BSDF is used to add Lambertian diffuse transmission.”
Given that we see shadows on the backside of fabric curtains, I’ll go with translucency and diffuse mix for the weave itself, with transparency mixed in with a thread mask. A Shoji (japanese paper door/wall) will light up the room, but you can’t see through it or make out the details behind it no matter how close you go.
I don’t see any transparency (semi transparency) going on for a sheet of regular paper - at all. Same with softbox and lampshades. The light coming out on the backside is too scattered to make out anything going on behind it. Put a sheet of paper right in front of your eyes and try reading your screen - not happening. Yeah, I know semi-transparent paper exist, but it’s not what we normally consider paper.
They don’t have “holes”, they have matter where light bounces around inside and end up in all directions at the backside. Diffuse and translucency are both a very simplified way of describing volumetric behavior (with no volume required). Since the result of SSS doesn’t light up anything, the only way to light up the backside in a scatterly fashion is using translucency. You can replace translucency with very high roughness refraction and rely on caustics for light transport, but the roughness doesn’t go high enough and will always be light direction dependent and loose a lot of energy.
Well, at least that’s my take on it.