Throughout history, the ability to deceive or manipulate the public was constrained by access — access to a printing press, a broadcast tower, a position of authority and, today, data. Despite lies spreading faster than ever, they are also exposed more quickly under scrutiny.
Fact-checkers, independent journalists, whistleblowers and everyday citizens now have digital tools to push back. Truth does not automatically triumph, but people are increasingly make it a point to insist upon it.
Following India’s general election in 2024, at when Prime Minister Narendra Modi won a very narrow mandate, three major state assembly elections took place: Haryana (2024) Maharashtra (2024) and Bihar (2025). Each electing state had its own political context, caste equations, political issues, grievances and aspirations.
Yet all three produced outcomes that seemed to align with uncanny precision toward one likely winner. The data trail is, at minimum, statistically remarkable and sometimes counterintuitive to conventional electoral logic.
What began as whispers of “strange numbers” turned into a wave of citizen-led analysis questioning outcomes that defied electoral gravity and ground realities.
While analyzing elections, data is crucial for understanding voting patterns. In data science, data is classified as labeled or unlabeled. Unlabeled data is raw, while labelled data is cleaned and structured. And clean, structured data reveals patterns.
Democracies, of course, are inherently noisy. Voter behavior is shaped by many micro-factors such as local issues, candidate appeal, turnout and even weather. Because voting reflects millions of independent decisions, the emergence of repeated, consistent patterns are highly unlikely under natural conditions.
When data looks too perfect and patterns repeat identically, it is no longer organic but curated or manufactured.
FPTP fairness
To assess electoral fairness, global institutions use the First-Past-The-Post (FPTP) system, employed in countries like the UK, Canada, the US and India. Metrics include vote share, efficiency ratios, close wins and turnout patterns.
However, in the last three assembly elections following India’s 2024 general election, the data was unusually uniform and improbably aligned across states. Out of ten FPTP metrics, anomalies appeared in eight. Let’s ştart with the recent Bihar election.
In elections, every number tells a story, NOTA (None of the Above) reflects voters who participate while rejecting all candidates. Unlike party votes, it is nonpartisan, independent and naturally random.
In most elections, NOTA varies widely. In the recent Bihar assembly election, however, 62 of 243 constituencies recorded NOTA counts tightly clustered between 3,000-4,000, (30 between 3,000-3,500 and 32 between 3,500-4,000) with eight constituencies showing identical counts in pairs.
Such uniformity is statistically implausible, with probabilities falling below 10⁻⁶. Historically, identical NOTA votes are a classic electoral red flag.
Too perfect to be random
The second key factor is high voter turnout. Historically in India and globally, high voter turnout signals anti-incumbency. Yet in Haryana (67.9%), Maharashtra (66.05%) and Bihar (66.91%), high turnout produced landslide victories for incumbents aligned with the Modi’s Bharatiya Janata Party-led National Democratic Alliance (NDA) — an inversion where high turnout favored those in power, a rare electoral pattern.
The third key indicator is vote-seat disproportionality. In large democracies, vote-to-seat conversion is usually messy, yet in these elections it appeared unusually optimized. In Bihar, 47% of votes yielded over 83% of seats for the BJP-led NDA; in Maharashtra, 49.6% produced 81.6% of seats; and with just 39.9% of votes, the NDA secured 53.3% of seats in Haryana, representing an extreme imbalance.
The fourth crucial indicator is the outcome of close races. In any competitive elections, winners decided by less than 5% of the vote are like the toss of a coin: sometimes the incumbent party wins, sometimes the opposition and it balances out.
But in all three states, the same party (Modi’s BJP) won the vast majority of such closely fought elections, defying randomness like a coin landing heads 45 out of 50 times.
The fifth indicator is the vote-seat swing relationship. In normal FPTP systems, a 1% vote swing yields a 3-6% seat gain. In Maharashtra, a 1.02% vote increase produced a 26% seat jump, a 25× amplification.
In Bihar, a +9.3% vote swing led to a +32.5% seat gain, an amplification of 3.5×. In a typical FPTP system, a +9% swing produces a +10–15% seat change. A +32.5% change is an aberration.
The sixth crucial metric is vote efficiency—how effectively votes convert into seats. In Bihar, 46.6% of votes produced 83.1% of seats (1.78× efficiency); Maharashtra reached 1.94× and Haryana 1.33×. In FPTP systems, efficiency ratios above 1.7–1.9 are statistical red-flags.
The seventh crucial indicator is voter turnout uniformity. Haryana (67.9%), Bihar (66.91%) and Maharashtra (66.05%)—states with very different political and social dynamics—recorded nearly identical high turnout clustered between 66-67%. In the absence of a national wave, such tight alignment across diverse states is anomalous.
The eighth crucial factor stable vote share but massive seat collapse for the opposition. In Bihar, the opposition received 38% of the vote, yet its seats dropped from 96 to 38, while the NDA alliance got 202 of 243 seats with 46.7% of the vote.
In Maharashtra, the opposition received 35.3% of the votes but only 49 seats, while the NDA won 235 of 288 seats with 49.6% of the vote—an enormous difference.
A peculiar pattern appeared across three states with strongly different backgrounds: high turnout, landslide victories, effective vote distribution, closed margins, gains for incumbents and a fragmented opposition. Such a degree of convergence at the national level in state elections is extremely rare.
It is also important to consider ground mood: anti-incumbency in Maharashtra, rural anger in Bihar and widespread dissatisfaction in Haryana. The results, however, completely defied expectations. When the weather forecast predicts a storm and the sky suddenly turns blue, you don’t blame the sky—you question the instruments.
Too many improbably events occurred in these three state elections to be dismissed as coincidence. Elections are not random, but they are not perfectly repeatable either. Eight out of ten rare red flags converged simultaneously—far too coincidence to ignore.
Democracy thrives on scrutiny, when citizens understand not just who wins but how and why. And democracy strengthens when some of the people refuse to be fooled, even for a time.
