Fine, _it works in practice but does it work in theory?!_

It just goes to your point about interpretation. Provided the variable is described accurately, the amount of variation explained is whatever it is and you can note that it’s not particularly impressive because of how the variable was constructed. 

To me, it’s not very important because the relationship passes the eye test so dramatically. 

Someone predicted the power law trend almost a decade ago and it’s largely held.

Undisciplined

Mmm, noo?  By compressing it, first by MA and then by log, you're stripping it of variation. So I suspect you get this stupidly high r-squared result pretty artificially