Think of it this way - the screen, be it on a view camera where it occupies the very same plane as the film will or in an SLR where it has the same relationship to the intended plane of focus as the film does, is a barrier (due to it's granular texture) to coherent rays of light. That is, it's a barrier that separates the image forming optical system from whatever optical system is used to view the image.
In order to focus the image, you must be able to see the focussing screen clearly. That means a good quality loupe in the case of a view camera, and for a 35 it means the ability to focus your eyeball sharply on an object about 3 feet away - the apparent distance most viewfinder optical systems present to the viewer.
That's what I had always assumed, but then I wondered why all the fuss with diopters... but then, I've never had such bad vision that I couldn't see the viewfinder!