The estimate is disputed. Most of them put it in the early thirties.

There was no announcement, because there was no instrument. Nobody counts sentences. What there was, afterwards, were three or four independent attempts to establish the year in which more text came into the world from generation than from typing, and they disagreed by about a decade, and the disagreement was itself the finding: it happened without producing a measurable edge.

The archivists at the deposit libraries noticed something before the statisticians did, and what they noticed was not volume. It was the disappearance of a particular kind of badness.

The record of a century is full of sentences that are wrong in specific ways — a regionalism, a misused word held confidently, a rhythm belonging to one person and no other, a paragraph that goes somewhere it should not have gone and is better for it. Those are how you date a document, attribute it, tell a forgery. They are the fingerprints.

They thinned first. Then they were rare. Then a document without them stopped being evidence of anything, because that was now what a document looked like.

One of the deposit librarians, interviewed near the end of her career, put it in a way that got quoted for a while. She said the archive had not been flooded. She said it had been smoothed, and that you cannot tell where a river has been once it is a lake.

The year has no name and no observance. Two later attempts to establish one were abandoned, in both cases because the working group could not agree which year, and in both cases the abandonment was written up in prose that could have come from anywhere.