In April 2024 the Indian government published 108 press releases. Across the other five full years of this whole record the April average is 392.

In May it published 160. That May average is 357. Nothing had gone wrong.





March is normally one of the two loudest months of the year, because the financial year closes. In 2024 it was 323, against a March average of 585 across 2020 to 2025. The quiet had already started.



While every other ministry went silent, one went the other way.



Forty-one in May, against an average of under ten for the year. Then zero in July.

There is the answer, sitting in the data. India held its general election that spring, and under India's election rules a government may not publicise its own work once voting is announced. The archive cannot show you that rule. It can show you every ministry stopping at once, and the one body running the election becoming the only one talking.



Most people reading any feed treat a gap as nothing happening. A gap is one of the loudest signals available.

Four to six hundred notices a month falling to 108 is not an absence of activity. It is a rule being obeyed. The activity continued — the offices were open, the work went on — but the publishing stopped, because publishing was the thing that was restricted.

So when a source you follow goes quiet, ask what rule might be operating. Silence has causes, and the causes are usually more interesting than the content you were expecting.



The election finding is only readable because one body moved opposite to everything else.

If all 26 ministries had fallen silent together you would know something happened but not what. The Election Commission climbing from 3 to 41 while everybody else dropped is what names the cause.

Second lesson: in any dataset, the exception explains the rule. When a broad pattern appears, immediately look for the thing moving against it. That one is carrying the information.



July 2024 carried 550 releases, above the July average of 507. December carried 570, with 21 releases titled Year End Review in it.

The machine did not ease back in. It resumed at full speed, and then closed the year with more of those self-reviews than any other December in this record.



Something held down by a rule rather than by circumstances tends to rebound once the rule lifts, and often overshoots.

Twenty-one Year End Reviews in December 2024, against 12 to 14 in every other December of this record, is a heavier catch-up than any other year required.

Third lesson: distinguish a fall caused by conditions from a fall caused by a rule. The first may persist. The second reverses on a date you can look up.

---



Silence is data. In the week to 25 September 2026, four of the 26 tracked ministries published nothing at all. Labour and Employment was one of them, after filing 18 items in July. That silence is small, but it is the same kind of fact.

The exception explains. In September 2026 the coal ministry hit its busiest month of that year, 34 releases, inside the year's quietest month overall at 356. One office rising inside a falling month is exactly the shape that pointed at the Election Commission in 2024.

A suppressed number comes back. Nothing in the current record is being held down by a rule. Worth remembering the next time a general election is announced: the feed will go quiet for about two months, and that quiet will mean a date, not a crisis.



For two months in 2024 the most informative thing in this entire archive was the material that was not published.

You cannot read an absence. You can only count one.

That is the whole argument for treating any public record as a series of numbers rather than a stack of documents. The documents tell you what someone chose to say. The counts tell you when they chose to stop.