Found this through search years later and it is still the best discussion of the question, so adding one thing. Practical: whatever you change, write down the date and the reason. In three months the reason is what you will have forgotten, and the reason is what makes the record worth having.

anna.melb_AU said:

It helps to ask what evidence would change your mind before you look at any.

Coming at anna.melb_AU’s question from a different direction. Take it one variable at a time. Almost every unanswerable question in these threads is unanswerable because three things changed in the same fortnight, and no amount of subsequent argument can untangle them after the fact.

The figures, for anyone assembling their own picture. One habit that pays for itself: post the method alongside the number. A figure without its method cannot be checked, and an unchecked figure is how this community accumulates folklore.

I would rather be corrected than agreed with, if it comes to it.

Closing the loop on my own question.

Update, and it is the boring outcome: nothing was wrong. The measurement I was reading had moved less than its own week-to-week variation, which somebody said upthread and I did not believe.

Admin said:

Update, and it is the boring outcome: nothing was wrong.

This is where I part company with the consensus forming above. I think the framing smuggles in the conclusion. Ask it the other way round and the obvious answer reverses, which usually means the question is doing the work.