Kerfox

OUR MISSION

The best way to revise is known. Almost nobody is taught it.

Fifty years of research says what actually moves a grade, and almost none of it reaches the person sitting down with a textbook the night before an exam. Kerfox is an attempt to hand a student the good version instead, free, from GCSE to a doctorate.

Why the gap is there.

The findings themselves are old, replicated and boring. Spread your practice out rather than stacking it. Answer from memory rather than re-reading. Mix the confusable things together rather than doing twenty of the same. Work through a solved example before you try one alone. None of it is controversial and all of it is the opposite of what revision usually looks like.

The reason it does not reach people is not that anybody is hiding it. It is that doing it properly is ADMIN. Knowing which topic is about to slip, keeping the two things you keep confusing next to each other, pitching the next question at the difficulty where you will just about get it - that is a scheduling problem, run daily, across everything you are studying. Nobody revising for six subjects is going to keep that ledger by hand, and they should not have to.

A phone is very good at scheduling problems. So Kerfox does the admin and hands back the studying: it decides what comes next and when, marks what you write against the marks a board would give it, and gets out of the way. What is left for you is the part that was always the point.

And it is free where it counts, which is not generosity so much as arithmetic. The students who most need somebody else to run the schedule are the least likely to be able to pay for one.

What we hold ourselves to.

Five of them, and every one names the page where you can check whether we are keeping it. A list of values nobody can audit is a list of adjectives.

  • Free where it counts

    Every lesson, every question and the memory engine behind them are free, and the plan is for that to stay true. Supporting us buys no feature, no seat and no priority - it buys the thing staying free for somebody else.

    Support us

  • Nothing ships unchecked

    42 separate checks run over the whole curriculum before a build is allowed out, and every one of them fails the build rather than printing a warning. A gate that warns is a gate somebody scrolls past.

    What gets checked

  • The method is public

    Every design decision here is traced to a published finding, cited and linked, including the several that cut against us. If you want to disagree with how Kerfox teaches, we would rather hand you the argument than make you guess at it.

    The research

  • Difficulty is the point

    If it feels easy you are probably not learning much. Questions are aimed at the band where you get roughly eight in ten right, on purpose, because an answer you were certain of teaches almost nothing and one you could never reach teaches nothing at all.

    How the pairing works

  • We say what we do not know

    Nobody has run a study on Kerfox, so the page where an efficacy claim would go says so in its first line and puts a zero where the headline number belongs. We would rather be caught having said it than caught not having.

    Our own studies

Three levels, three different jobs.

Kerfox runs from GCSE to a doctorate, and it would be easier if that were one thing done at three sizes. It is not. What a student is being marked on changes completely on the way up, and the app has to change with it.

  • GCSE

    Where the mark scheme starts crediting particular words rather than correct ideas, which is a skill nobody teaches on purpose and everybody is marked on.

  • A-Level

    Written against the four boards’ own specifications - AQA, OCR, Edexcel and WJEC - rather than against a general idea of the subject, because a topic one board examines and another does not is a term of revision either wasted or missing.

  • University

    One shelf up, where there is no mark scheme to be faithful to and the job becomes derivations, real problem sets and the honest difficulty of the thing itself.

How it is made.

Content is written against the boards’ own published specifications and examiner reports, not from memory of a subject. Figures are simulated rather than illustrated: where a caption says the steps are the same length, the drawing generates them the same length, because a figure that merely looks about right is an argument nobody checked.

Then the machines get it. Constants are compared against their real values, question generators are run hundreds of times to prove a wrong option cannot equal the right one, the derivation marker is made to mark our own answers and then made to fail on broken ones, and the whole thing stops at the first disagreement.

Why we built it this way

WHERE IT IS UP TO

Kerfox is in its first testing round, and the waitlist is how the first seats go out. Nothing is charged for and nothing is promised beyond one email on the morning it opens.

Join the waitlist