The subject of those notes was not a Model Code quarrel or a press-conference slight. It was the machinery of the franchise itself — who is added to the electoral roll, who is struck off it, who may apply to get on, who holds the password to the database, and whether a statutory form can be rewritten by a portal update. By the time the newspaper printed the record, the Commission’s Special Intensive Revision had already removed more than thirteen crore names from draft rolls in thirty States and Union Territories.

Within hours the Opposition demanded Kumar’s removal, a Supreme Court-monitored inquiry and a halt to the revision. Some leaders asked for his arrest and for fresh polls. The Bharatiya Janata Party called the uproar a smear and the internal notes a sign of vibrant democracy. The Commission itself did not deny that the fourteen notes existed. It called them suggestions, said differing views are normal, and insisted that every final decision of the past year had been unanimous. It did not say who changed Form 6, who authorised a wave of appeals in West Bengal, or why the full three-member body did not meet for months.

This essay treats that investigation as a starting point, not a verdict. It sets the notes against the history of the Commission, the 2023 law that chose its present members, the legal design of unanimity inside a three-member bench, the software that now holds the voter list, the citizen who discovered she was no longer on it, and the election of 2029, which will be run by some version of this institution whether or not anyone resigns this week.

IThe record

The Election Commission is, and has been since 1993, a three-member constitutional body. The Chief Election Commissioner is first among equals. He does not have a veto. That is not Opposition talking-point; it is the holding of a Constitution Bench in T. N. Seshan v. Union of India (1995). The 2023 appointment statute added a cultural instruction: transact business unanimously “as far as possible,” and settle procedure and the division of work only by unanimous decision. The Express notes are what it looks like when both sentences are tested at once.

Sandhu and Joshi’s objections clustered around five themes.

The first was process. Communications and orders, they wrote, were going out in the name of “the Commission” without the Commission having approved them. Joshi invoked Article 324 and the 2023 Act. In a body that is supposed to speak as a body, a letterhead is not a small thing.

The second was Form 6, the statutory application a citizen uses to register as a new voter. In mid-2025 a new mandatory question appeared on the online form on ECINet: whether the applicant, or a parent or grandparent, figured on the roll prepared in the previous Special Intensive Revision. Joshi had warned in May that the form could not be altered without the Union government amending the Registration of Electors Rules, 1960. Sandhu had agreed. The question went live anyway. On 13 August Sandhu called the change “unauthorised and illegal” and said it must be removed immediately. He had learned of it from a newspaper report.

The third was the Special Intensive Revision itself — the additions, deletions and restorations. The SIR began in Bihar in June 2025 and was rolled out nationally. Draft deletions crossed thirteen crore. The two Commissioners recorded that they were being kept in the dark about operational decisions that were, in law, the Commission’s to take.

The fourth was West Bengal. After the Supreme Court in February 2026 took the last stage of the Bengal SIR away from the Commission and assigned it to judicial officers, lakhs of appeals were filed. Of about 27 lakh deleted Bengal electors, some 22 lakh filed appeals to climb back on; another 16 lakh appeals sought to throw other people off. Sandhu asked, in writing, who had authorised the Election Commission itself to file appeals against voters whom judicial officers had already included, and who had actually filed them. Neither he, nor Joshi, nor the West Bengal Chief Electoral Officer, he recorded, had been told the basis or the process.

The fifth was the machine. Joshi wrote in May of the “gradual centralisation of the electoral roll database” and asked for an audit to ensure that only statutory authorities held change credentials. Sandhu wrote in August that Electoral Registration Officers, District Election Officers and Chief Electoral Officers did not have “proper and complete access” to the ERONet portal, and that the Director General (IT) had no legal authority to restrict the access statute gives those officers. When a work-allocation order further concentrated IT powers in Delhi, both Commissioners wrote separately to Cabinet Secretary T. V. Somanathan. Sitting Election Commissioners do not ordinarily petition the executive they are meant to referee. That they did so is part of the record.

The Commission’s reply of 23 September is also part of the record. It did not dispute the existence, dates or contents of the fourteen notes. It redescribed them as suggestions, observations, technical opinions and internal checks-and-balances. Every final decision of the past year — the SIR included, along with some forty other initiatives — had been unanimous. Highlighting fourteen notes while ignoring the larger body of approvals, it said, gave only one part of the picture. Digital platforms, it added, operate under strict data security protocols, and local officers function according to statutory powers.

The Commission acknowledged the notes and called every final decision unanimous. It did not explain how fourteen recorded objections fit inside unanimity, who changed Form 6, who filed the Bengal appeals, or why the full Commission did not meet from May to early September.

That silence is not proof of a conspiracy. It is also not an answer. The rest of this essay is about what the unanswered questions sit on top of.

IIHow the referee was built

Article 324 vests the “superintendence, direction and control of elections” in an Election Commission. The text is almost silent on how many people that Commission should contain. For the first four decades of the Republic it contained one. The Constituent Assembly imagined simultaneous elections once every five years. One referee seemed enough. He was appointed by the President on the advice of the Union government. There was no statutory selection committee, no Leader of the Opposition, no Chief Justice.

T. N. Seshan, CEC from 1990 to 1996, ended the age of neglect. He enforced expenditure ceilings, introduced the photo identity card, postponed polls on law-and-order grounds, and spoke of politicians in language they were unused to hearing from a civil servant. He also concentrated power in a single office in a way that frightened the government that had to live with him. Additional Election Commissioners were briefly appointed in late 1989, withdrawn in January 1990, and appointed for good from 1 October 1993. Seshan went to the Supreme Court. The 1995 judgment upheld the multi-member Commission and settled two points that matter now. The CEC is not a superior officer with a veto; the three are equals, and if they cannot be unanimous the majority binds. And it was “wrong to think that the two ECs were pliable persons who were being appointed with the sole object of eroding the independence of the CEC.”

The second sentence has aged less comfortably than the first. For most of the years after 1993 the Commission presented a single face. Dissent existed but was rare. The best-known modern precedent is Ashok Lavasa, who in 2019 recorded notes of dissent on clean chits given to senior leaders of the ruling party on Model Code complaints and later left for the Asian Development Bank. Lavasa’s dissent was about speeches. It was not about the list that produces the count. That is why September 2026 feels different even to people who remember him. The Express notes touch the roll.

There is an irony the history makes visible. The multi-member body was created, in part, to restrain a strong CEC. In 2026 the complaint from two Commissioners is that a strong CEC has restrained them. The instrument designed as a check is being described, by two of its occupants, as a rubber stamp. Whether that description is fair is a present-tense question. That the description could be made at all is a past-tense achievement of the 1993 design.

IIIWho chooses the referee

Every crisis of the Election Commission eventually becomes a crisis of how its members were chosen. On 2 March 2023, in Anoop Baranwal v. Union of India, a five-judge Constitution Bench held that the long practice of the Union government alone advising the President on CEC and EC appointments was constitutionally unsustainable. Free and fair elections, the Court said, are part of the basic structure. A referee appointed solely by one of the teams is a structural conflict. Until Parliament wrote a law under Article 324(2), appointments would be made by a committee of the Prime Minister, the Leader of the Opposition in the Lok Sabha, and the Chief Justice of India.

Parliament wrote a law within months. The Chief Election Commissioner and Other Election Commissioners (Appointment, Conditions of Service and Term of Office) Act, 2023 came into force on 2 January 2024. It kept the Prime Minister and the Leader of the Opposition. It replaced the Chief Justice with a Union Cabinet Minister nominated by the Prime Minister. A Search Committee chaired by the Law Minister shortlists names from a pool of serving or retired secretaries. The Selection Committee decides by majority. Two of three votes belong to the government of the day.

The Opposition member can dissent, and has. Adhir Ranjan Chowdhury dissented on the March 2024 appointments of Gyanesh Kumar and Sukhbir Singh Sandhu. Rahul Gandhi dissented in February 2025 on Kumar’s elevation as CEC and on Vivek Joshi’s appointment. Dissent on the selection committee does not stop an appointment. It records that the appointment was not consensual.

All three current members are retired IAS officers, as the statute effectively requires. Kumar’s career ran through Kerala, Kashmir-related assignments and Ayodhya-related work before he entered the Commission. Sandhu and Joshi were themselves selected under the new law. The Opposition’s present line — that two “handpicked” Commissioners have now rebelled against a third “handpicked” CEC — is therefore also an admission that the 2023 process did not produce a monolith. The government’s line — that internal notes prove a living, deliberative institution — uses the same biographical fact to the opposite effect.

A Constitution Bench has since been asked to look again at the 2023 Act. In the week of the Express report, Justice Dipankar Datta repeated the old axiom: the Commission must not only be independent, it must be seen to be independent. Perception, in election law, is not a public-relations problem. It is half the job.

IVThe revision

Electoral rolls decay. People die, move, turn eighteen, acquire a second address, or were never eligible. A Special Intensive Revision is, in the abstract, the least controversial job an Election Commission has. The current SIR began in Bihar in June 2025 and spread. By September 2026 draft rolls in thirty States and Union Territories had lost more than thirteen crore names. India’s electorate sits in the range of ninety-five to a hundred crore. A deletion rate of that size, even as a draft figure that claims and objections can still repair, is large enough to change the character of constituencies.

The Commission’s public case is that the deletions catch duplicates, dead electors, migrants who have left, and ineligible entries — including those the ruling party describes as infiltrators. The Opposition’s public case is that the deletions fall disproportionately on the poor, on migrants without documentary trails, on Muslims in some States, and on voters in States the BJP does not govern. Both cases can be partly true at once. A genuine clean-up will always remove some legitimate voters who cannot produce papers on the Commission’s timetable. A politicised clean-up will always be able to point at some genuine duplicates. The Express investigation does not settle which description fits 2025–26. What it shows is that two of the three people legally responsible for the exercise did not believe they were in control of it.

Three design choices turned a bureaucratic revision into a national argument. The first was the burden of proof. In a classical revision the State comes to the voter. In large parts of this SIR the voter was required to re-establish a claim, sometimes with documents that State governments themselves issue and that the Commission was not always willing to treat as conclusive. When the State demands proof of a right the Constitution already conferred, the poorest and the most mobile lose first.

The second was sequencing. Starting in Bihar, then expanding into States heading into assembly cycles, made every deletion look like a campaign event. West Bengal became the most litigated theatre; the Supreme Court’s transfer of the last mile to judicial officers is itself a judgment on trust.

The third was opacity about the draft-to-final pipeline. A draft deletion is not a final deletion. Claims, objections, appeals and restorations exist precisely so that the thirteen-crore figure can shrink. But if software will not record an Electoral Registration Officer’s decision to restore a name — as field officers in Goa told the Express — the pipeline is not a safeguard. It is a decoration.

None of this requires a conspiracy. Large information systems fail in the direction of their incentives. If the political reward for a “clean” roll is higher than the political cost of a wrongful deletion, the system will over-delete and under-restore. The two Commissioners’ notes are consistent with that diagnosis even if they never used the word.

VA checkbox that outran the rules

Form 6 is dull until it is not. It is the application prescribed by the Registration of Electors Rules, 1960. The form is a creature of delegated legislation. The Election Commission administers it. The Union government, by amending the Rules, changes it. That distinction is the whole of the Form 6 controversy.

The new mandatory question converted a citizen’s individual right to register into a genealogical inquiry. A first-generation voter, a migrant whose parents never figured on any Indian roll, a person whose surname is spelled six different ways across six documents, now faced a gate the 1960 Rules did not erect. The Commission can defend the question as an anti-fraud filter. The two Commissioners’ legal point is prior to that defence: even a wise filter, if it changes a statutory form, is still an illegal filter until the Rules are amended.

Most disputes inside the Commission can be boxed as policy disagreement. Form 6 cannot. Either the Commission has the power to rewrite a prescribed form by portal update, or it does not. Former CEC O. P. Rawat, speaking after the report, called the change “absolutely wrong and illegal.” If he is right, every registration collected under the altered form carries a defect. If he is wrong, two sitting Commissioners misunderstood the most basic instrument they administer. The Commission’s reply did not say who added the question, on whose instruction, or under what legal power. Courts, when they see this record, will not ask whether the mood in Nirvachan Sadan was collegial. They will ask whether the 1960 Rules authorised the field.

There is a practical sequel. Citizens who were turned away from Form 6, or who completed it under a question the Commission later quietly adjusted, now form a class. Some will never know they were filtered. Some will find out in 2029. The legal system is bad at retrofitting a right to vote onto people the software already forgot. That is why a checkbox is never only a checkbox.

VIThe software state

Indian election law still reads as if the electoral roll is a local document. The Electoral Registration Officer is a statutory authority. The District Election Officer and the Chief Electoral Officer of the State sit above her. Appeals climb a local ladder. The software was supposed to be a clerk, not a principal.

Joshi’s May note and Sandhu’s August note describe a different architecture. Credentials to change the roll, they suggested, were concentrating in the Commission’s IT division in Delhi. A Director General (IT) was, in Sandhu’s words, exercising a restriction she had no legal authority to exercise. A work-allocation order issued without the full Commission’s approval threatened to harden that concentration.

Goa is the parable. Officers on the ground found ninety-seven people eligible. The software would not record the decision. The names stayed off. Officers told the voters to file Form 6 as if they were new. In a small State, where being on the roll is also the key to public-health schemes, a missing name is not only a missing vote. It is a missing dialysis slot. Eight emails in seven days from Goa staff produced no unlock. That is what centralisation looks like when it fails: a statutory officer reduced to a supplicant of a helpdesk.

Defenders of the new stack will say that a national database is the only way to catch a voter registered in two States, that cybersecurity requires a single custodian, and that State CEOs have historically been uneven. All of that can be true. It does not answer the legal question the Commissioners posed. Statute locates the power to include and delete in named local officers. Software that overrules those officers without a Commission decision is not a technical upgrade. It is a silent amendment of the Representation of the People Act.

There is also an audit problem. Joshi asked for an audit to ensure that only statutory authorities held change credentials. The Commission’s public reply spoke of “strict data security protocols.” It did not say whether the audit happened, who would conduct it, or whether the two Commissioners would see the report. In a system that now holds the franchise of a billion people, “trust the protocol” is not an argument. It is a request for faith.

When the ERO finds a citizen eligible and the portal refuses to write the name, the Constitution has been outvoted by an access-control list.

VIIUnanimity, majority, and the missing meeting

Indian public conversation still talks about “the Election Commissioner” in the singular. The law has not, since 1993. The Seshan judgment and Section 18 of the 2023 Act generate four possible states.

State A is genuine unanimity: all three agree, a meeting records it, instructions issue. This is the Commission’s description of the last year. State B is recorded majority: they differ, they vote two-to-one, the majority prevails. This is what the 1995 judgment contemplated and what the 2023 Act permits when unanimity fails. State C is silence presented as unanimity: no meeting, or a meeting that does not vote, and instructions issue in the Commission’s name. The two Commissioners’ notes describe State C and call it unauthorised. State D is dissent filed and ignored: notes exist, they are acknowledged as “suggestions,” and the original instruction stands. This is the Commission’s 23 September description of the fourteen notes, married to a claim of State A.

State D is the unstable one. A suggestion that has been rejected is, in ordinary language, a dissent. Calling it a suggestion after the fact is a classification choice, not a fact about the document. If the full Commission did not meet from May to August, as the Express reported — the next full meeting coming on 9 September, after four months — then neither unanimity nor majority can be shown for decisions taken in that window. Sparse meetings make both rules difficult to demonstrate.

Every collegial body grows a chair who speaks for it. That is administratively inevitable. It becomes constitutionally dangerous when the chair’s voice is treated as the body’s voice without a vote. The CEC controls the agenda, the secretariat, the press note and, in the current stack, a great deal of the IT. Those are not small advantages. They do not need to be a formal veto to function as one.

The remedy the two Commissioners reached for — writing to the Cabinet Secretary — tells its own story. They did not, so far as the public record shows, force a recorded vote and publish a minority note. They went to the executive’s highest civil servant. That is both a sign of how blocked they felt inside the building and a paradox: the independence of the Commission is not strengthened when its members petition the government to referee their own referee. A healthier design would be boring. Fixed calendars of full-Commission meetings. A rule that no instruction issues in the Commission’s name without a minute. A published dissent, as the Supreme Court itself uses. None of that requires a new Constitution. It requires the three people now in the building to behave as if the 1995 judgment were still in force.

VIIIThe political hour

The Opposition’s first move was to split the institution. For months the INDIA bloc had attacked “the Election Commission.” After the Express report it attacked Gyanesh Kumar. Rahul Gandhi called the combination of the ruling party, the RSS and the Commission an “act of treason” and revived the phrase “vote chori.” Congress described Kumar as the Prime Minister’s puppet. Arvind Kejriwal demanded his arrest and the cancellation of recent elections, calling the SIR the nation’s biggest scandal. Mamata Banerjee sought a repoll. Uddhav Thackeray demanded arrest and fresh polls. Omar Abdullah asked what remains of democracy if trust in the election is lost. M. K. Stalin said Kumar must go and the SIR must be halted. The common three-point demand by 24 September: remove the CEC, put the Commission under a Supreme Court-monitored inquiry, pause the revision.

The BJP’s first move was to defend the institution by redescribing the dissent. Sambit Patra called the uproar a smear aimed at discrediting constitutional bodies. Internal notes, he said, were proof of a healthy democracy, not of dictatorship. What mattered was that the SIR had been an exercise in unanimity. He reached for history — R. K. Trivedi, Seshan’s own contempt for colleagues he called donkeys and rascals — to argue that disagreement inside the Commission is older than this government.

Two things are new in this exchange. The first is the Opposition’s decoupling. By blaming the CEC rather than the Commission, the Opposition is trying to salvage the office for the day it might win an election. That is tactically rational. It is also an admission that a three-member body can be legitimate even when one member is not. Whether the public accepts the decoupling is another question. Most voters do not parse CEC from EC. They parse “the election people.”

The second is the government’s comfort with process language. “Notes are normal” is a strong reply if the notes were about the colour of the voter-slip. It is a weaker reply if the notes say a statutory form was illegally altered and a database was centralised against written advice. Process language without point-by-point rebuttal leaves the specific charges standing. That is why the Express’s unanswered questionnaire will travel farther than either party’s press conference.

International desks treated the story as a democracy story. That circulation does not decide an Indian election. It does set the frame in which 2029 will be observed. A Commission that spends the next three years answering “was the roll fair?” in every foreign interview has already paid a cost, regardless of who is right.

IXThe citizen who lost her vote

Abstract numbers conceal the only fact that ultimately matters. A vote is a personal right. When thirteen crore names come off a draft, some of those names belong to people who should not have been on the roll. Some belong to people who have voted for three decades and cannot understand why this year they must prove they exist.

Goa makes the abstraction concrete: eligible voters, a portal that would not write the finding, health schemes tied to enrolment, eight emails that produced no unlock. West Bengal makes it political: 27 lakh deletions, 22 lakh appeals to climb back on, a Supreme Court that no longer trusted the Commission to finish the last mile. Delhi’s draft roll, published in the same season, left out 47.7 lakh of 1.45 crore electors, with some constituencies seeing more than forty percent of electors not included. Odisha, Kerala, Kashmir, the Hindi heartland: each State has its own ratio of needed deletion to disputed deletion. No public, independently audited, State-wise table yet exists that would let a citizen see which is which. In the absence of that table, every party will cite the geography that helps it.

The Commission’s safeguard is the claims window. In theory a wrongfully deleted voter files a form, attends a hearing, and returns to the roll. In practice the window is short, the documents demanded are uneven, the software is the same software that deleted the name, and the voter who works a daily-wage job does not have three mornings for a hearing. A right that can be recovered only by a person with time, literacy and transport is a right that has already changed its social composition.

This is the quietest and most important angle of the investigation. The notes from Sandhu and Joshi are not, in the end, about their dignity as Commissioners. They are about whether the person at the booth still meets the person the Constitution recognised. If the two Commissioners are right that they were locked out of the process that writes that recognition, then the lock-out was not an internal HR problem. It was a franchise problem.

XAfter the rift

Institutional crises in India rarely produce institutional reform. They produce a news cycle, a court listing, a committee, and then the next election. The risk in this case is that 2029 arrives with the same three-member body, the same 2023 appointment arithmetic, the same ECINet permissions, and a voter list whose contested deletions have been frozen into finality. At that point the argument will no longer be about notes. It will be about whether the loser accepts the count.

Some proposed remedies will not work. Impeachment of a CEC is constitutionally possible and politically almost impossible. It requires a special majority the present Opposition does not have. Demanding the CEC’s arrest is rhetoric. Cancelling completed assembly elections on the basis of an internal dissent would set a precedent every future loser would abuse. Treating the Express report as a conspiracy complete with a named mastermind will convince only those already convinced. On the other side, treating fourteen recorded objections as a compliment to internal democracy will not rebuild trust among the voters who cannot find their names. “All decisions were unanimous” is a claim that now has to be demonstrated with minutes, not asserted with a press note.

What would actually move the needle is less theatrical.

First, publication. The Commission can, without admitting fault, publish a State-wise, constituency-wise account of draft deletions, restorations, rejected claims and pending appeals, with a plain-language explanation of the top reasons for deletion. Sunlight does not require a court order. It requires a decision.

Second, an independent technical audit of ECINet and ERONet: who holds change credentials, which officer’s decision the software can overrule, whether the Form 6 change left a defect in the 2025–26 registrations. The two Commissioners asked for this audit in writing. Commissioning it now would be cheaper than defending its absence in 2029.

Third, a sitting calendar. The full Commission should meet on a published schedule. Instructions that alter forms, permissions or the division of work should not issue between meetings. This is Section 18 of the 2023 Act, taken seriously.

Fourth, the appointment statute. As long as two of three votes on the Selection Committee belong to the government of the day, every controversial CEC will be received as a government CEC. Restoring the Chief Justice to the committee — the Anoop Baranwal interim — would not depoliticise the office. It would share the blame. Shared blame is how referees survive.

Fifth, the Supreme Court. Several benches already have pieces of the SIR. A Constitution Bench already has the 2023 Act. The Court cannot run elections. It can insist that the body that does run them be able to show its working.

Democracies lose elections in the count and lose legitimacy in the roll. India has, for thirty years, been able to boast that its referee, however battered, was accepted by winners and losers. That boast was always partly myth. Seshan was hated. Lavasa was isolated. The 2023 Act was written to keep the government inside the room where referees are chosen. The Express notes are the first time the myth has been contradicted from inside the room, in the commissioners’ own sentences, about the list that makes the count possible.

Whether September 2026 becomes a turning point or a footnote will be decided by what is published, audited and amended before the next roll is frozen. The past gave India a strong Commission by giving one man too much power and then spreading that power across three. The present has shown that three is not a number, it is a practice. The future will test whether that practice can be recovered in time for a hundred crore people to believe that when they stand in line in 2029, the name the officer finds is still theirs.