46 Lakh Names, One Repeated Mistake: Census 2027's Caste Question

A Census Act That Never Mandated a Caste Taxonomy The Census Act, 1948,

gives the government wide power to ask citizens almost anything and compels them to answer - but it says nothing about how the answers to an open-ended question like caste must be structured or classified. That silence is not new; it is the same silence that swallowed the 2011 Socio-Economic and Caste Census.

What is new in Census 2027 is that India is choosing, deliberately, to walk back into it - for the first time attempting a general caste count since Independence, with the identical unstructured "open field" design that produced 46 lakh different caste names in 2011, names never sorted into any usable category and, ultimately, never released.

The

Rule as Written, the Rule as Practiced The rule as written is simple: SC and ST respondents pick from a state-specific drop-down list, because those categories are constitutionally defined and administratively fixed under Articles 341 and 342.

Everyone else "Enter[s] Caste name" freely, per Question 10(C) of the notified schedule. In practice, this means two respondents from the same community, using different spellings, sub-caste names, gotras or surnames, will generate entries that a computer cannot recognise as identical - precisely the failure mode the government itself cited when explaining why SECC 2011's caste data was withheld for "errors."

Where the NPR Shadow Falls

Layered onto this is the second half of the story: questions on parents' religion, birthplace, Aadhaar, mobile, voter ID and driving licence - fields that mirror the 2019 National Population Register rehearsal form almost exactly.

The NPR itself has no independent constitutional basis; it draws its authority from the Citizenship Rules, 2003, framed under the Citizenship Act, 1955 and its data has long been feared as the administrative backbone of a future National Register of Citizens.

Congress leader Jairam Ramesh's charge that the government has "buried" the caste census through methodology, while reviving NPR-adjacent questions through the back door, captures the actual constitutional tension here: two exercises with entirely different legal bases and objectives, run through a single mandatory questionnaire, without Parliament ever separately debating whether they should be.

The exam-relevant insight isn't that caste is being counted - that itself is a historic first. It is that the government had the administrative capacity to build a standardised caste taxonomy (as it did for SC/ST) and chose not to extend it to the rest of the population, meaning the technical failure of 2011 was a choice available to avoid, not a limitation the Registrar General's office lacked the tools to solve.