I worked as a journalist for over a decade -- CBS, Time, CNET, Wired, etc. -- before founding
https://recent.io/ and every news organization I can think of updates articles after publication.
This practice predates the Internet by decades; wire services called updates "writethrus." Updated copy would be labeled internally for editors as 2nd-lede writethru, 3rd-lede writethru, with perhaps some details about what was changed or added, with those internal notes typically not being published for readers. So east coast papers would print a different version of the story than later-deadline west coast papers. The Washington Post did this in print form, with stories in the early "bulldog" edition often modified and expanded before appearing in the final edition. News has always been a snapshot in time.
There is significant pressure on reporters to post news quickly to capitalize on social sharing and search -- even if the quick first post is one or two paragraphs with little review by editors -- and update their articles quickly. This was the official policy in at least one newsroom where I worked. That can and does lead to inaccuracies and hasty reporting, though errors do tend to be corrected quickly. But the linked article does not accuse the NYT of factual errors.
It is true that it would be near-trivial for news organizations to have a history tab appear on each article that shows older versions -- I've suggested this and other features before, like providing people quoted in the article an automatic right of reply in a text field underneath the story. The reason we don't see history tabs is probably a combination of legal risk (imagine that a green reporter writes that John Doe was convicted of murder, and John Doe was the prosecutor, not the defendant), little demand from readers, embarrassment about errors in early versions, and newspaper-era newsroom thinking.