How this site's answers are checked
Every previous-year answer here is checked against two independent sources before it is published. Where they disagree, or where the official record itself is unclear, the question is held back rather than guessed. This is what that means in practice, and where the line actually sits.
Two independent answer keys, not one
Each transcribed question carries two answers from the source material: the coaching institute's own printed key, and the official NTA answer key published separately. A script checks every question in the bank against the official key before it can be published — not a spot check, every row, every time content changes.
When the two sources disagree
The official NTA key wins. It has happened on a handful of questions across the papers on this site — the coaching key was simply wrong on those — and the worked solution says so explicitly rather than silently using the "expected" textbook answer.
When NTA itself couldn't settle it
A handful of questions each year are dropped from the final key entirely — no option matches, or more than one does, and NTA marks every candidate correct rather than picking one. Those questions are held back from this site rather than assigned an answer that isn't really settled. Right now that's 9 questions out of 1340 transcribed, sitting out for exactly that reason.
Formulas render, they aren't typed as text
Every question, option and worked solution passes through the same KaTeX pipeline before it reaches you — the source is Markdown with LaTeX, and what you read is typeset, not a wall of backslashes. A build fails if a formula doesn't parse.
A worked solution isn't assumed just because a question is published
1327 of 1331 published questions carry a full worked solution today; the rest show the correct option without one yet, rather than a placeholder pretending to explain it. Ask only ever quotes a passage that has actually been written.
Chapter weightage only counts complete papers
A paper held at a third of its questions would pull every chapter it happens to touch upward and every chapter it misses down, invisibly. Only papers held at 95% or more of their questions count toward chapter-wise weightage, and the coverage of every paper is shown on that page rather than hidden behind the final number.
What this process does not catch
- Transcription of the question itself. The answer is checked by script against the official key; the stem and options are typed from the source booklet and read back by a person, not independently re-OCR'd against a second copy of the paper. A transcription slip in the wording is possible even when the answer index is verified correct.
- The prose of a worked solution. The correctness of the underlying science is checked as carefully as any textbook explanation is checked — by a person who knows the subject — but there is no equivalent of the answer-key script for prose: no independent process re-derives each explanation from scratch and confirms it word for word.
- Coverage. Not every past paper is transcribed yet, and not every transcribed question has a worked solution. What is published is checked; what isn't published yet just isn't published yet.