[This is the final post in a three-part series by Rudraksh Lakra. Parts One and Two are linked.]
This is the third essay in the Programming Democracy series. The first examined how Assam’s 2023 delimitation exercise restructured electoral geography through gerrymandering and opaque boundary-drawing. The second examined how West Bengal’s Special Intensive Revision (“SIR”) of electoral rolls generated large-scale exclusion through algorithmic classification, unstable software systems, and procedurally burdensome verification mechanisms. This essay draws the two together, identifies what they share, and asks what their convergence reveals about the conditions under which democratic competition is organised in India today and the role of technologies.
The Assam delimitation and the Bengal SIR were different instruments deployed in different states. One redrew constituency boundaries to determine who could win within a given electorate. The other revised electoral rolls to determine who constituted the electorate at all. One operated through maps and geographic data. The other operated through databases and verification software. Yet examining both exercises together reveals a shared grammar: consequential decisions about the conditions of electoral competition were made before polling day, through administrative and technocratic processes operating beneath the procedural language of maintenance. These were shielded from independent scrutiny through inaccessible formats and opaque methodologies.
The parallels between the two exercises run deeper than their institutional form. In Assam, the draft delimitation notification was released in June 2023, with objections required within three weeks, a timeline that made meaningful public participation extremely difficult. In Bengal, the “logical discrepancy” category was introduced midway through the SIR, and instructions were repeatedly altered at times through informal WhatsApp communications, while the election followed within weeks of the electoral rolls being frozen (see here, and here). In both cases, compressed timelines ensured that the exercises acquired practical finality before legal challenges could meaningfully intervene.
The methodology of both exercises was withheld in remarkably similar ways. In Assam, constituency boundaries were released only as static PDF documents lacking latitude and longitude markers, making independent computational analysis almost impossible without manually reconstructing each boundary. In Bengal, the final electoral rolls were released as scanned PDF images with CAPTCHA-gated downloads, averaging vastly larger file sizes than machine-readable equivalents while containing none of the underlying structured data. In both exercises, the Election Commission possessed the data in analysable form but released it in formats that obstructed scrutiny. India administers Aadhaar and UPI at a national scale. Publishing more accessible formats requires little effort. The choice to withhold machine-readable formats was therefore a governance decision rather than a technical limitation.
The demographic direction of both exercises also followed a similar trajectory. In Assam, cracking divided Muslim voter concentrations across constituencies in Nowboicha, packing consolidated Muslim-majority constituencies into a single seat in Barpeta, and stacking reduced the Muslim share of the Barpeta electorate from roughly 60 percent to 35 percent. In Bengal, the correlation between logical discrepancy flags and Muslim population density across districts reportedly stood at 0.80. In Muslim-majority districts including Murshidabad, Malda, and Uttar Dinajpur, discrepancy flags operated at nearly three times the state average. In both states, demographic restructuring emerged through systems presented as neutral administrative exercises.
The burden in both exercises also fell upon the same populations. In Bengal, voters flagged by ERONET were required to establish their legitimacy through forms, hearings, and supporting records within compressed timelines. The burden weighed most heavily upon the mobile, the poor, women whose surnames changed after marriage, Adivasis, Dalits, and communities whose citizenship had already been politicised through the infiltrator discourse. In both exercises, participation increasingly depended on satisfying documentary standards that were unevenly distributed across society. In Assam, the NRC process had already compelled residents to submit extensive family and ancestry documentation tracing lineage to pre-1971 records, thereby producing a highly granular demographic database of the state’s population. Combined with electoral data and GIS mapping systems, this created conditions for constituency restructuring with extraordinary precision. The resulting boundaries were drawn in ways that enabled the BJP-led alliance to potentially secure a legislative majority even without winning the popular vote.
These episodes also reveal growing strains within constitutional institutions themselves. In Assam, the Supreme Court declined to stay the delimitation exercise before the 2024 general election and did not hear the challenge on merits in time for judicial intervention to meaningfully affect the electoral process. The petition challenging the SIR reached the Court in July 2025. At that stage, the Court had several months before the Bihar elections, and later, before West Bengal’s electoral rolls were frozen, within which it could have addressed the foundational constitutional questions raised by the exercise. It chose a different course. Instead, as Vasudev Devadasan has observed, it engaged in “dialogic review”: supervising the implementation of the SIR rather than adjudicating its legality. Successive orders directed the acceptance of Aadhaar, required the publication of deletion lists, and extended deadlines. Yet the central constitutional challenge remained unresolved before the Bihar election, leaving excluded voters without any effective remedy before polling.
A similar pattern unfolded in West Bengal. On 29 January 2026, the Court reserved judgment on the petitions challenging the legality of the SIR. This left sufficient time to deliver a merits ruling before Phase II of the SIR, which included West Bengal and eight other states and three union territories. Alternatively, the Court could have granted interim relief preserving the voting rights of excluded individuals pending final adjudication. It declined both courses of action. This occurred even as the Court itself expressed concerns regarding the algorithmic systems underpinning the SIR and acknowledged that the appellate process could address only a small fraction of pending cases before polling day. Across Assam’s delimitation exercise, Bengal’s SIR, and the wider rollout of the SIR framework across multiple states, the Court effectively preserved the status quo by postponing adjudication of the underlying constitutional questions until after the electoral process had become practically irreversible. This fits with the logic of the “Executive Court”, where institutions retain the vocabulary of constitutionalism while increasingly aligning with executive priorities.
The Election Commission of India is often criticised for active partisanship, but the deeper concern is structural rather than individual. As Gautam Bhatia argues, the ECI is an example of a weakly entrenched institution. Although constitutionally guaranteed, its independence and autonomy are not explicitly protected, leaving it vulnerable to executive influence. This vulnerability became particularly visible through the Chief Election Commissioner and other Election Commissioners (Appointment, Conditions of Service and Term of Office) Act, 2023, which made appointments to the Commission effectively executive-controlled. The legislation departed from the framework laid down by the Supreme Court in Anoop Baranwal v. Union of India (2023), where the Court had sought to strengthen the independence and impartiality of the ECI. Yet despite the clear constitutional implications, the Supreme Court declined to stay the legislation, a response consistent with the logic of the “Executive Court” model.
Several broader lessons emerge from these developments.
First, the architecture precedes the election: Electoral outcomes are now shaped long before polling day through administrative and technocratic processes involving constituency boundaries, demographic databases, electoral rolls, verification systems, and algorithmic sorting. In Assam, political competition was reorganised through delimitation and demographic restructuring of constituencies. In Bengal, the electorate itself was reconstituted through the SIR before a single vote was cast. Elections continue to occur formally at the ballot box, yet the conditions under which those elections take place are increasingly determined upstream through institutional and technological design. This is where the idea of structural authoritarianism becomes useful. Political power operates not simply through overt coercion or direct prohibition, but through infrastructures that organise democratic participation in advance and shape who enters the electoral arena, on what terms, and with what degree of institutional vulnerability.
Secondly, technology expands the scale of political management while obscuring its political character: GIS mapping, demographic profiling, algorithmic modelling, and centralised electoral databases permit electoral engineering at a scale that would previously have been administratively impossible. Yet these same systems also perform an important rhetorical function: they transform deeply political choices into apparently technical exercises. A constituency boundary cutting across rivers and disconnected settlements appears as a cartographic irregularity rather than demographic engineering. An algorithm flagging a Nobel laureate because of an age discrepancy reads as a software error rather than a structural problem within voter verification. The political consequence is absorbed by the machine, and the machine cannot be held accountable. This produces a compounding asymmetry within democratic competition. Incumbents possess access to state databases, digital infrastructures, institutional machinery, and the administrative capacity to deploy these systems at scale. Opposition parties, researchers, journalists, and ordinary citizens encounter systems they can rarely independently audit or effectively challenge. Accountability becomes diffused across software architectures, opaque methodologies, undisclosed workflows, and inaccessible datasets. The most consequential decisions in an electoral process, who is represented where and who is permitted to vote at all, increasingly become the decisions least open to public scrutiny or meaningful reversal. Digital structural authoritarianism operates precisely through this redistribution of power into infrastructure and procedure.
Thirdly, the burden of proof becomes the instrument of exclusion: In Assam, communities were required to establish a linkage to earlier electoral rolls and, in many cases, prove citizenship through documentary evidence connected to pre-1971 ancestry records for NRC. This data most probably formed one of the key informational foundations underlying the later delimitation exercise. In Bengal, voters flagged by ERONET had to establish their legitimacy through forms, hearings, supporting documents, and compressed appellate procedures. In both exercises, the formal language of electoral inclusion remained intact while the operational logic shifted toward suspicion and verification. The constitutional presumption underlying universal adult suffrage, namely that adult residents are entitled to political participation unless lawfully disqualified, gradually gave way to a system in which citizens repeatedly had to establish their own eligibility through bureaucratic compliance and documentary production.
This exclusionary logic became intertwined with the discourse surrounding “infiltrators”, where certain populations were increasingly treated as suspect unless they could repeatedly establish documentary legitimacy. In Assam, the delimitation exercise unfolded after years of political rhetoric centred upon demographic change, migration, and the protection of indigenous” Assamese communities. In West Bengal, the same logic surfaced through the SIR, most explicitly in Amit Shah’s declaration in Parliament that the government’s policy was to “detect, delete and deport.” Electoral exclusion, therefore, increasingly appeared as part of a wider administrative trajectory linking voter registration, citizenship scrutiny, demographic management, and civic belonging. Electoral exclusion increasingly became connected to broader questions of citizenship, welfare access, and civic belonging. These anxieties intensified further after West Bengal BJP leader Suvendu Adhikari reportedly stated that individuals excluded through the SIR process would lose access to government welfare benefits, and that the government would begin proceedings to deport alleged illegal immigrants. Electoral exclusion, therefore, acquired implications extending far beyond the franchise itself.
Fourthly, opacity functions as a structural feature: In Assam, constituency maps were released only as static PDFs without machine-readable coordinates or shapefiles, making independent analysis extremely difficult. In Bengal, the electoral rolls were similarly released as scanned PDF images, effectively photographs of printed pages accompanied by restrictive download limits that rendered computational analysis burdensome. The flagging logic of ERONET remained undisclosed despite the Election Commission acknowledging concerns regarding the reliability of its de-duplication software. The absence of transparency limited meaningful public scrutiny before electoral rolls and constituency structures acquired practical finality. Opacity, therefore, became integral to how these exercises functioned because inaccessible methodologies restricted both independent auditing and effective legal challenge. It also sharply limits our understanding of the actual electoral consequences of the Bengal SIR. The data necessary to conclusively evaluate the impact of the exercise, including booth-level results, a full accounting of additions alongside deletions, and clarity regarding who was removed, why they were removed, and how they may have voted, remains unavailable in public form. Digital structural authoritarianism relies heavily upon this relationship between technological governance and restricted visibility. Infrastructure acquires political force partly because the systems governing democratic participation remain inaccessible to those governed by them.
Finally, structural effects can outlast individual elections: Constituency boundaries, demographic classifications, electoral databases, and verification systems are difficult to revise once embedded within administrative infrastructure. They continue shaping political competition across multiple electoral cycles, producing durable asymmetries within democratic contestation long after the immediate controversy surrounding them fades.
These developments need to be understood against the backdrop of “competitive authoritarianism”, a term developed by Steven Levitsky and Lucan Way, and a framework that Christophe Jaffrelot has used to describe India under Prime Minister Narendra Modi. In this model, democratic institutions formally persist, and elections remain competitive, yet the ruling party acquires systematic advantages that progressively tilt the political field in its favour.
What distinguishes the current moment from ordinary electoral friction is captured in a line from Hemingway’s The Sun Also Rises. Asked how he went bankrupt, Mike Campbell replies: “Two ways. Gradually, then suddenly.” Democratic erosion frequently unfolds through a similar trajectory. Asymmetries accumulate incrementally — through administrative control, technocratic infrastructure, funding disparities, media concentration, and the weakened autonomy, especially of guarantor institutions — until the competitive balance of the political system has quietly shifted. Algorithmic governance and opaque administrative procedures fold into this same logic, forming a broader architecture through which democratic contestation is organised and, in the organising, constrained. The “suddenly” arrives when that architecture becomes visible: through exercises such as the Assam delimitation and the SIR, unprecedented in scale and compressed into an electoral timeframe that leaves little room for scrutiny or redress. The result is a system in which constitutional forms are preserved while the conditions necessary for genuinely free and equal political contestation are gradually — and then, with moments like these, suddenly — eroded.
