Source : INDIA TODAY NEWS
Author’s note: This essay reconstructs and critically tests a thought experiment I put together as a framework analogous to the ongoing Special Intensive Revision of electoral rolls, and the criticism the exercise has drawn. ChatGPT was used as a tool to poke holes in my argument and dig up supporting research material. This process was conversation-led rather than prompt-driven. ChatGPT was then used to structure that conversation into an essay.
advertisement
Imagine a closed system inhabited by a finite number of people. Within it operates a single factory that processes uncleaned wheat, separates usable grain from chaff and supplies the entire population. There is no competing factory and no alternative source of wheat. The people depend upon it for something indispensable to their lives. They also own shares in the company that operates the factory.
Suppose the factory receives 100 metric tonnes of material each month, containing 90 tonnes of good wheat and 10 tonnes of chaff. A well-run operation should identify and remove the 10 tonnes of chaff while preserving the 90 tonnes of food. Its quality cannot be measured merely by the purity of what leaves the factory. It must also be measured by how much usable grain reaches the people who need it.
Now imagine that the factory instead rejects 50 tonnes, although 40tonnes of that rejection is perfectly good wheat. Its published output may be remarkably free of impurities, but it has deprived the population of almost half the wheat available to it. Management might attempt to retrieve the grain from the discard pile in a second round of sorting, and then sort the rejects again. Such reprocessing consumes time, labour and resources. It may also never fully repair the original mistake. The ideal remains a first pass that separates chaff without needlessly discarding wheat.
The factory is an analogy to the Election Commission of India‘s SIR. Wheat represents eligible voters. The chaff represents legitimately ought not to remain on a particular electoral roll – deceased persons, duplicates, people no longer ordinarily resident there, or persons otherwise ineligible.
The monopoly analogy is instructive: just as the public cannot obtain its wheat elsewhere, citizens cannot turn to a competing institution to maintain an authoritative electoral roll. The Commission exercises that public responsibility within a constitutional and statutory framework.
TWO CEOS: INCOMPETENCE AND DESIGN
Consider the first possible explanation for the factory’s waste:incompetence.
Its chief executive has devised an inaccurate process that systematically rejects good wheat. Families receive less food;some go hungry. Because consumers are also shareholders, they suffer a second injury: if the operation is inefficient and costly, the company’s performance and the value of their holdings may deteriorate. The promoters, recognising managerial failure, remove the CEO. In this version, the injury is serious, but it follows from administrative incapacity rather than a deliberate attempt to deprive the population.
Now consider a second explanation: design.
The CEO deliberately restricts the wheat available for sale. As the only supplier, the factory enjoys market power. Under suitable conditions, it can use scarcity to raise prices, increase profits and present shareholders with an apparently healthier financial statement.
The promoters may reward the CEO for generating returns, even as the same people —shareholders and consumers — struggle to obtain enough wheat. An institution charged with serving a population has begun to serve another incentive instead. The apparent success of the company conceals the failure of its public purpose.
The analogy proposes two possible explanations for a flawed SIR: the erroneous exclusion of eligible citizens may be the product of bad administration or of a process deliberately designed to make some citizens less likely to remain enrolled.
Note that, rather than being a fact-finding exercise about the ECI, the hypothesis is about incentives. The distinction is worth examining because its implications for institutional accountability differ. Incompetence demands correction and oversight. Deliberate political discrimination, if established, would demand a more searching investigation and legal remedies.
But this hypothesis must be read with a caveat, that there are limits to the market comparison.
A monopoly cannot necessarily charge any price it wishes: demand, income, regulations and the risk of unrest matter. A high price does not automatically produce higher corporate value. More fundamentally, the dividend from excluding voters would not be a higher price for a product. It would be a possible electoral advantage for those who benefit from the resulting composition of the electorate. Thus, the two-CEO device is most useful as a contrast between accidental and purposive exclusion, not as a literal economic model of election administration.
WHAT DOES AN ACCURATE ELECTORATE ROLL MEASURE?
The first objection to the factory comparison is straightforward.
Perhaps the discarded material really is chaff. Removing dead, duplicate or otherwise ineligible entries is a legitimate and necessary function. A claim of mass wrongful exclusion cannot be established by the size of the discard pile alone. The fact that someone voted previously is relevant evidence of an earlier registration, but not conclusive proof that the person remains eligible for the same constituency’s role today. People move; records change; some entries may have been erroneous to begin with.
But that objection does not settle the question. The object of a revision is not to maximise deletions. It is to construct the most accurate electoral roll: to include everyone who qualifies and exclude those who do not. A system that is excellent at eliminating ineligible entries but poor at retaining eligible citizens has failed one half of its task. The distinction is familiar in statistical classification. A legitimate voter wrongly removed is a false negative; an ineligible entry wrongly retained is a false positive. Reporting only the number of entries removed says little about either error rate.
There is a more demanding question: on whom should the burden of detecting and repairing errors fall? In the wheat factory, it would be extraordinary to insist that every hungry household prove that individual grains in the waste pile had been edible before management examined its sorting equipment. Likewise, when a public institution controls the official roll, it should be able to demonstrate the quality of its decisions – through clear reasons, accessible correction procedures, defensible data and systematic verification – rather than relying solely on each excluded citizen to identify an error and win readmission.
Perfect classification may be unattainable on a national scale. Even a conscientious system can make mistakes, and inclusion errors can dilute legitimate votes or distort an election. The relevant question is therefore not whether errors are logically possible, but how they are anticipated, measured, minimised and remedied, and whether burdens fall disproportionately on particular groups. Critically, restoration after an election cannot return a ballot that an otherwise eligible citizen was unable to cast when it mattered.
THE 13.39-CRORE QUESTION: CATEGORIES ARE NOT PROOF
This distinction sits at the centre of the controversy over the national SIR figures. In its October 2026 clarification, the ECI said that 13.39 crore electors had been placed in categories arising from the revision process across its first three phases, rather than accepting the description that all 13.39 crore had been conclusively or unlawfully deleted.
The reported breakdown was approximately 6.30 crore permanently shifted, 3.01 crore absent or untraceable, 2.80 crore dead, 1.02 crore duplicate and 0.26 crore in other categories, out of some 95.40 crore reviewed. These are administrative classifications, and draft-stage figures must not be treated as a uniform set of final exclusions.
The numbers answer an important question: how has the Commission described the entries it flagged? They do not, by themselves, answer a second question: how reliably were people assigned to those categories, and how many eligible electors were wrongly left out? The claim that the Commission has published no breakdown would therefore be inaccurate.
The stronger criticism is that a breakdown of stated reasons is not an independent measure of their correctness. For that one needs disaggregated evidence, quality checks capable of public scrutiny, correction rates, and the fate of appeals and disputed cases.
The concern is not purely philosophical. Reporting has documented grievances from previously registered electors and difficulties surrounding their removal or restoration.
On 29 September and in early October 2026, the Commission instructed officials in states and Union territories where the exercise had been completed to conduct a special drive to identify and enrol eligible people left out of the rolls. Former Chief Election Commissioner SY Quraishi welcomed the renewed effort to find omitted voters but questioned the use of Form 6, ordinarily framed as an application for new voters, to restore people who had been registered before. A corrective drive is evidence that the possibility of wrongful exclusion is being addressed. Yet, it is not, by itself, a count of confirmed wrongful exclusions or proof of their cause.
For the wheat factory, the analogous question is multifold: how much chaff was labelled as such and how much good wheat entered the discard stream, who detected it, how much was recovered and how much never reached the population?
THE INCENTIVE TO EXCLUDE, AND THE PROBLEM OF PROOF
The original thought experiment becomes politically sharper when the CEO’s incentives are reconsidered. For an incumbent political party, the hypothetical benefit of a manipulated electoral roll is obvious: retaining people more likely to support it while removing people less likely to do so could alter electoral competition. If an incumbent sought such an outcome, rewarding an exclusionary design might make more sense than tolerating random incompetence.
Again, a caveat: an incentive to interfere is not evidence that the incentive has been acted upon. Nor does it establish that the Election Commission, which is constitutionally distinct from the government of the day, has collaborated in such interference.
At least three explanations must remain open until evidence separates them.
First, errors might arise without political intent, through flawed databases, overly rigid documentation, rushed schedules or mistaken field verification. Second, a formally neutral rule might have unequal effects because access to documents, mobility, residence and administrative assistance vary across groups. Third, exclusions might be intentionally directed or knowingly tolerated to favour an electoral interest. The same observable imbalance could arise through different mechanisms; resolving the distinction requires more than the presence of a political incentive.
The most informative evidence would therefore look for patterns that cannot readily be explained by valid eligibility factors, differences in population movement, local administrative conditions or the overall political swing. It would also investigate who made the relevant policy decisions, what they knew about emerging errors and how they reacted to warnings. The question is not simply whether the factory is wasting wheat, but whether the wasted grain is concentrated in places that systematically benefit one set of stakeholders — and, if so, why.
WHAT PRANNOY ROY’S DEKODER ANALYSIS CLAIMS
A recent DeKoder analysis presented by psephologist and broadcaster Prannoy Roy brings empirical evidence into that debate. In an account published on 9 October 2026, the analysis compared West Bengal constituencies with different rates of SIR deletions and the subsequent change in party support between the 2021 and 2026 Assembly elections.
In the 25 constituencies with the highest reported deletion rates, it identified a roughly 10.5-percentage-point swing against the Trinamool Congress (TMC). In the 25 with the lowest deletion rates, the reported swing against the TMC was about 5.3 points. The association was presented as consistent with the concern that greater deletion had accompanied greater electoral damage to the TMC, while benefiting the Bharatiya Janata Party (BJP).
Roy also drew attention to who appeared to have been deleted.
According to the analysis, women accounted for 62% of deleted names, compared with 38% for men. It also estimated that about 30 lakh Muslim voters were removed from the rolls, against the 19.8 lakh who would have been removed had their deletion rate matched that of non-Muslims.
The estimated deletion rate among Muslims was thus 52% higher than among non-Muslims. Roy’s argument is that these apparent disparities matter electorally because the affected groups have not supported the principal parties in identical proportions in past voting patterns.
These claims add something to the original factory analogy. The argument goes beyond imagining that good wheat may have been discarded or that a manager could gain by discarding it. It identifies a purported pattern of exclusions and electoral shifts that can be tested.
It must be noted, however, that the numbers, even taken at face value, show correlation rather than causation. Constituencies differed in many respects besides deletion rates. Aggregate demographic estimates do not establish the identity or voting intention of every removed individual, and the analysis does not itself demonstrate that a decision-maker engineered the pattern.
There is also an ecological risk. Observing that a constituency with many deleted names swung towards one party does not mean that those deleted would all have voted for the other party. To move from a correlation to a causal claim, researchers would need to address pre-existing constituency differences, turnout, migration, urbanisation, the particular grounds for deletion, changes in public opinion and the distinction between gross deletions and final unrepaired errors. This is a task that requires urgent attention.
THE WEST BENGAL CEO’S RESPONSE
The West Bengal Chief Electoral Officer rejected DeKoder’s interpretation on 9 October 2026, describing the claims as misleading and speculative. The officer cited results in the five constituencies with the highest deletion levels, saying four were won by the TMC and one by the BJP. In the five with the lowest deletions, the BJP won all five.
The office argued that these outcomes did not support the claimed political narrative. It further disputed the figure that women constituted 62% of deleted electors, saying its data on deletion of invalid entries, excluding adjudication cases, put women at around half. On religion, it said the Commission did not hold deletion data classified by individual electors’ faith, and therefore challenged estimates presented as Muslim deletion rates.
The officer emphasised procedural protections: notice, an opportunity to be heard, a reasoned order from an Electoral Registration Officer and avenues of appeal. The office maintained that SIR was a lawful exercise meant both to preserve valid entries and to remove invalid ones. These are substantial objections, and they belong in any fair appraisal of the DeKoder claims.
They do not, however, answer every empirical question. Looking at the five most extreme seats is not equivalent to testing the relationship across the 25-highest and 25-lowest groups, much less across every constituency. The outcome of who won a seat is not the same measurement as the change in vote share: a party can lose ground yet still win.
In a follow-up video responding to official counterclaims, Roy doubled down on analysing vote share swings rather than seat victories, noting that the BJP achieved a 10% vote swing in its favour in the top five deletion seats compared to just 2.5% in the bottom five. He explained that TMC retained four of those top seats only because they were high-margin safe seats in 2021 that absorbed the swing.
Equally, if DeKoder’s gender calculation includes adjudication deletions and the official counter figure excludes them, the two percentages may not share a denominator. And the absence of religion-coded official records does not by itself show that estimates based on other data are false; in fact, their methodology, assumptions and error margins become crucial. Without access to harmonised underlying data, neither the DeKoder estimates nor the office’s rebuttal should be treated as a definitive causal test. Again, one must underscore the urgency of an impassioned empirical analysis of the data, the entirety of which the ECI must make available.
The useful way forward is a reproducible comparison: agree on which deletion stages count, publish constituency-level denominators, distinguish provisional exclusions from final removals and restorations, disclose the statistical method used to infer demographic identity, and test the relationship across the full distribution of constituencies with sensible controls. An institution confident in the reliability of its process has a strong public-interest reason to make such scrutiny possible.
THE CONSTITUTIONAL QUESTION: WHAT’S THE REVISION FOR?
The most important part of the wheat analogy is ultimately not the CEO or even the disputed statistics: it is the purpose of the factory.
The factory exists to supply wheat to the people; the separation of chaff is a means to that end. Similarly, an electoral roll exists to make participation in representative democracy possible for those entitled to participate. Removing ineligible entries is necessary to protect the integrity of that entitlement, but exclusion cannot be treated as an achievement independent of inclusion.
The constitutional argument is more precise than a simple claim that any mistaken deletion automatically invalidates an entire SIR.
Article326 locates elections to the Lok Sabha and state legislative assemblies in adult suffrage, subject to constitutionally permitted disqualifications. Article 325 guards against exclusion from electoral rolls on grounds only of religion, race, caste or sex. Parliament’s election laws and the Commission’s powers under Article 324 supply the machinery for maintaining the rolls.
In Association for Democratic Reforms v Election Commission of India, decided on 27 May 2026, the Supreme Court upheld the Bihar SIR as a lawful and proportionate exercise within the statutory framework, and described its twin purposes as including all eligible electors and excluding ineligible persons. It also recognised safeguards of notice, inquiry, hearing, reasoned decisions and appeal;non-inclusion at the draft stage does not automatically amount to final deletion.
That ruling complicates any claim that inclusion and exclusion can be ranked as a matter of settled legal doctrine. They are both constitutional objectives. Yet it does not eliminate the normative argument that inclusion is the democratic end and exclusion a protective instrument. A state can legitimately investigate doubtful entries without turning verification into an obstacle course for people who are entitled to vote. If an eligible citizen is ultimately excluded without lawful justification, the constitutional and statutory entitlement to participate is implicated, even though the legal consequences and remedies depend on the facts and the process followed.
The harms on both sides are real but not identical. An invalid entry can undermine electoral equality, particularly if used to cast an unlawful vote. An eligible person wrongly deleted loses direct access to political participation. If that exclusion lasts through polling day, later administrative corrections cannot recreate the missed vote. That asymmetry makes it reasonable to demand particularly rigorous safeguards before an existing elector is finally deprived of registration. It does not justify ignoring invalid entries; it means that success must be judged against both forms of harm rather than the deletion total alone.
THE SHAREHOLDERS WHO MAY LOSE THEIR VOTES
The deepest feature of the original analogy is that the people who consume the wheat are also shareholders in the company. They are damaged not once, but twice. They lose what the institution is supposed to supply, and they lose influence over how it is governed.
An electoral system contains a still sharper possibility: people excluded from the electoral roll lose the very instrument through which they can help change the political leadership that presides over public institutions.
It’s true that the shareholder analogy is not exact. Citizens are not owners of the Election Commission in a corporate sense. The Commission is not a government department controlled by whichever party is in office. And electoral oversight does not work like promoters hiring and firing a CEO.
The point is political rather than corporate. Citizens collectively authorise public power through elections. Wrongful exclusion can weaken their ability to exercise that authorisation and to seek accountability through the ballot. Where exclusions fall unevenly, that feedback loop may be distributed unevenly too.
This is why the proposed audit should ask more than how many names were removed. It should establish how many exclusions were erroneous; how many were corrected before voting; how long corrections took; which groups and localities bore the heaviest burden; whether reasons were supplied in individual cases; whether appeals were realistically accessible; and whether the process’s error profile changed once problems became known.
Random independent sampling, transparent reversal and restoration statistics, and publication of appropriately privacy-protected local data would permit the public to distinguish administrative failure from patterned discrimination more convincingly than either official assurances or suspicion alone.
WHAT REMAINS OF THE THOUGHT EXPERIMENT?
It would be disingenuous to claim that it contains no holes – Several remain.
A real electorate is not fully closed: people turn 18, die, move between constituencies, lose or acquire eligibility, and sometimes enter the rolls late. A first-pass classifier can never be guaranteed perfect. The grain supply metaphor makes losses tangible, but legitimate electoral roll revision is not a one-way filtering operation. It also requires adding newly eligible and previously omitted citizens.
The CEO’s profit motive is only a metaphor for political advantage, and a pattern that benefits a party cannot, on its own, establish who intended it. DeKoder’s estimates are contested, and the West Bengal election office has supplied a different interpretation of the reported data.
But it’s important to acknowledge that those qualifications do not destroy the central argument – they discipline it.
The case for investigating wrongful exclusion need not mean proving a conspiracy; it should depend on the public institution’s duty to demonstrate that it can both maintain roll integrity and protect rightful enrolment.
The stronger version of the analogy is that a monopoly entrusted with an indispensable democratic function cannot measure success by the volume of material it rejects, and cannot shift the entire cost of identifying its mistakes to the people it serves.
If competent administration can still wrongly exclude eligible citizens, the institution must show how it catches and corrects those errors. If exclusions correlate with politically consequential patterns, the institution must make it possible to test whether the correlation is explained by legitimate factors, unequal administrative effects or deliberate action. Both propositions follow without presuming a verdict in advance.
A wheat factory that produces pure grain while leaving good wheat in the waste bin, and its shareholders hungry, has misunderstood its purpose. An electoral roll revision that celebrates the removal of names without a comparably rigorous account of who remained entitled to vote risks the same confusion.
Inclusion and exclusion are both necessary for electoral accuracy. But the democratic question that gives the exercise its meaning is whether every eligible citizen is still able to exercise their right to vote.
– Ends
SOURCE :- TIMES OF INDIA
