How India's bureaucrats turned a promise to count caste into a masterclass in quiet sabotage
IN MAY 2010 India's ruling Congress party did something it had resisted for decades: it agreed, under pressure from campaigners, to count caste in the national census. Newspapers hailed it as a breakthrough. Five years later the data had vanished into a bureaucratic fog, officially declared too messy to publish. No minister ever said no. No law was repealed. The count simply disappeared, one procedural memo at a time.
This is the pattern that Trina Vityatil, a sociologist, calls "bureaucratic deflection": the art of saying yes in public while making sure, in private, that nothing changes. Rather than refuse a reform outright—an act that invites scrutiny and blame—bureaucracies bury it in paperwork, ambiguity and administrative drift. The reform survives on paper. It dies in practice. India's aborted caste census is as clean an example of the technique as social science is likely to find.
Numbers are never neutral
It is tempting to think of a census as a mere tally, a bloodless exercise in counting heads. Scholars know better. Bruce Curtis, a historian of statistics, describes census-taking as, in his words, "identifying political subjects and centralising knowledge"—a way for the state to decide which differences among citizens matter enough to record. Categories are never given in advance; someone chooses them, and the choice has consequences. What a census counts becomes real in a way that what it ignores does not.
In India this makes caste data uniquely dangerous to the status quo. Satish Deshpande, a sociologist, has written of an elite ideology of "castelessness", in which the inherited advantages of upper castes are recast as individual merit, while the disadvantages of lower castes are treated as their defining feature. A caste census threatens that fiction. It would show, in hard numbers, how thoroughly privilege clusters at the top. Little wonder that a bureaucracy staffed disproportionately by the upper castes has, for generations, quietly resisted collecting the data that might expose it.
Stage one: exile
The sabotage began almost as soon as the political promise was made. Between 2010 and 2011 officials at the Registrar General's office and the home ministry manoeuvred the caste count out of the main census entirely, merging it instead with an unrelated poverty survey to create something called the Socio-Economic and Caste Census (SECC).
The reclassification looked technical. Its effects were not. The regular census operates under a robust legal mandate with dedicated staff and a synchronised national timetable. The SECC had none of that: it was handed to different ministries, ran on no fixed schedule, and took 21 months to complete in Kerala and 46 in Rajasthan. Responsibility, and with it accountability, evaporated the moment the count changed its name.
Stage two: design by omission
Once caste had been shunted into the SECC, its architects set about ensuring the data would be useless. The questionnaire dropped the two categories—"Other Backward Classes" and "General"—needed to sort more than three-quarters of the population into any policy-relevant group. Everyone else was funnelled into a catch-all "Other" box, for which no preset list of caste names was ever compiled, and no fixed spellings were specified. A tick-box marked "no caste/tribe" offered a tidy escape hatch. And nobody bothered to explain how caste should be recorded among Muslims and Christians, communities in which caste hierarchies persist but map awkwardly onto Hindu categories.
None of this was accidental incompetence, in the argument of critics: it was incompetence with a purpose.
Stage three: enumerators fill the gaps
Vague instructions do not stop fieldwork; they simply hand discretion to whoever is left holding the clipboard. In India's case that meant poorly paid, poorly trained enumerators and data-entry operators, who resolved the ambiguity according to their own assumptions.
In affluent neighbourhoods, many simply ticked "no caste" whenever a household objected to the question, quietly encoding elite discomfort as official fact. In several Muslim-majority areas, enumerators marked "no caste" for entire streets on the assumption that Muslims have none, or recorded sectarian labels—Sunni, Shiite, Roman Catholic, Methodist—in a box meant for caste. A survey that could not tell an enumerator what to write ended up recording whatever the enumerator already believed.
Stage four: too messy to matter
The final trick was circular but effective. Having designed a form destined to generate chaos, officials then pointed to the chaos as proof the exercise had failed. The SECC produced 4.6m distinct caste entries, with up to 40 spellings for a single caste in some states. In 2015 the government published the survey's economic data and shelved the caste figures indefinitely, citing unreliability—even after an expert panel found only 1.1% of entries actually contained errors. The data that might have quantified caste privilege was, in the end, simply never released.
A pattern, not an accident
What makes bureaucratic deflection effective is precisely that no single decision looks like sabotage. Renaming a survey, dropping a category, skipping a training session, tolerating sloppy fieldwork—each step is defensible in isolation. Only in sequence do they add up to the quiet strangulation of a policy that the political class had already agreed to.
Campaigners draw an obvious lesson: reforms that threaten entrenched hierarchies cannot be entrusted to the hierarchy's own administrators. They are pushing for independent oversight bodies, staffed by anti-caste scholars and activists, to design the next count, vet its categories and audit its data before officials get another chance to lose it. Whether India attempts a fresh caste census remains an open political question. What is no longer in doubt is how the last one was allowed to fail.
This article is unfolding the reality. Discerning!
ReplyDelete