Shader array is good, thanks for pointing that out! At least that would save you from duplicating masks that are aligned in a grid. Scaling is still an issue, though.
So let's just say I have a number of rings of different sizes (as masks) that are dispersed along the horizontal plane in some non-grid order. My idea was that I could construct the initial ring and then multiply it and scale it and place it wherever I want dynamically using the math functions (not by simply duplicating the nodes and changing their params). If all the rings are supposed to have the same width I could make it an input param (constant), so that I would have to change it only once in the project instead of changing the size in each SSS node. Makes sense?
More and more it seems to me that an image-based mask may just be necessary in this situation if I want to avoid SSS hell