I was going to ask the stupid question "why bother with special lenses to introduce unsharp parts if you can just destroy details in software still anyway" but

I was going to ask the stupid question "why bother with special lenses to introduce unsharp parts if you can just destroy details in software still anyway" but I think I realized the answer, in case anyone else has that question also: introducing it later depends on knowing which parts were further away. Pictures (typically) don't have depth/distance information for each pixel.