The government wants a caste census. The trouble is, it has exactly two ways to conduct one - and each has already failed once, in a different century.

The RGI Isn't Choosing Between Two New Ideas

• It's Choosing Between Two Old Colonial Failures In 1901, Herbert Hope Risley, the Census Commissioner of British India, tried to solve a problem that sounds strikingly familiar: how do you count caste in a society where caste identities are fluid, local and endlessly varied? His answer was to impose order from above - ethnographic surveys, anthropometric measurements and a rigid, ranked classification that took loosely held local identities and fixed them into permanent administrative categories. It worked, in the narrow sense that it produced clean data. It also did lasting damage, turning caste from a fluid social fact into a hard legal one - a transformation the RSS is explicitly invoking today when it objects to a drop-down menu.

The Same Two Failure Modes, a Century Apart

India's current dilemma is not a new problem looking for a new solution - it is the same 1901 problem, still unsolved, now presenting the RGI with the same two doors British administrators tried and found wanting. Door one is Risley's door: classify caste from above, using pre-set lists.

That produces order but, per the RSS's own argument, produces it by freezing a colonial-era architecture in place - exactly Risley's method, just run through a modern drop-down menu instead of an anthropometric survey. Door two is what happens when you refuse Risley's approach entirely: let people self-declare.

India tried that too, in 2011 and 46.7 lakh distinct caste names later, the government quietly shelved the raw data because no policy apparatus could use it.

Where the Parallel Breaks Here is the

difference that matters for 2026: Risley had the coercive authority of a colonial state and could simply fix categories by fiat, public objection notwithstanding. Today's RGI has no such freedom - the RSS's objection is being raised and heard before the Census even begins, in a way an 1871 or 1901 census commissioner would never have entertained.

That is a genuine gain in accountability. But it also means the RGI cannot simply pick the historically "cleaner" of the two options, because the cleaner option (a drop-down menu) is precisely the one facing the strongest political resistance for reproducing colonial logic - while the more "neutral" option (self-declaration) has already been empirically proven unworkable at administrative scale.

What this really tests is whether India in 2026 can invent a third method that neither colonial India nor the UPA-era SECC ever attempted - perhaps a verified, community-consulted classification system distinct from both Risley's imposed lists and 2011's unstructured chaos.

Until that exists, the RGI is not choosing between progress and colonial legacy. It is choosing which of two colonial-era failure modes to repeat and hoping the third time produces usable data without the political cost either predecessor paid.