33 Lakh Voters Questioned: What Delhi's SIR Notices Reveal About Its Electoral Rolls

Periodically, India's Election Commission orders a Special Intensive Revision (SIR) — a door-to-door audit of the voter rolls where field-level officers visit households, match records against family structures and previous entries, and flag inconsistencies. When something doesn't add up, a notice is issued.

In 2026, the Chief Electoral Officer (CEO) of Delhi published those notices as PDFs — one file per polling part, across all 70 assembly constituencies and 13 districts. 14,947 PDFs in total.

We downloaded and parsed all of them into a single table. What came out was a ledger of 33,11,746 notices — every person in Delhi's electoral roll who was flagged for a discrepancy.

Explore the data yourself: We built an interactive dashboard covering all 5 tabs — Overview, Geography, Reasons, Demographics and Data Quality. Open the Delhi SIR Notice Ledger →

The Scale: One in Five Delhi Voters

Delhi's electoral rolls carry roughly 1.55 crore registered voters. The SIR process questioned 33 lakh of them — roughly 1 in every 5.

That is not a rounding error or a bureaucratic technicality. It means the CEO's systems — cross-checking this year's roll against last year's, validating names, ages, and family relationships — threw a flag on 21% of the city's voter base.

Why Were They Flagged? Ten Reasons, One Dominant

A notice can carry multiple discrepancy tags. Across 33 lakh notices, there were 37 lakh reason-tags — meaning many people were flagged for more than one problem simultaneously.

The single biggest category:

"Unmapped with Last SIR" — 13.8 lakh notices (41.6%)

This is the most consequential flag. It means the system could not link the current entry to any record in last year's roll. The voter either didn't exist in the 2025 roll, or the linking algorithm — which tries to match on name, age, address, and family relationships — failed to find a match.

Three things can cause this: a genuine new voter who enrolled this year, a voter who moved and re-enrolled without deleting the old record, or a ghost entry — a name that was never a real person. The notice process is how the commission decides which it is.

Name Mismatches — 13.4 lakh combined

The second and third largest reasons are name-related:

• Parent Name Mismatch — 7.8 lakh (23.7%)
• Self Name Mismatch — 5.6 lakh (16.9%)

Together, name discrepancies account for more than 40% of all notice-tags. This is a chronic problem in Indian voter rolls: names are transcribed differently across documents — the voter's Aadhaar, ration card, and electoral roll may all spell the same name differently, making automated matching hard.

Family Relationship Red Flags — 5.9 lakh

The SIR system also validates biological plausibility of family entries:

• Progeny Age Gap < 9 Months (2.1 lakh) — two siblings listed with a birth gap under 9 months
• Parent Age Difference < 15 (1.6 lakh) — parent and child less than 15 years apart
• Progeny ≥ 6 (1.4 lakh) — more than 6 children listed under one parent
• Parent Age Difference > 50 (0.8 lakh) — parent listed as more than 50 years older than a child
• Grandparent Age Difference < 40 (0.7 lakh) — grandparent and grandchild less than 40 years apart

These checks catch either data entry errors (wrong ages typed) or deliberate manipulation (fictitious family clusters used to bulk-add voters).

Geography: Where Are the Notices?

Not all of Delhi is flagged equally.

By district

North East leads with 4.58 lakh notices (13.8% of Delhi's total), despite having only 8 of 70 constituencies. West (3.51 lakh) and East (3.40 lakh) follow. New Delhi district — which covers the government enclave, Lutyens' Delhi, and Cantt — has the fewest at just 37,594.

The gap between the most- and least-noticed districts is more than 12x, which is far wider than you'd expect from population differences alone.

The reason mix shifts by geography

This is where the data gets interesting. In most districts, "Unmapped with Last SIR" is the dominant reason — typically 45–55% of a district's notices. But in North East constituencies like Seelampur, Babarpur, Mustafabad, and Karawal Nagar, Parent Name Mismatch overtakes Unmapped as the top reason.

In Seelampur (AC 65), for example:

• Unmapped: 8,034 notices
• Parent Name Mismatch: 13,356 notices

And in Mustafabad (AC 69):

• Unmapped: 17,055 notices
• Parent Name Mismatch: 24,479 notices

One plausible explanation: these are dense, predominantly Urdu-speaking areas where names are transliterated into Roman script through different conventions (Mohammad vs. Mohammed, Abdul vs. Abdool). The matching algorithm interprets transliteration variations as mismatches.

Constituency breakdown

Among the 70 constituencies, the extremes tell a story of Delhi's uneven urbanisation:

Matiala — a constituency covering rapidly developing peripheral areas of South West Delhi — tops the absolute count among all 70 constituencies at 1.18 lakh notices.

Demographics: Who Gets Noticed?

Age

The median age of a noticed voter is 40 years. The age distribution peaks sharply in the 25–44 range, consistent with Delhi's overall age structure. There are, however, some striking outliers at the tail:

• 1,527 noticed voters are listed as over 90 years old
• 25 are listed as over 100
• The oldest entry in the dataset is 125 years old

Whether these represent genuine supercentenarians or clerical errors (a 2-digit year misread as 4, or an entry never deleted after death) is exactly the kind of thing the SIR process is designed to resolve.

Gender

The gender split is nearly balanced: 52.0% male (17.2 lakh) and 47.9% female (15.9 lakh). 228 voters are recorded as Other.

The gender ratio shifts noticeably by reason. "Unmapped with Last SIR" skews female (61.0% female), while "Parent Name Mismatch" skews male (73.6% male). The blank-reason category skews overwhelmingly male (97.0% male), which may point to data collection patterns in specific field offices.

Data Quality: What the Notices Themselves Get Wrong

When you parse 14,947 PDFs into a single table, you also capture the errors in the source documents. Several are worth flagging:

The blank-reason problem is concentrated in a few constituencies — Matiala (2,843 blanks), Okhla (2,180), Mustafabad (1,969) and Vikaspuri (1,861) together account for nearly 15% of all blanks. Whether these represent PDFs that were generated before reasons were populated, or a systematic gap in those areas' data pipelines, the CEO's office would need to investigate.

What This Means

A notice in the SIR process is not a deletion. It is a question mark — an automated flag that kicks off a field verification, after which the voter can submit documents to confirm their entry. Most entries flagged survive scrutiny.

But the scale — 33 lakh notices in one city, in one revision cycle — is a measure of how much uncertainty exists in the rolls, and how much work the commission's field machinery has to do every year just to keep the list honest.

Three things stand out:

1. Name matching is the weakest link. The 40%+ share of name-mismatch notices suggests the underlying matching algorithm is too sensitive to spelling variation. Better transliteration normalisation would reduce false positives substantially.

2. Peripheral constituencies carry a disproportionate burden. Areas like Matiala, Bawana, Mundka, and Badarpur — which have seen rapid in-migration — have the highest per-part notice rates. In-migration creates "Unmapped" entries naturally; the current process puts the burden of resolution on new arrivals who may not know they need to act.

3. North East Delhi's name-mismatch pattern deserves a dedicated study. The reversal of reason ranking in Seelampur, Mustafabad and Babarpur is too systematic to be random. Whether it is a transliteration artefact, a data-entry convention difference, or something else requires deeper investigation.

The Data, Open

The dashboard we built is fully interactive — filter by district, switch between Overview, Geography, Reasons, Demographics, and Data Quality tabs, and hover on any element for exact counts.

Open the interactive Delhi SIR Notice Ledger →

The underlying data is aggregated; no individual names or EPIC numbers are included. The source PDFs are public, published by ceodelhinet.nic.in, downloaded 26 September 2026.

This analysis was produced by InnerKore. If you use the data or the findings, link back and credit the source.