It is all about haptic feedback, and low latency between action and outcome.

You can twist virtually with good precision. But, what is much harder is the feel, state of the knob, particularly when it is released and then gripped again.

Mechanically, the structure of the knob can take some energy input, and it serves as a mechanical pivot, or fulcrum, depending on how people use a knob.

Without all those physical things, people lack the complex frames of reference needed for fine, "thought is action" type control.

> You can twist virtually with good precision. But, what is much harder is the feel..

This is especially true in environments where gloves are worn.

In addition, the essential ship controls must work in rough weather, so you want something you can grasp to keep your hand steady. So long as airplanes have human pilots, I do not suppose the physical stick, rudder pedals and throttles are going away.

Yes. And state comms. Where that input device is matters.

> actions like turning/twisting do not work well on touch screens

Have you ever used an iPod (capacitive touchpad) scroll wheel?

A good physical knob is still better, but a touch wheel can be made pretty decent.

The iPod wheel has a physical barrier acting as an affordance for your finger to trace. It certainly works better than it would just floating in space in the middle of a touch screen.

I wonder if it's much easier to process the signal (at the level of measuring and interpreting the capacitances) if you assume a one-dimensional user input. I'm not sure what signal processing happens when sliding on a touchscreen/trackpad.

A iPod touchwheel can be used without looking at it. It's much more like a knob than a touchscreen display, even if it uses similar technology.