Well done Arenamontanus! (Arenamontanus on these forums makes no mystery of actually being Anders Sandberg, and he's one of the authors)
Using a log-uniform distribution fitted to a collection of point-estimate values with different orders of magnitude is still making an assumption (i.e. why log-uniform?) but it's certainly better that using point-estimates.
I also like how neutral the paper is - it's not trying to push either thesis (there is or isn't life out there), it's just about the maths.