The Largest Thing India Does Well

Pritiraj BrahmaFounding Partner, Pritiraj & Partners · 1 February 2026

India moves 64.2 crore people through a polling booth in six weeks with 39 repolls — the most impressive administrative act on earth. In the year that followed, one state’s roll lost 47 lakh names, the reasons were published only under court order, and 9,968 deletions carried no reason at all. The machine that collects the vote is superb. The institutions that decide what it is worth have not been rebuilt in decades.

ProvenancePublished 1 February 2026; revised 28 August 2026 to argue from attributed figures throughout. Turnout and deployment figures are from the Election Commission’s statements of 3 June 2024 and its published dataset of December 2024, which differ on the participation total (64.2 crore against 64.64 crore); both are stated. Bihar revision figures are from Commission press notes and statements between June and October 2025, and deletion-reason counts from Bihar Chief Electoral Officer records cited in Supreme Court proceedings; the two summary totals in circulation, 47 lakh net and 69 lakh gross, describe different quantities. Electoral bond values are as furnished by the State Bank of India to the Court in March 2024. The Joint Committee timeline runs to 28 August 2026, at which date no report had been presented.

India moved 64.2 crore people through a polling booth in six weeks in 2024, with 39 repolls against 540 five years earlier. It is the most impressive administrative act on earth. In the year that followed, a single state’s electoral roll lost about 47 lakh names, the reasons were published only after the Supreme Court ordered it, and part of the ledger carried no reason at all. India has built a superb machine for collecting the vote and has not rebuilt the institutions that decide what the vote is worth.

Begin with the thing a critical journal should say first, because it is true. In the summer of 2024 the Election Commission of India conducted the largest voluntary logistical operation ever attempted, almost without incident. The operating rule is that no voter should travel more than two kilometres to vote, which means polling parties walk into forests, cross rivers and set up a booth for a single elector. This essay is about what surrounds that act — a register a citizen can trust, representation that reflects the present population, competition financed in the open. That gap is the subject of this journal.

Six findings anchor this analysis:

1. The conduct of the poll is at global standard, on the Commission’s own numbers. Briefing the press on 3 June 2024, the Chief Election Commissioner recorded 64.2 crore voters including 31.2 crore women, from an electorate of 96.88 crore; the Commission put final turnout at 65.79 per cent. Some 1.5 crore polling and security personnel and over 68,000 monitoring teams were deployed, and 39 repolls were held against 540 in 2019. The Commission’s published dataset of December 2024 gives participation as 64.64 crore rather than 64.2 crore; the difference is one of compilation stage and we state both rather than choose.
2. The register, not the booth, is where the system is now contested. Bihar’s Special Intensive Revision ran from 24 June to 30 September 2025. The roll opened with about 7.89 crore electors and the final roll carried about 7.42 crore — a net reduction of roughly 47 lakh names in a single state in fourteen weeks. The draft roll of 1 August had removed 65.63 lakh; by 30 September, 21.53 lakh had been added on Form 6 and a further 3.66 lakh removed. In October the Chief Election Commissioner described the exercise as having deleted 69 lakh voters, and announced that a pan-India revision would follow.
3. The reasons for removal were published only after a court required it, and part of the ledger carries no reason. On 14 August 2025 a Supreme Court bench of Justices Surya Kant and Joymalya Bagchi directed the Commission to publish district-wise lists of excluded electors on official websites by 19 August and to publicise their availability. Of 366,742 names deleted from the draft roll, Bihar Chief Electoral Officer records cited in those proceedings account for 62,441 as deceased, 212,999 as having shifted and 81,334 as duplicates — leaving 9,968 removed with none of those reasons recorded. Ten thousand names is a small share of the total and a complete answer to the question of whether the ledger reconciles.
4. A “final” roll was revised six days after being declared final. The Commission’s press note of 30 September 2025 gave the post-revision electorate as approximately 7.42 crore; its note of 6 October restated it as 7,43,55,976, adding 1,63,619 service voters. The correction is legitimate and small. That it was needed at all, days after publication and weeks before a state election, is a statement about the register’s tolerances rather than about anyone’s intent.
5. Representation has been frozen for half a century. The Lok Sabha has stood at 543 seats on the delimitation that followed the 1971 census. The Forty-second Amendment froze allocation in 1976; the Eighty-fourth Amendment of 2001 extended the freeze until after the first census taken after 2026. India therefore elects a house whose composition reflects a population distribution more than fifty years old, while the population has roughly doubled and redistributed itself sharply between states.
6. The financing instrument was struck down and not replaced. A five-judge bench of the Supreme Court struck down the electoral bonds scheme on 15 February 2024, holding that anonymous corporate funding violated the voter’s right to information under Article 19(1)(a). The State Bank of India had issued bonds worth ₹16,518 crore across 30 tranches since 2018, and told the Court that 22,217 bonds were purchased and 22,030 redeemed between 1 April 2019 and 15 February 2024. Two and a half years on, no disclosure architecture has been legislated in its place.

The register is the real battleground

Every democracy’s integrity rests on a list. India’s contains close to a billion names, must be maintained where tens of millions move for work, and must be inclusive enough that no citizen is excluded and accurate enough that no name appears twice. Those requirements pull against each other, and the tension is political rather than technical: from outside, a removed name is indistinguishable between a duplicate correctly deleted and a citizen wrongly struck off.

Bihar demonstrates both halves. The Commission reported in July 2025 that roughly 22 lakh electors had been found deceased, seven lakh enrolled in more than one place and about 36 lakh migrated or untraceable. A roll carrying 22 lakh dead names needed cleaning, and saying otherwise is not a serious position. But the same exercise produced two summary totals in public circulation — 47 lakh and 69 lakh — that describe different quantities and were never published with a reconciliation; a district-wise list of the excluded that appeared only under judicial direction; and 9,968 deletions for which the published categories give no reason.

The remedy is not a better algorithm. It is procedural, and the Court has already stated most of it: publish additions and deletions in full with reasons, at a granularity that permits independent verification; give a struck-off citizen a fast and real appeal rather than a form; and revise on a fixed statutory calendar rather than in the months before a contest. A roll revised in public on a known schedule is defensible. One revised close to an election will be contested whether or not it was done correctly — and the contest itself corrodes what the Commission has spent seventy years building.

Reform energy is spent on the calendar

The most advanced electoral reform in the country is an argument about when elections are held. The high-level committee chaired by former President Ram Nath Kovind reported in March 2024, having consulted 47 parties of which 32 supported simultaneous polls. The Bill followed in December 2024 and went to a Joint Committee whose tenure has since been extended three times without a report. The Commission has estimated a requirement of roughly 30 lakh integrated voting units.

There are respectable arguments for the change: cost, administrative disruption, and the effect of a permanent campaign on hard decisions. But note what the sequencing reveals. Twenty months of parliamentary process have gone into the frequency of elections. Over the same period, the register lost 47 lakh names in one state and gained a court order; the financing architecture struck down in February 2024 has not been rebuilt; and the freeze on representation approaches its expiry with no settled successor. The reform that commands legislative energy is the one that does not require the political class to constrain itself.

The counter-case, honestly stated

Three arguments cut against this analysis, and the first cuts against this Review’s founding claim.

First, the election may not generalise. The lesson drawn here — that the Commission succeeds because the task has an unambiguous deadline, a single accountable authority and a result that cannot be fudged — may not transfer. An election is a short simultaneous event with a binary output, staffed by officers seconded from other departments, under continuous adversarial surveillance by competing parties with both the means and the motive to publicise any failure. Schooling, electricity and public health are continuous services with contested quality measures and no adversarial monitor with a stake in exposing shortfalls. That the Commission’s lesson lifts into sectors lacking an adversarial audit is the weakest joint in the argument this Review is built on, and it is more honest to name it in the first essay than to leave it implicit across forty.

Second, a roll that carried 22 lakh dead names and seven lakh duplicate entries was not a defensible register, and the Commission is entitled to say that the alternative to a contested revision was an uncontested inaccuracy. That is right. It argues for revision conducted early, published fully and defended in public — not against revision.

Third, publishing a representative’s record may produce the behaviour it measures rather than the behaviour it wants. Attendance and questions asked are countable and only loosely related to usefulness. This Review argues elsewhere that systems optimise for whatever they count; a legislative scorecard invites exactly that distortion, and the recommendation below should be read with that warning attached rather than as a clean win.

What we would do

1. Put roll revision on a fixed statutory calendar, away from election dates. With full publication of additions and deletions by reason and district, and a time-bound appeal for anyone removed. Bihar shows both why revision is necessary and what the absence of a published ledger costs: two totals that do not reconcile, and a court having to order disclosure of who was struck off. A pan-India revision has been announced; the publication standard should be settled before it begins, not litigated during it.
2. Publish a deletion reconciliation with every revised roll. Opening electors, deletions by reason, additions by form, closing electors, and the residual that fits no category — one table, one page, per district. The 9,968 unexplained deletions in Bihar were discoverable only because petitioners went looking. A standing format makes the question routine rather than adversarial.
3. Settle the financing architecture before the next general election. Near-real-time disclosure of contributions above a modest threshold, party-level expenditure limits with audited returns, and an enforcement body with the capacity to act. ₹16,518 crore flowed through the instrument the Court removed; that demand has not disappeared, and money that cannot move through a disclosed channel does not stop moving. The vacuum benefits incumbents of every colour, which is why it persists.
4. Pair any change to the electoral calendar with interval accountability. If elections are to be less frequent, the years between them must carry more scrutiny, not less. A Joint Committee that has spent twenty months on frequency could report on frequency and interval accountability together, and would produce a materially better settlement for the citizen if it did.
5. Protect the Commission’s standing as a first-order concern. The most valuable institutional asset India holds in this domain is a referee both sides have historically accepted. Bihar shows how quickly that acceptance becomes the subject of litigation. It is easier to lose than to rebuild, and nothing on the reform agenda is worth spending it on.

India has built the most impressive voting machinery in the world and surrounded it with institutions that have not been rebuilt in decades. Politics is the promise; governance is the delivery. The Commission does not achieve the near-impossible because Indians are better at elections than at electricity or schooling, but because the task has an unambiguous deadline, a single accountable authority, and a result nobody can fudge. That is the argument this Review will make repeatedly, in different sectors, with different numbers: India’s delivery failures are failures of institutional design — tasks without owners, targets without deadlines, and results nobody is required to publish.