Overview / Standards
Standing policy
Editorial standards
These rules are deliberately separate from any one study, so they can be held against every page on this site — including pages written later, and including pages written by someone else. They were set before the findings, not after them.
What counts as evidence
Every claim traces to a book, an edition and a printed page. A score without cited evidence is not a score, and is not published.
The corpus is graded into three tiers and each claim inherits the tier it rests on. Native text — a publisher's own text layer — supports exact quotation; only 15 of 167 current books have one. English OCR supports quotation after re-verification by exact string match against the extracted text. Urdu and Sindhi OCR is theme-grade: it supports statements about structure, chapter coverage and proper nouns, and it is never quoted at the sentence level, anywhere, for any purpose. That tier covers 76 of 167 current books — see what is missing.
The consequence is stated plainly rather than buried: where a claim depends on something not appearing in Urdu-medium material, this site says not established, never absent. Degraded Nastaliq OCR cannot distinguish the two, and pretending otherwise would let the method's weakest point produce its strongest-sounding claims.
Quotation and copyright
Passages are excerpted for criticism and analysis, with citation, and page images are reproduced as evidence for specific claims. That is ordinary scholarly practice and it is how the argument is made. The textbooks themselves are not republished. They carry explicit reservations of rights, and free distribution to schoolchildren is not a licence to redistribute. Every record instead carries a full bibliographic entry and a credited link to the source the copy was obtained from.
Counter-evidence
Where a board contradicts itself, both sides are shown, in the same section, at the same prominence — not in a footnote and not in a closing paragraph. A finding that has counter-evidence and hides it is a worse finding, not a stronger one.
The same applies to negative results. A tested hypothesis that fails is published with the same prominence as one that succeeds; two of this project's longest analyses report that the thing they went looking for is not there.
What this method cannot show
It cannot show what any child believes. The corpus shows what is taught, what is repeated, in what order it arrives, and what pupils are required to reproduce in an examination. It does not observe classrooms, actual examination papers, or what any pupil takes away. Where an argument depends on an unobserved step, that step is named as unobserved.
Arguments about consequences are made from structure — what recurs across subjects and years, and what is examinable — never from inferred psychology. Predictions are labelled as predictions and are published with the evidence that would refute them.
A computed inventory of every gap — unfound books, unreadable text, unscored axes and excluded material — is published at what is missing, and is regenerated from the catalogue on every build so it cannot fall out of date with what the study holds.
Numbers
- No headline number without its denominator and its date. Axis coverage is uneven by design, and a mean over two books is not a board figure. Denominators are printed beside every mean on this site, and thin ones are marked.
- Every figure is computed from the published dataset at build time, not typed into prose. This site has already been wrong once because a count was hardcoded; that route is now closed.
- No single ranking of the boards is offered. Two defensible composite indices disagree about the order, and reporting one alone would mislead. Both are published with their parts.
- Scores are measurements, not verdicts. A high score records a property of a text. It is not a judgement of a board, an author, a teacher or a province.
Corrections
Errors are corrected in place, visibly, and the correction stays on the record. Nothing is quietly patched. The corrections log lists every error this project has found or been shown, what changed, and whether any published claim moved — including the ones that were embarrassing.
A correction is logged whether or not it is convenient for the argument. One of the entries in that log withdraws a headline finding after a reader pointed out it was false.
Right of reply
Any board, author, publisher or individual described on this site may reply, and the reply will be published beside the finding it concerns — not on a separate page, not summarised, and not edited for content beyond length and legality. A reply does not require this project to agree with it.
If a factual error is identified, it is corrected and logged. If a characterisation is disputed but not factually wrong, the dispute is published alongside the characterisation and both stand. If material is asked to be removed, it will be removed, and the removal will be recorded publicly — what was removed, who asked, and when. That record is itself part of the reference work; a silent deletion would be a worse outcome for a reader than the original disagreement.
Write to [email protected]. The most useful challenge names the page, the specific claim, and what the correct position is — ideally with the board, book, edition and printed page.
Independence
This project has no funder, no institutional sponsor, no political affiliation and no commercial interest in its conclusions. It was not commissioned. Nobody has reviewed it before publication, which is a real weakness and is stated here rather than left to be discovered — the findings have not been through peer review, and the corrections log is currently the only external check that has operated on them.
If this project ever stops
Reference works fail badly when they disappear. If this one is no longer maintained, the commitment is that the URLs stay alive and the pages stay as they are, with a dated banner saying the work stopped and when. A stale reference that says it is stale remains useful; a dead link destroys every citation anyone ever made to it.