The typical publishing methods kind of favors that approach of publishing new articles instead of updating existing ones though, for better or worse.

Maybe science journalism should just adopt a wiki-model instead, where there is one article per "subject" then any new (confirmed?) information/data goes into that, and interested people can subscribe to updates there instead.

Wikis generally have much better long-term maintenance given the right individuals running it, compared to a "publication journal" where things tend to get out of date eventually, with no way of actually seeing when old articles get updated.

No problem with publishing new articles, as long as they're properly contextualized and link to their predecessors – and the latter updated to link to the new information as well.