I was similarly wondering whether the author has tried taking a subset of the data and seeing if the result still comes out the same. It seems mildly suspect th

I was similarly wondering whether the author has tried taking a subset of the data and seeing if the result still comes out the same. It seems mildly suspect that, in each case, it just so happens that we've just about reached the limit. By excluding recent data, there can't be a bias to recent values