Essay · Section A29 September 2026

A rumour only has to be believed; the truth has to be proved.

Early in 1857 a story moved through the cantonments of the Bengal Army. The cartridges of the new Enfield rifle, which a sepoy had to bite open, were said to be greased with the fat of cows and pigs. The authorities denied it, then withdrew the greased cartridges and allowed sepoys to grease their own. It made no difference. By the time the explanation arrived, Mangal Pandey had fired at his officers at Barrackpore and the regiments at Meerut had risen. Historians have since found that the story was not wholly false: animal tallow had been used, and no one had asked where it came from. That detail matters. The rumour did not win because it was invented. It won because the sepoys already distrusted the Company, and the Company had given them no reason to believe its denials.

This essay argues that the contest between rumour and truth is unequal by nature. A rumour needs only a listener who is ready for it; the truth needs evidence, time and a witness whom people trust. We cannot make the truth as fast as a lie, and we should not try to do so by appointing the State as the judge of what is true. What a society can do is build witnesses before the crisis: institutions whose word is trusted, platforms that slow down what is dangerous, and citizens who have learned to ask for proof.

Why does a rumour need so little? Because its listener supplies most of the proof. In Othello, Iago never produces evidence of Desdemona’s infidelity; he produces a handkerchief and a hint, and Othello’s jealousy does the rest. Shakespeare put the point exactly: “Trifles light as air / Are to the jealous confirmations strong / As proofs of holy writ.” On 21 September 1995, reports that idols of Ganesha were drinking milk spread across India and to temples abroad within a day. Scientists explained on television that surface tension and capillary action drew the liquid from the spoon, and the explanation, though correct, travelled much more slowly than the wonder. A study by Soroush Vosoughi, Deb Roy and Sinan Aral, published in Science in 2018, examined about 126,000 stories on Twitter and found that falsehood spread “farther, faster, deeper, and more broadly than the truth”. The main reason was not automated accounts. It was people, who share what is new, surprising and frightening.

Technology has not changed the psychology; it has changed the cost. In July 2018, five men were beaten to death in Rainpada, a village in Dhule district of Maharashtra, after messages on WhatsApp warned of child-lifters. The messages were false, and more than twenty people died in similar attacks across India that year. WhatsApp responded by limiting how many chats a message could be forwarded to. Today the forgery is no longer a crude message but a video. In Ujjain and at two universities in Punjab, AI-generated clips and unverified messages have just set crowds against the police and against each other. A quieter version of the same problem appears in generative image models. Trained on an archive in which the 1980s were photographed mostly by the privileged, they produce a past that looks upper-caste and urban. A machine that has absorbed our prejudices returns them as though they were neutral facts, which makes it a very efficient spreader of rumour.

The truth, by contrast, needs a witness, and a witness has to earn trust. In 1998 The Lancet published a paper by Andrew Wakefield linking the MMR vaccine to autism. It took years of work by other scientists, and an investigation by the journalist Brian Deer, before the journal retracted it in 2010. By then vaccination rates had fallen and measles had returned in several countries. The Mahabharata shows what happens when the most trusted witness is used for a lie. Yudhishthira, who was known never to lie, said “Ashwatthama is dead” and added softly that it was the elephant. Drona put down his weapons. The epic says that Yudhishthira’s chariot, which had always moved just above the ground, touched the earth from that moment. Courts, science and a free press are our versions of that witness. Their value depends on the belief that they would not bend the truth even when it would be useful.

There is a serious counter-argument. A rumour is not always a lie. The sociologist Tamotsu Shibutani, in his book Improvised News (1966), described rumour as news that people improvise, the explanation people build when official channels are silent or not trusted. The sepoys’ rumour was partly true. During the Emergency, when newspapers were censored from June 1975, many Indians relied on the BBC and on word of mouth, and much of what they heard was accurate. Even the truth-tellers make mistakes: the story that Orson Welles’s radio play The War of the Worlds caused mass panic across America in 1938 was greatly exaggerated by the newspapers that reported it. A society that sets out to eliminate rumour will usually end up eliminating dissent. That is why the Bombay High Court in 2024 struck down the amendment to the IT Rules that allowed a government fact-check unit to label content about the Union government as false. The judge who gave the deciding opinion found the provision vague and liable to chill speech. John Stuart Mill argued in On Liberty (1859) that “He who knows only his own side of the case, knows little of that.” A State that decides what is true for its citizens takes away their practice in deciding it for themselves.

The objection does not defeat the essay’s argument; it defines the remedy. The answer to rumour is not a ministry of truth but a better supply of trusted truth. First, public institutions must speak early and plainly, because silence gives rumour its opening; a district administration that publishes its road-widening plan and its reasons before the bulldozers arrive leaves less room for a forged video. Second, platforms can add friction without censorship: limits on forwarding, visible labels on synthetic media, and quick takedown under law, as Shreya Singhal (2015) allows, for content that incites violence. Third, those who build AI systems should be required to audit them for caste and gender bias, instead of certifying themselves. Fourth, and most lasting, is education. Finland has built media literacy into its school curriculum, and its citizens are regularly found among the most resistant in Europe to misinformation. A citizen who habitually asks “How do we know?” is the only fact-checker who can work at the speed of a rumour.

The Company’s denials in 1857 failed not because they came late but because the people hearing them had no reason to trust them. That is the real lesson. Truth cannot win a race against rumour; it can only be prepared in advance. It needs witnesses who have earned trust before the crisis, and citizens who expect proof. A rumour asks only to be believed. A democracy has to ask for more, of its institutions and of itself.