A minister stood in Parliament and named a number: 20 million. A statistician sat with the same country's Census data and found a different number: 2.7 million. Both claim to describe the same border, the same people, the same decade - and only one of them can be treated as fact.

The Number That Wasn't There A minister told Parliament that 20 million illegal Bangladeshi migrants live in India. A Census-based analysis, using the country's own enumeration data, finds 2.7 million people born in Bangladesh - documented and undocumented combined.

That is not a rounding gap. It is nearly a sevenfold difference and it forces a genuinely uncomfortable question for a democracy: which number gets to shape policy and on what basis? The case for taking the political estimate seriously starts from something real - the state has a constitutional duty to secure its borders and knows things intelligence and enforcement agencies gather that a household Census cannot fully capture: people actively evading enumerators, using forged documents or settling in ways designed to stay invisible.

A number like 20 million may function less as a precise count than as a signal of a genuine, felt security concern along a long and historically porous border. Dismissing it outright risks dismissing the underlying anxiety too.

The case for demanding the evidentiary standard is that Census data is not merely one estimate among many - it is the only instrument in the country built, specifically and rigorously, to count people, cross-checked by its own Post Enumeration Survey for exactly the kind of undercount concern critics raise.

That survey found a national omission rate of just 2.3% and a lower 0.84% rate specifically in the Eastern border zone - the area a real infiltration surge would concentrate in most. The age structure of the existing Bangladesh-born population is ageing, not growing; the geographic concentration in five border states has held steady even as the total figure fell.

None of this is what a hidden population of tens of millions would look like in the data. Where this leaves us: a claim of this scale - one that shapes citizenship law, electoral roll revisions and how millions of Indian citizens along the border are treated by the state - cannot rest on an unverified political estimate when a rigorous, cross-checked counter-instrument exists and says otherwise.

The security concern is legitimate; the specific number used to justify sweeping policy is not supported by the evidence available. A democracy that lets a claim shape law without meeting the evidentiary bar the state's own statistical machinery sets is not being cautious about security - it is lowering the standard of proof for its most consequential decisions and doing so precisely where the people affected have the least power to contest it.