Summary
The Hindu digitised Delhi's draft electoral roll from the Election Commission's Special Intensive Revision and estimated religion using a published name classifier. It found that Muslims, about 13.6 percent of electors by that estimate, were 20.3 percent of those flagged for "logical discrepancies".
The skew was sharpest on five reasons linked to family size. An unnamed senior election official told the paper that the flagging was done by the Commission's centralised ERONET software, not by Electoral Registration Officers.
The eight reasons were never formally notified before the Phase 3 direction of 14 May 2026 and have never been explained. The real issue is due process: an unpublished software rule has quietly displaced the statutory officer's judgment and placed the burden of proving eligibility on the elector.
WHY IT MATTERS FOR UPSC & STATE PCS
On 14 October 2026, The Hindu published a data analysis of Delhi's Special Intensive Revision draft roll. It showed that the Election Commission's ERONET software flagged electors for logical discrepancies at markedly different rates across communities, with the heaviest skew on family-size criteria. The Commission did not respond to the paper's questions.
Standard News
The Roll Belongs to an Officer
- the Flag Came From a Machine Article 326 promises the vote to every citizen aged 18 or above who is not disqualified by law. It does not say who decides, in a given case, whether that promise holds. The Representation of the People Act, 1950 answers that question. It gives the preparation and revision of each constituency's roll to an Electoral Registration Officer and Rule 20 of the Registration of Electors Rules, 1960 requires that officer to inquire into claims and objections. The design is deliberate: a human officer, a hearing and a reasoned decision. Delhi's revision shows what happens when that design is quietly bypassed.
The rule as written The law lists the grounds that remove a person from the roll: not being a citizen, being declared of unsound mind by a court or being disqualified for corrupt practices.
Eligibility otherwise depends on age and ordinary residence. In Lal Babu Hussein v. Electoral Registration Officer (1995), the Supreme Court held that an officer cannot strike a name or demand particular documents, without giving the elector a fair opportunity to be heard. Family size appears nowhere in that list.
The rule as practised According to an unnamed senior election official quoted by The Hindu, the "logical discrepancy" flags in Delhi were raised by the Commission's centralised ERONET software, not by EROs. Five of the eight reasons turn on family structure. The paper reports, for example, that one reason flags an elector when six or more electors named the same person as parent in the 2002 roll. The Hindu then estimated religion with a published name classifier, reported to be about 97 percent accurate. On that estimate:
- Muslims are about 13.6 percent of electors but 20.3 percent of those flagged for discrepancies.
- Under the five family-size reasons, 152 of every 1,000 Muslim electors were flagged, against 56 of every 1,000 others.
- Under the six-or-more-children reason alone, the Muslim share was 58.2 percent. The control case is telling. Bihar's revision did not use these checks and there Muslims were 18.4 percent of deletions against a 17.7 percent share of the adult population, roughly proportionate. The religion figures are an estimate, not recorded data. But the mechanism does not depend on intent. A proxy correlated with fertility will skew by community wherever fertility differs. NFHS-5 shows that gap narrowing, at 2.36 against 1.94, but not closed.
Where the
burden quietly moved A notice is not a deletion. It does something subtler: it requires the elector to prove eligibility on a criterion the law never created. The eight reasons first appeared in writing only in the Phase 3 direction of 14 May 2026, had been used in West Bengal before that without announcement and have never been publicly explained.
That is the constitutional gap. Article 324 gives the Commission superintendence over the rolls, but the statute places the judgment in an officer who must hear the elector. Neither text anticipated a filter that decides before any officer looks.
Quick Facts
Key numbers & takeaways — revise these first
-
Delhi's Special Intensive Revision draft roll holds 97,53,577 electors.
-
About 33.12 lakh electors received notices: 13.79 lakh could not be linked to the previous roll and 19.33 lakh were flagged for logical discrepancies under eight reasons.
-
By The Hindu's estimate, 318 of every 1,000 Muslim electors and 197 of every 1,000 other electors were flagged for logical discrepancies.
-
Under the "six or more children" reason, about 1.4 lakh electors were flagged and 58.2 percent of them were Muslim by the paper's estimate.
-
In Bihar's revision, which did not use the discrepancy check, Muslims were 18.4 percent of deletions against a 17.7 percent share of the adult population.
-
NFHS-5 (2019-21) puts the total fertility rate at 2.36 for Muslim women and 1.94 for Hindu women.
-
Article 326 provides for elections on the basis of adult suffrage.
-
Rule 20 of the Registration of Electors Rules, 1960 requires the Electoral Registration Officer to inquire into claims and objections.
Connect the dots for your UPSC preparation.
Standard news covers the event. Log in to read our comprehensive analysis and uncover the hidden constitutional, structural, and ethical dimensions of this topic:
Why none of the eight discrepancy reasons maps onto any statutory ground of disqualification and what that means if a notice ends in deletion.
How the Bihar revision, which did not use the discrepancy check, works as a natural control group and what its near-proportionate deletions reveal.
The Dutch childcare-benefits scandal, where a risk algorithm's neutral-looking criteria brought down a government and the parallels and limits for India's rolls.
A concrete reform package: publishing filter criteria, keeping EROs in the loop and auditing filters for disparate impact before use.
Included in this analysis
Join thousands of aspirants analyzing the news deeply.
Unlock Premium — Rs.699 AnnuallyDon't have an account? Sign up for free