>>100546319I did. Took ~3K years.
From my conclusion, I will say that I am not a statistician. In reality, this is probably a terrible poll, there definitely needs to be some kind of t-test, and it might not come to the right conclusion.
However, after hand-tallying the times from Holodex, converting them to PDT, and trying to figure out how to find the IQR in excel, I made it.
https://files.catbox.moe/fr40zb.xlsxThe main things missing from it are some special events that I deliberately left out, like the 3D debut, and most of the collabs. (I may have missed some, but I am not going to check)
From the looks of the median and IQR (which is probably more accurate for this) it appears that there has been a shift towards later streams. The two datasets are decently close, they both have ENReco and a JP server in them, but there probably are factors that haven't been accounted for.
