers are priced like dishonest ones.
On Uweza: Revenue counts as evidenced only when an operator’s message backs it, and every other entry is labelled, so a founder with real sales can show them.
What it costs us: An honest founder without messages looks weaker than they are, until they attach them.
Small groups are easy to re-identify, even without names.
On Uweza: A funder sees no median, share or average for fewer than 3 participants, and never a name.
What it costs us: A small cohort shows its funder almost nothing.
A system stays viable when a signal of trouble reaches the people who can act, without waiting for the next report.
On Uweza: A participant silent for 30 days rises to the top of the cohort list, with how long, and the rule that put them there is printed.
What it costs us: Staff are shown problems before they have been asked about them.
A record that depends on an institution disappears when the institution does.
On Uweza: The Passport belongs to the person and opens from a link anyone can check. An institution cannot own or withdraw a learner’s record.
What it costs us: We say no to institutions who are paying us and want control of it.
One position we do not explain
A child who signals they are not safe takes precedence over everything else on the platform at that moment. We do not describe how it works, because explaining the mechanism publicly would inform the wrong reader.