The Special Intensive Revisions (SIR) since June 2025 have removed over 11 crore names. The process has been criticized for false-positive errors (removing citizens from the voter rolls), but also for its onerous documentation requirements and high-effort processes.
An Electoral Roll Operating System is
required. India can maximally automate the SIR and establish a high-integrity
process going forward. Most of India's adults can be added into voter rolls from
existing records without a document demand. The remaining will be reached by a
welfare-first outreach programme that enrols them in the schemes they are
missing and registers them to vote.
The system will have five functions: Identity, Citizenship, Residence, Eligibility, & Lifecycle Maintenance. It will continuously maintain the electoral roll rather than requiring repeated rebuilding through periodic SIR exercises.
Data inputs
The government has direct access to these
data sources that feed the proposed system. The system, itself, will present UI
for data inputs and updates. This system makes it a citizen-friendly and
transparent mechanism. Reliance on numbers such as Aadhaar, PAN, Mobile,
etc. and their interlinks avoid the problems of name-matching. The Voter ID is
a number itself.
|
Source |
Scale |
Proves |
Role
in engine |
Limits |
|
Aadhaar |
145.32 crore generated (Sept 2026) |
Identity, current address |
Identity
key and dedup; residence evidence |
Not citizenship; dead holders not always
deactivated |
|
PAN, income tax |
Over 50 crore PANs (2020) |
Identity, father's name |
Family
edge |
Issued to foreigners too |
|
Mobile |
128.2 crore mobile subscriptions (June
2026) |
Contact; KYC identity; dominant
night-time location |
Outreach
channel, consented OTP and home-location code for precinct |
Subscriber is not always the user; tower
precision varies; needs a legal basis (the ask) |
|
Passports |
About 10% of the population holds one |
Nationality for travel; issued after
police verification |
Strong
anchor, rebuttable |
MEA said in June 2026 a passport is a
travel document, not conclusive proof of citizenship; skews urban and
better-off |
|
Births & deaths registration (CRS) |
99.1% of births and 99.4% of deaths
registered in 2024 |
Birthplace, parents; death |
Anchor
for under-26s; death purge for all |
Weak for adults born before ~2000 |
|
MHA citizenship grants |
Small |
Citizenship, directly |
Anchor
for registered and naturalised citizens |
None material |
|
Old electoral rolls (2002–04 intensive
revision) |
Every elector of that era |
Prior enrolment as a citizen |
Anchor:
elector or child of one |
Pre-2003 names need fuzzy matching across
scripts |
|
School records (UDISE+, admission
registers) |
Most adults under ~45 |
Date/place of birth, parents' names |
Anchor (ECI already accepts educational
certificates) + family edges |
Older rural women least covered |
|
Driving licence (Sarathi) |
Issued by state transport offices |
Identity, date of birth, address |
Residence corroboration |
Issued to foreign residents too;
addresses often stale |
|
Bank, Jan Dhan, insurance, fintech (incl.
PF, PPF, etc.) |
Jan Dhan: 59.09 crore accounts (Aug 2026) |
Identity, address |
Continuity of residence only |
KYC is mostly Aadhaar, not independent
evidence |
|
ASHA / HMIS birth records |
Births since ASHA launch in 2005 |
Birthplace, mother |
Corroborates the 18–21 cohort — exactly
the cohort needing both parents' status |
Fragmented; not a legal birth record |
|
Ration (NFSA), LPG, pension, MGNREGA |
NFSA: 81.35 crore people on 20.29 crore
cards |
Household composition, long-run address |
Family
graph and continuity of residence |
2.49 crore cards cancelled since 2020,
some wrongly |
|
Immigration records (Bureau of
Immigration entry/exit, FRRO registrations, visas, OCI cards) |
Every foreign national lawfully in India;
biometrics mandated under the Immigration and Foreigners Act 2025 |
Foreign nationality; long absences abroad |
The
only adverse-evidence source: a verified match is referred to the ERO |
Name-only matches are unreliable; needs a
document-number or biometric match |
|
Land revenue records (record of rights,
mutations), registered deeds, property tax |
99.81% of records of rights computerised;
37 crore+ parcels with a Bhu-Aadhaar (ULPIN) number (Jan 2026) |
Decades of presence; holdings passed from
parent to child |
Family
edge and pre-1987 presence for older rural adults |
Skews male and landed: the landless
majority of T3 is absent; title disputes; urban mapping only begun |
The First Ask: mobile home location for the precinct
- What is asked. Telecom operators compute, for each Aadhaar-linked subscriber, the dominant night-time location over the trailing 180 days. They return only a polling-part code (or a short list of candidate parts) and a pattern flag: stable, seasonal, moved or split. No tower logs, timestamps or trajectories leave the operator.
- Why. Aadhaar and roll addresses lag real moves by years. Mobile is the only signal that is both near-universal and current.
- Accuracy limits. The subscriber is not always the user: a SIM in a husband's name is used by his wife, a parent's SIM by a student. Tower location is coarse (hundreds of metres in dense cities, kilometres in rural areas) so in cities it narrows the choice to a few parts and the address decides. With several SIMs, use the Aadhaar-linked number with most night-time activity. People with no phone or a shared phone fall back to declaration and records.
- Rule of use. Mobile location can raise a prompt and break a tie between records.
- Legal basis needed. No current law lets ECI obtain subscriber location for roll maintenance; the Telecommunications Act 2023's disclosure powers are framed around public safety and emergencies. The ask is an amendment to the RP Act 1950 (or rules under it) that names the derived field, limits its purpose to the roll, deletes it after each revision, provides for audit, and lets anyone opt out to declaration-only. The derived-only, opt-out design is what makes the use defensible under Puttaswamy; expect it to be challenged regardless.
The Second Ask: presuming citizenship from residence records
The electoral-roll problem exposes a citizenship-status problem that cannot be solved by demanding documents from people who were never documented. Parliament therefore must decide whether sufficiently long, continuous state-recorded residence should create a rebuttable presumption of citizenship.
A one-time legal presumption should treat a
long, continuous record history in India as citizenship, so that the the least-documented adults (likely the most vulnerable) are not pushed into exception processes to find
documents that were never issued.
- What is asked. A rule under the Citizenship Act or the RP Act 1950: an adult with continuous, independent records in India for a set period, and no adverse immigration record, is presumed a citizen for electoral registration. The presumption is rebuttable only by evidence, with the burden on the state. The threshold is 12 years, matching the residence Section 6 already requires for naturalisation, counted from the person's own first record across at least two independent sources.
- Scheme age is not record age. Aadhaar, NFSA and Jan Dhan are all 12 or more years old, but each person's history starts at their own enrolment. Jan Dhan grew from 14.72 crore accounts in 2015 to 59.09 crore in 2026, so most accounts are younger than the scheme. UIDAI and the other custodians can count exactly how many adults clear 12 years from enrolment dates; that is the pilot's first query.
- The trade-off. It will include some long-resident people who entered without documents and have no immigration record. At a 12-year threshold this is in substance a regularisation: it grants by presumption what Section 6 grants by application, without Section 6's other conditions, including renouncing another nationality (India does not allow dual citizenship). That is the policy choice Parliament makes; the counterweights are that the presumption is rebuttable and the immigration-record check still runs.
- What it changes. People with long record histories move from T2 and T3 to automatic inclusion with no doorstep check, and anyone who clears the threshold no longer needs to be matched by name to pre-Aadhaar documents. The pilot measures how many people move.
- Carve-out. Assam is governed by Section 6A of the Citizenship Act (the 25 March 1971 cut-off) and the NRC process and needs separate treatment.
Adverse evidence
Immigration records are the one source that
can show someone is not a citizen, so
they are the one source allowed to trigger review. A verified match to a
foreign national's visa, FRRO registration or OCI card goes to the ERO as a
referral. Exit records for Indian citizens work the other way: someone living
abroad is still a citizen, so they are routed to overseas-elector registration
under Section 20A of the RP Act 1950.
Coverage estimates
An estimated 85-96 crore adults (83–94%) resolve from data alone; 9-15 crore need outreach. These are planning ranges built from the source coverage above, not measurements. The system, as it pilots and deploys, will replace these estimates with actual data.
|
Tier |
Evidence
state |
Est.
adults (crore) |
Who
is typically here |
Action |
|
T1 Established |
Documentary anchor on self or parent:
legacy roll, passport, CRS birth, school certificate, MHA grant |
60–68 |
Settled households; most urban middle
class; under-26s with birth records |
Auto-confirm; pre-filled Form 6 or EPIC
confirmation |
|
T2 Corroborated |
No direct anchor, but family edge to a T1
citizen, or 10+ years of independent multi-source records from childhood |
25–28 |
Rural households with ration and MGNREGA
histories; older adults born before birth registration |
BLO confirms at doorstep; no document
demand |
|
T3 Incomplete |
Aadhaar plus short or fragmented
footprint; no family edge |
8–12 |
Inter-state migrants, women who moved at
marriage, urban informal workers, nomadic and denotified tribes, transgender
persons, care-leavers |
Welfare-first outreach |
|
T4 Unlinked |
No Aadhaar, or records that cannot be
linked |
1–3 |
Homeless people, remote forest and hill
habitations, people with severe disability, elderly without family |
Welfare-first outreach, in person |
Migrants and Duplicates
Migrants
The roll's largest accuracy error is
location, not citizenship: people still registered where they used to live, and
the same person registered in two places. Mobile-derived home location is the
one source that shows where a person sleeps now, so the design uses it for the
precinct, subject to the legal ask below.
Three
kinds of movers need three different answers.
"Migrant" covers people whose ordinary residence has not changed and
people whose residence has.
|
Pattern |
What
the data shows |
Ordinary
residence (RP Act s.20) |
Engine
action |
|
Seasonal migrant anchored to home (brick
kilns, sugarcane cutting, construction seasons) |
Night-time location returns to the home
cluster every year; family stays on the home ration card |
Home: temporary absence does not end ordinary residence |
Keep the home entry; no prompt to shift;
tell them polling dates in advance |
|
Settled mover still registered at the old
place |
Night-time location in the new place for
6+ months with no return; ration drawn there; children in a new school |
New
place |
One-tap prompt to shift (Form 8); the old
entry closes as part of the transfer |
|
Split or ambiguous (weekly commuters,
households split between village and city) |
Two clusters, neither dominant |
Person's
declaration |
Ask; the declaration wins |
The 6-month threshold and the "returns
every year" test are proposals for the pilot to calibrate.
One
person, one entry. Sections 17 and 18 of the RP Act
1950 bar registration in more than one constituency or more than once in one.
Duplicates arise because a new Form 6 does not close the old entry. The fix:
- Find them. Dedupe all state rolls on the Aadhaar key, with photo and demographic similarity matching for entries not linked to Aadhaar.
- Pick the right one. Keep the entry that matches current ordinary residence: the declaration first, then mobile home location, then other records.
- Close the other as a transfer. The person is told by SMS and has 30 days to object. They lose their second entry, never their vote.
- Stop new ones. Every new Form 6 checks the national index and becomes a shift, closing the old entry in the same transaction.
- Offer an amnesty. Duplicates surrendered in the first cycle face no action for false declaration under Section 31.
Duplicates
Reaching the unlisted: welfare first
The people in T3 and T4 are thin in the
data because they are thin in the state's services, so the first visit offers
them services, not scrutiny. Each enrolment also generates the independent,
dated records that later support their voter registration — the welfare visit
is the evidence-building visit.
Who
visits. People the household already trusts: ASHA
and anganwadi workers, SHG members under DAY-NRLM, and the Booth Level Officer
as part of the same team, not alone. In cities, add ward offices, labour unions
and NGOs working with construction, domestic and gig workers. Each team carries
Aadhaar, banking and CSC enrolment kits.
The
visit, in order:
- Ask what the household lacks. A checklist screens for each scheme below.
- Enrol on the spot in whatever the household is eligible for and missing.
- Fix the records that will matter later: late birth registration, the woman's address moved from her natal to her marital home, the child's school record linked.
- Offer voter registration last, with Form 6 pre-filled from what was just gathered. Declining is fine and recorded only as "not now".
- Close the evidence gap with help, not demands. The team identifies which accepted document is nearest to hand and helps obtain it; free legal aid from the District Legal Services Authority covers anyone who needs a certificate or an inquiry.
- Return in 90 days to deliver cards and finish what is pending.
What
the team can enrol people in:
|
Programme |
Gap
it fills |
Record
it leaves |
|
Aadhaar enrolment or update |
T4 has no identity key |
Identity; current address |
|
Late birth registration (RBD Act) |
Adults born before universal registration |
Birthplace and parents — the core Section
3 fact |
|
NFSA ration card; One Nation One Ration
Card for migrants |
Food security; portability |
Household composition; residence |
|
PM Jan Dhan account |
Direct benefit transfers |
Dated KYC; continuity |
|
e-Shram registration |
Informal and migrant workers |
Occupation; work location |
|
Ayushman Bharat PM-JAY |
Health cover |
Household listing |
|
PM Ujjwala LPG connection |
Clean cooking fuel for women |
Woman-held, address-linked record |
|
NSAP old-age, widow and disability
pensions; UDID card |
Elderly, widowed and disabled adults |
Age, residence, disability status |
|
Transgender certificate (national portal) |
Identity in self-identified gender |
Identity document that matches the person |
Rules that keep it compassionate
- No citizenship questions at the door. The team records facts: birthplace, parents, years at the address. The engine and the ERO apply the law later.
- Language and dignity. Scripts in the local language, women team members for women-headed households, and home visits for the elderly and disabled.
Safeguards and governance
The two errors are not symmetric: wrongly
removing a citizen takes away a constitutional right, while a non-citizen on
the roll is a vote that the existing objection process can catch. The
safeguards are weighted accordingly.
|
Safeguard |
Mechanism |
|
Proportionality (Puttaswamy) |
Only fields needed for the Section 3 test
and residence are queried; location only as a derived home-location code
computed inside the operator, never raw telemetry; no religion, caste or
income fields |
|
Federation, not aggregation |
Custodians answer yes/no attestations
through a consent layer on India Stack (DigiLocker, Account Aggregator
pattern); ECI holds results, not source data |
|
Purpose limitation (DPDP Act 2023) |
Attestations are usable only for
electoral registration; logs record every query |
|
Transparency to the person |
Tier and reason codes visible on the
Voter Helpline app; one-tap correction request |
|
Public reason-coded statistics |
Every addition and removal published by
part, with reason, so parties and civil society can audit before the roll is
final |
|
Independent audit |
A random sample of T3/T4 people is
visited by an independent team each cycle to measure how many were eligible
citizens |
|
Disparity monitoring |
Gender ratio, age-cohort and
SC/ST-reserved-constituency coverage tracked by district; a district
breaching thresholds pauses removals |
Sources
•
UIDAI
enrolment dashboard — 145.32 crore Aadhaar generated, updated 28 Sept 2026
•
Special
Intensive Revision (Wikipedia) — Phase I and II electors and deletions; May
2026 Supreme Court ruling
•
SIR
Phase III deletions (SuperKalam, 1 Sept 2026) — over 6 crore names, 17.04%
•
India
has 99.1 crore voters (Deccan Herald, Jan 2025) — electors and gender ratio
•
SC
directs ECI to accept Aadhaar for Bihar SIR (The Week, Sept 2025) — Aadhaar
is identity, not citizenship
•
CRS
Report 2024 (Drishti IAS) — 99.1% birth and 99.4% death registration
•
PMJDY
at 12 years (Outlook Money) — 59.09 crore accounts
•
Ration
card deletions since 2020 (Outlook) — NFSA 81.35 crore people, 20.29 crore
cards
•
TRAI
June 2026 subscriber data (TelecomTalk) — 128.2 crore mobile subscriptions
•
PAN–Aadhaar
linking (Patrika, 2020) — 50.95 crore PANs allotted
•
Passport
is a travel document, not proof of citizenship: MEA (The Goan, June 2026)
•
1.5
crore passports and services delivered in 2025 (News On Air) — about 10% of
the population holds a passport
•
The
road to DILRMP 3.0 (The Week, Aug 2026) — 99.81% of records of rights
computerised; 37 crore+ Bhu-Aadhaar parcels
•
New
immigration rules mandate biometric data of foreigners (Tribune) —
Immigration and Foreigners Act 2025 and its rules
No comments:
Post a Comment