Could a national exam survive its own paper leaking?
Every year more than twenty lakh students sit NEET-UG on one printed form, so the whole exam rests on a single sheet staying secret. Keeping any one artifact safe at that scale is a genuinely hard ask, and it is nobody’s fault in particular when it slips. Re:Neet is an open attempt to reason toward an alternative one idea at a time, and it is a lot more fun to think through together. Try a preset to feel the format, then read the specification.
Why should one sheet carry so much?
When a single identical form decides tens of lakhs of ranks, the whole exam has to travel a long physical chain of custody, and it stands or falls with that one sheet. In 2024 a leak was found to be localised and the Supreme Court saw no systemic breach, yet the 2026 cycle was still cancelled and re-run, because once trust in the single form slips, there is little to do but sit the whole exam again. This feels less like a story of blame than a sign that we are asking a lot of one piece of paper.
So here is a question worth sitting with. Could we keep exactly what the exam measures and still remove that single point of failure? One idea we will follow: draw every paper late from a large, calibrated bank, so a stolen sheet would tell you almost nothing.
If everyone sees a different paper, are the papers still fair?
It is a fair worry: once no two candidates share a form, those forms still have to measure the same thing. A sampling algorithm can assemble every paper to one blueprint, and raw scores can become percentiles within each shift, so a rank reflects how ready you are rather than which paper you happened to draw. JEE already runs this normalization today, and shared anchor questions would be the gentle upgrade that sharpens the very top and comparisons across years.
Computers mean shifts, and that turns out to be okay
India cannot seat twenty lakh candidates at once, so a computer-based exam would run in shifts, much as JEE already does. At an illustrative three lakh concurrent seats, over twenty lakh candidates fit into about eight sessions across three or four days, with a paper fallback wherever certified centres are not ready yet. The interesting part is not the seats, it is making scores fairly comparable across shifts.
What if anyone could help write it?
Picture a national question commons where anyone can contribute candidate questions, gently capped at 100 per mobile number a year so no single coaching shop can flood it. The pool could be split into chunks, each vetted by a different institution, so no committee ever sees the whole set, and a month before the exam the entire pool could go public for practice. A fuller version would add a one-way air gap after intake, quarantine, and expert rewrite, so live questions are unpublished variants and a contributor could never tell whether their question was used. Those firewall stages are the roadmap we are working toward, not what runs today.
A few choices are worth measuring, not debating
The difficulty mean, the marking scheme, and how much rote learning is really worth are all judgment calls, so rather than defend them on a slide, the design measures them on the live bank. If you picture a student’s recall as the slice of a public pool they can cram in a month, something surprising falls out: make the pool large enough and memorization stops predicting the score at all, while genuine understanding still ranks. It is far more convincing to watch that happen than to be told, so please poke at the knobs yourself.
60% → 2%
Recall you would need to win a seat as the public pool grows from ten thousand to a quarter-million questions, holding a month of cramming fixed
150,000
Questions you would have to memorize to stay competitive at that larger pool, which is beyond anyone in a single month
2.5%
Seat share we model, close to NEET's fifty thousand among twenty lakh candidates
Follow along as the bank grows
Occasional notes on new questions, presets, and open practice releases. No spam.