> That presupposes knowledge of more than one universe to even start to enumerate the possible parameters deciding the set
I think you entirely missed my point (because you didn't read the link!). You could use the universal prior, that is, you could enumerate all possible computer programs (e.g. Turing machine programs) starting from the shortest and find the shortest one that generates our universe. Now, 2^L where L is the length of such program is the prior probability of our universe.
So to determine e.g. the average value of the gravitational constant across all universes, you would measure the gravitational constant in all universes and calculate a weighted sum where you weight the value of the constant by the prior probability of the corresponding universe. The same for the occurrence of life, etc.