you are making an argument about the rate at which noise overwhelms an error prone measure. you are saying that 25 years isn’t enough. i’d say that about a decade is easily enough to render comparisons problematic to the point of uselessness. 1/

that’s not “my empirical argument breaking down”. it’s a dispute over the noisiness of a measure. you might be right or i might be, but perhaps it’s better to characterize the dispute charitably and accurately. /fin