We build books for a child we will probably never meet. She is in Class 9. She lives in a Tier-2 town, or a village an hour past it, and she goes to a Hindi-belt government school or a low-fee private one with a hopeful English name over the gate. She reads English — she can decode it, parse it, answer questions in it — but she does not live in it. English is her second language, worn over a first. When we design a page, she is the person on the other side of it. Everything we make either reaches her or it doesn’t.
A while ago we were checking a chapter that opens with a simple game. The line said something like: the stone-paper-scissors hand-game you’ve played since you were small. It’s a warm line. It assumes intimacy — a game everyone knows in their hands before they know it in words.
We ran it past our quality checks, including AI reviewers built to catch exactly this kind of thing. Every one of them waved it through. One literally annotated it: Rock-Paper-Scissors — universally familiar, no issue.
That verdict is wrong, and it took us embarrassingly long to see why. The reviewers weren’t being careless. They were being consistent — consistent with a reader who was not our reader.
Every readability tool works the same way: it rewards short sentences and common words. By that measure, the line scores beautifully — small words, simple syntax. And completely beside the point. Readability measures the shape of a sentence. It is silent on the one thing that decides whether a line lands: does the reader recognise what it’s pointing at? Rock-Paper-Scissors is three small words and zero meaning if you’ve never played it — and familiarity with it doesn’t track age or intelligence or reading level. It tracks exposure: growing up around Western media, American school stories, a particular online childhood. Our reader’s childhood had kabaddi, gilli-danda, lagori, the odd-even toss before a cricket match. Not this.
Here is the part that unsettled us most: familiarity with that game tracks almost exactly the privilege these books are built to reach past. The line wasn’t just unfamiliar to her. It was unfamiliar in a way that quietly sorted readers by background — and welcomed the ones who needed the book least.
The deeper lesson is that “familiar” is not a property of a sentence at all. It’s a relationship between the sentence and a specific reader. “Is this familiar?” is an incomplete question; the complete question is “familiar to whom?” And every time the “to whom” is left unspoken, something fills it in anyway. A human reviewer fills it in with themselves. An AI reviewer fills it in with the centre of gravity of its training — the global, English-speaking, online default. Either way, the unspoken reader is the privileged reader. Not out of malice. Just because that’s who is standing there when nobody specifies otherwise.
So the fix was not a better metric. It was a named reader. We wrote her down in real detail — the things she reliably knows (the school bell, the canteen, the kirana, the weekly market, carrom, the games that actually fill an Indian recess) and the things she may not (lockers, prom, a “fall semester,” and yes, Rock-Paper-Scissors) — and we now make every judgment from inside her rather than from above her. Not “is this familiar?” but “would she recognise this, unexplained?” The point of writing her down is to stop every evaluator, human or machine, from quietly substituting itself.
Did it work? The test that matters is the one that hurts: shown the original line — the warm, score-perfect line every reviewer had blessed — the recalibrated check caught it immediately. From inside the right reader, the line lights up: she hasn’t played that game “since she was small.” She may never have played it at all. The chapter hadn’t changed. Only the vantage had. The vantage was everything.
One nuance keeps this honest. The tempting conclusion — cut everything unfamiliar — is lazy in the other direction. That game, it turns out, sits in the official curriculum; the syllabus leans on it. So the rule we settled on is small and load-bearing: the examples we are free to choose, we choose from her world. The ones the curriculum fixes in place, we keep — and teach, in two friendly lines, assuming nothing. Unfamiliar isn’t a death sentence. Unfamiliar-and-unintroduced is.
We think about that Class 9 reader constantly. She is willing. She is able. The only thing standing between her and the idea on the page is whether the page was written for her — or for the reader every tool assumes by default. We’d rather build for her. But first we had to stop letting our own instruments forget she existed.