Which is in the lens, not in the camera. Build MF focusing rings into MF lenses, AF-optimized focusing rings into AF lenses, and that problem is taken care of.
You have a point about focusing screens, but those can be made changeable, as they are in professional cameras. Want MF lenses? Use an MF-optimized screen.
I used to think the same way as you do until I got a T90. There you have an MF-optimized camera with a brilliant user interface that breaks down all the complexity of the camera into a dial controlled by a few buttons, where you normally use just two or three, and after two days of using the camera you have it in your reflexes where everything is and your fingers will find the right place automatically. It would translate 1:1 into an autofocus camera (and in fact it has). Try one for a few days.
I think the main problem in this kind of thread is that people look at low-end digital cameras in their comparisons. In a professional camera suddenly you get dials again instead of menus, and you can control the camera while it's at your eye. Granted, your dial doesn't have printed shutter speeds on it, but otherwise it's all there.
The interface problem in a digital camera is managing complexity. In a simple 1970s-style SLR you can translate directly from every control to a function simply because there are so few functions. But people want extra features and have wanted them since the 1980s, even if some people don't want them. Camera designers had the problem of managing this complexity already in the 1980s and there were few cameras that did this well already back then, well before the digital age. In a digital camera, which is intrinsically more complex than an FM3a, I'd take the interface of an F5 over an FM3a any day.