Essay · Section A18 September 2026

A document is only as true as the institution that issues it.

A construction worker in Odisha walks to his ration shop and presses his thumb to a scanner. The machine reads the ridges of a living hand and releases his grain. Somewhere in another government database, the same man has been dead for months — a death certificate issued in his name, a nominee paid two lakh rupees, a file closed. The paper says he is gone; his thumb says he is here. Both are official. Only one of them is true.

We tend to think of a document as a container of fact, the way a bottle contains water. It is not. A document is a promise made by an institution: that someone with authority looked, verified, and is prepared to be answerable for what is written. The ink carries no truth of its own. What it carries is the credibility of whoever signed it. When that credibility is sound, a piece of paper can do extraordinary things — it lets strangers trade across oceans, lets a stranger's child inherit land, lets a citizen vote in a constituency where no one knows her face. When the credibility rots, the same paper continues to look identical, and that is precisely what makes it dangerous.

The record is one of the oldest technologies of the state, and it has always been an instrument of power before it was an instrument of convenience. Kautilya's Arthashastra devotes whole chapters to the superintendent of accounts, because a kingdom that cannot count cannot tax, and one that cannot tax cannot defend itself. The Domesday Book was compiled not to describe England but to make it legible enough to extract from. India's own colonial land settlements did something subtler and more lasting: they took custom, memory and shared use, and converted them into title. Whatever the village had known for generations became true only if it appeared in the settlement register. Millions of tenurial relationships that had existed for centuries simply ceased to exist because a clerk did not write them down. The document did not record reality. It replaced it.

That replacement is what makes the integrity of issuance a constitutional question rather than an administrative one. Consider three failures that differ in subject and are identical in structure. A medical officer certifies a death that did not occur, and a welfare fund pays out on it. A police officer prepares the file authorising a detention after the person is already in custody, so that the satisfaction the law requires is manufactured to fit an act already done. An electoral authority marks a third of a city's voters for deletion under a category it has never defined, so that the affected citizen cannot know what case to answer. In each, the paperwork is flawless. The signatures are genuine, the formats correct, the registers updated. What has failed is not the document but the institution behind it — and because the document looks exactly as it should, the failure is invisible to every system downstream that trusts it.

Digitisation has raised the stakes of this old problem in a way we have not fully absorbed. An error on paper used to stay where it was made. A wrong entry in a tehsil register inconvenienced one family until someone corrected it. An error in a linked database propagates at the speed of the network and, worse, becomes self-confirming: the death record in one system invalidates the pension claim in the second, which fails the eligibility check in the third, and by the time the living man protests, three independent-looking systems agree he is dead. He is asked to prove otherwise against a consensus that is really a single error, copied. The phrase every citizen has learnt to dread — that the computer is not accepting it — is the sound of an institution abdicating to its own record.

It is tempting to conclude from this that we should trust documents less and human judgement more. That conclusion would be a serious error, and the honest essay must say so. The alternative to the register is not freedom; it is patronage. Where entitlement depends on being recognised rather than recorded, the person who suffers most is exactly the person with least social standing — the migrant, the widow, the landless, the low-caste claimant whom nobody in the office knows and nobody vouches for. Documentary systems, for all their cruelty, were built to displace precisely that discretion. India's own experience with muster rolls, direct transfers and digitised land records shows real gains against leakage that no amount of official goodwill had previously delivered. The problem with the file is not that it exists. It is that it has become unanswerable.

What a trustworthy system needs, then, is not fewer records but contestable ones. Three properties do most of the work. First, reasons: a decision that affects a right must state the ground it rests on, because an unstated ground cannot be challenged and therefore cannot be wrong. Second, independence: registers must be cross-checked against other registers maintained by different authorities with different incentives, since the fraud that is invisible inside one database is usually obvious the moment two are compared — a man cannot be dead in the welfare system and authenticating his fingerprint in the food system on the same morning. Third, and most neglected, a remedy the affected person can actually reach. A right of appeal that requires a labourer to travel to a district headquarters, hire a lawyer and file within thirty days is, for him, not a remedy at all. It is a formality that protects the institution from criticism rather than the citizen from error.

There is a deeper principle underneath these mechanics, and it is worth stating plainly because it is the one that administrative practice most often forgets. The human being outranks the record. A record exists to describe a person; when the two disagree, the burden of proof belongs to the record. Our systems invert this every day. They ask the living man to prove he is alive, the citizen to prove she belongs, the acquitted to prove he is not a habitual offender. Each inversion is defended as prudence, and each quietly shifts the cost of the state's own mistake onto the person least able to bear it.

Gandhi's talisman asked the administrator to recall the face of the poorest man he had seen and to test every decision against whether it would help him. The test can be restated for an age of databases without losing anything: can that man contradict the state's file about his own life, and be believed? Where the answer is yes, documents are what they were meant to be — a servant of truth, scaled up so that it can reach people the state will never meet. Where the answer is no, the file has stopped describing the citizen and started defining him, and no volume of digitisation, dashboards or seamless integration will make that arrangement either efficient or just. The measure of a republic is not how well its records are kept. It is how easily an ordinary person can correct them.