Relative size (distance) is often correlated with diffusion/softness, as if the only thing you have to worry about/change is the distance of the light/modifier... which is wrong because distance significantly impacts the light output requirement, the resulting light falloff, and significant characteristics of the resulting image.

Using modifiers/diffusion is also commonly referred to as "making the light source larger." This is also wrong. What diffusion actually does is break a singular strong light source into a much greater number of smaller/weaker light sources... and it is the position/angle from which those smaller light sources see the subject that allows them to see/light around edges, creating the "soft light" wrap/fill effect.

But if the diffusion is placed far enough away to where the angle is too shallow for them to see around edges (~4x the modifier's size), then only the parallel rays see the subject and it will again be "hard light."

Thinking of a softbox as a larger array of light sources, rather than just a singular/larger light source, also helps to make more sense out of techniques like feathering the softbox, crating/using uneven diffusion, etc...

Likewise, because the diffusion is creating a greater number of light sources; it is moving farther away from being a point light source and it won't exactly follow the Inverse Square Law... especially at much larger sizes/shorter distances (but the ISL is usually still close enough to be useful).

Diffusion isn't an absolute requirement for "soft light", e.g. a bounce umbrella also breaks the light into many more light sources that can see around edges.