Author: Chanchal Yadav
College: Babu Banarasi Das University
To the Point
Ask most people when India last ran a full caste count, and the honest answer traces back to 1931. Every census since independence has tracked Scheduled Castes and Scheduled Tribes specifically, but the wider caste map of the country has stayed officially unrecorded for nearly a century. That is about to shift. On 30th April 2025, the Cabinet Committee on Political Affairs approved bringing caste enumeration back into the national census, timed for the 2027 exercise — which will also mark India’s first fully digital census. On its face, this reads as a bureaucratic update. Underneath, it sets two constitutional interests directly against each other: the government’s authority to gather data for planning welfare programmes, and a citizen’s right to keep something as personal as caste to themselves, a right the Supreme Court traced to Article 21 in its 2017 Puttaswamy ruling. What follows is an attempt to weigh both sides honestly, without pretending the answer is obvious in either direction.
Use of Legal Jargon
Before going further, a short glossary of terms that recur throughout this piece: fundamental right, reasonable restriction, proportionality test, legitimate state interest, informational privacy, data fiduciary, creamy layer, ultra vires, policy domain, and statutory backing. Each is explained the moment it appears, so a reader without a legal background should be able to follow the reasoning without stumbling over vocabulary.
Abstract
Nearly a hundred years without a fresh caste count has consequences. Reservation policy, OBC sub-categorisation, and welfare targeting have all leaned on data that is either badly dated or stitched together from uneven state-level surveys. The 2025 decision to reintroduce caste enumeration under the Census Act, 1948 is meant to correct that — replacing guesswork built on a 1931 baseline, which the Supreme Court has already called obsolete, with something statistically current. But there is a genuine complication. Caste in India is not a neutral demographic label; it carries generations of social hierarchy and discrimination behind it, and turning it into a compulsory field in a centralised, digitised government record raises fair questions about what becomes of that information afterward. This article traces the statutory origin of the census, applies the Supreme Court’s three-part privacy test from Puttaswamy — legality, necessity, proportionality — assesses how much real protection the Digital Personal Data Protection Act, 2023 provides, and reviews how the Supreme Court has actually ruled across a series of 2026 orders. The conclusion, in short: the census clears the legal bar for its own existence, but the harder question of what happens to the data afterward has been left open rather than answered.
The Proof
1. The Statutory Foundation Behind the Census
No new law was needed to set this in motion. The Census Act, 1948 already gives the Union Government broad power to collect demographic and social information, caste included, through enumerators sent house to house. This rests constitutionally on Entry 69 of the Union List, which places census-taking squarely under Parliament’s exclusive jurisdiction. The April 2025 Cabinet decision, then, was not fresh legislation — it was the government choosing to use an authority that already existed. That distinction gives the 2027 exercise sturdier legal ground than earlier attempts, notably the Socio-Economic Caste Census of 2011, which was run through the Ministry of Rural Development rather than the Census Act, and whose results the Supreme Court later dismissed as too unreliable to act on.
2. Why Privacy Becomes Relevant Here
None of this privacy debate would exist without the 2017 Puttaswamy judgment, so it is worth pausing there. A nine-judge bench, unanimously, held that privacy sits within Article 21’s guarantee of life and personal liberty. The judgment did not, however, treat privacy as an absolute shield. It set three conditions for any state action that restricts it: a valid law must authorise it, that law must serve a legitimate purpose, and the method used must be proportionate — meaning the least intrusive option that still achieves the goal. Caste does not fit the usual categories of sensitive data like health or financial records, but given its historical link to social stigma, most legal scholars still treat it as sensitive personal information. A mandatory, nationwide, door-to-door disclosure of caste — with penalties under the Census Act for refusal or false declaration — is exactly the kind of state action the Puttaswamy test was designed to scrutinise.
3. Measuring the Census Against the Puttaswamy Standard
Is there a law authorising it? Yes, without much dispute — the Census Act covers this directly.
Does it serve a legitimate purpose? Very likely. Better-targeted welfare delivery, a defensible basis for OBC sub-categorisation, and reservation figures grounded in something more current than 1931 all align with concerns the Supreme Court itself raised in Indra Sawhney and later in Ashoka Kumar Thakur, where it criticised reliance on stale caste data.
Is the method proportionate? This is the point on which opinions genuinely diverge. Opponents argue that assembling one large, centralised, digitally linked caste database — ahead of a fully mature and tested data-protection regime — is not obviously the gentlest way to collect this information. The disagreement is not about whether to count people; it is about safeguarding what gets collected once the counting is done.
4. Where the Data Protection Framework Falls Short
The Digital Personal Data Protection Act, 2023 received presidential assent in August 2023, though its provisions have come into force in stages — most from November 2025, with a few not activating until 2027. Section 17 is the relevant provision here: it exempts certain government functions tied to sovereignty, security, and public order from the Act’s usual consent and notice requirements. One clause stands out — Section 17(2)(a) — which allows the Central Government to exempt any “instrumentality of the State” from the Act in its entirety, through nothing more than an executive notification, without any case-specific justification required. Several legal commentators have noted that this sits uneasily against the Puttaswamy proportionality standard, and against the more carefully calibrated safeguards the Court itself insisted on decades earlier in the telephone-tapping case, PUCL v. Union of India, later reaffirmed in Anuradha Bhasin. The practical upshot: the census has clearance to proceed, but the very statute meant to guard the resulting data still carries an unusually broad exit clause.
5. What the Courts Have Actually Ruled, 2025–2026
The Supreme Court has been asked more than once to halt this exercise, and each time it has refused. In February 2026, it declined to take up a petition challenging the procedure for recording and verifying caste data. In April 2026, a bench headed by the Chief Justice dismissed a plea seeking a direction to stop the census entirely. In May 2026, another PIL was rejected, with the Chief Justice observing that the matter “exclusively comes within the policy domain” and that there was nothing constitutionally objectionable about the government identifying backward communities through census figures. What these rulings have not yet addressed is the data-handling side of the exercise — retention periods, access restrictions, security design — under a genuine proportionality review. Each order so far answers whether the census may happen, not whether it is being conducted with adequate safeguards. Separately, the Patna High Court has already upheld Bihar’s 2023 caste survey as constitutionally valid, though that ruling remains under appeal before the Supreme Court.
Case Laws
1. Justice K.S. Puttaswamy (Retd.) v. Union of India, (2017) 10 SCC 1 — The nine-judge bench that recognised privacy as part of Article 21 and established the legality-necessity-proportionality framework now applied to any state action touching caste data.
2. Indra Sawhney v. Union of India, 1992 Supp (3) SCC 217 — Upheld 27% OBC reservation subject to an overall 50% ceiling, and emphasised that identifying backward classes needs periodic revisiting rather than a one-time determination.
3. Ashoka Kumar Thakur v. Union of India, (2008) 6 SCC 1 — Upheld OBC reservation in higher education with the creamy-layer exclusion, while directly criticising continued reliance on 1931 caste figures as outdated — an argument that indirectly supports the case for a fresh census.
4. People’s Union for Civil Liberties (PUCL) v. Union of India, (1997) 1 SCC 301 — Built procedural safeguards into telephone-tapping powers to protect Article 21 against unchecked surveillance; regularly cited as the missing template the DPDPA’s government exemptions lack.
5. Anuradha Bhasin v. Union of India, (2020) 3 SCC 637 — Refined the proportionality standard further, requiring the State to adopt the least intrusive option reasonably available — a bar critics argue Section 17(2)(a) of the DPDPA fails to meet.
6. Patna High Court, State of Bihar & Ors. (2023) — Upheld Bihar’s caste-based survey as valid, holding its purpose to be the legitimate identification and upliftment of backward communities; the matter is presently on appeal before the Supreme Court.
7. Supreme Court orders, February through May 2026 — A series of dismissals of PILs against caste enumeration in the 2027 Census, consistently treating the underlying decision as an executive policy matter rather than one requiring judicial override.
Conclusion
Set aside the politics, and the legal position is relatively clear: the caste census has a genuine statutory basis in the Census Act, and the Supreme Court, every time it has been asked, has declined to interfere with the government’s decision to conduct it. That much is settled. What remains unresolved is everything that follows the enumerator’s visit — how the data is stored, who is permitted to access it, how long it stays on record, and whether the DPDPA’s fairly generous exemption powers would hold up against a focused proportionality challenge if one were ever brought directly before the Court. Framing this purely as “privacy against welfare” understates the issue — most observers would likely agree that accurate caste data serves a legitimate function. The real fault line lies between a defensible policy objective and a data-protection framework that has not yet demonstrated it can manage information this sensitive, at this scale, responsibly. As 2027 approaches, the legal contest will probably move away from whether the census should happen at all and toward whether it is being carried out with proper care — and that is where Article 21’s privacy guarantee will face its more meaningful test.
FAQ
Q1. What separates a caste census from the reservation system?
A caste census simply records how many people belong to each caste. Reservation is the actual affirmative-action mechanism under Articles 15(4), 16(4), and 46. The census can shape how reservation policy is designed, but it does not itself grant or modify any quota.
Q2. Is it constitutionally acceptable for the government to record people’s caste in a census?
Broadly, yes. The Census Act, 1948 provides clear authority for this, and the Supreme Court has treated the decision as falling within executive policy. That does not, however, amount to unconditional approval — how the data is gathered and later handled still has to satisfy the Puttaswamy test.
Q3. Does the Article 21 right to privacy prevent the State from collecting caste data altogether?
No. Privacy is a fundamental right, but a qualified one — it can be restricted where a valid law exists, the purpose is legitimate, and the means are proportionate. The same reasoning applied to other privacy-restricting state actions applies equally here.
Q4. Does the Digital Personal Data Protection Act, 2023 genuinely guard caste data against misuse?
Only in part. It establishes baseline standards of necessity and proportionality, but Section 17 — particularly Section 17(2)(a) — allows the Central Government to exempt entire government bodies from the Act through a simple notification. The safeguard exists in principle, but with a considerable gap left open in practice.
Q5. Has any court actually stopped the caste census from proceeding?
No. Across 2025 and into 2026, the Supreme Court has consistently declined to halt or modify the exercise, treating it as an executive decision. It has not yet ruled on a direct challenge to the security of the data itself.
Q6. Why did the SECC 2011 numbers fail, and how is this attempt different?
SECC 2011 operated outside the Census Act, under a separate ministry, and ended up affected by overlapping classifications, manual errors, and incomplete coverage — the Supreme Court eventually deemed the data unusable. The current exercise runs directly under the Census Act, giving it firmer legal standing and, at least on paper, a more standardised digital process than SECC ever achieved.
