Actually this paper is basically re-proposing S2F in a different dress.

He claims the main mechanism for the floor is the inverse of issuance schedule but that leads an exponential because 1/2^-N is an exponential over the long term.
But as PlanB did he also claims there is power law in the data and the 2 don't go along at all unless you realize that when you look at clusters or few data points you can fit anything through it and call it a day.
I still like the entire proposal of using physics arguments for studying Bitcoin.