I think the touch pad is going to be one of those things people will just have to try, and it might turn out to be an awesome addition to the input tools at hand. Or it might one of those things that sounded good in theory, but is just a pain in practice.
One thing to consider is that things like editing video are already mouse centric, and the keyboard shortcuts augment the interface. You don't leave your hands in home position.
If you're working on the laptop, using the trackpad, you're already in the mode where you are going back and forth, and your hands need to move around.
When using a mouse, trackball, etc, you can get into workflows based on having one hand operating the mouse, while the other does the keyboard shortcuts. So there could be very interesting uses with two hands.
What makes it not so compelling to me, is that I do serious work with my laptop closed, and a big monitor (I'm music producer/audio engineer).
The amount of information you need to get on the screen at once, if possible, for media authoring applications, requires a relatively rigid focus on the screen. Having my head hunched over a little laptop display is not going to work for extended sessions.
The problem of creating physical interfaces that work well with the screen relates to the abstraction that is involved in GUIs. You have to keep steady focus on the screen, keeping track of the mouse, and what you are doing. You enter kind of a trance, pretty quickly, where you have accepted a two dimensional representation of a complex structure, and are operating within its framework.
This divorces the metaphorical actions of the user in the GUI, which is what we care about, from the physical actions of the body.
For a new user interface to work, it will have to be incorporated into this metaphorical framework the user operates in.
If it deviates to much from the abstractions of the interface, it breaks the trance required to get into the flow, and will simply not be used.
I suspect one of the issues will be if users can actually quickly look down to see what they are touching, or perceive enough with peripheral vision, without losing track of where they are on the screen. It's a reasonable location for the touch bar. As the user gets used to where controls on the touch bar are located, the visual cues might be enough to consistently target the virtual control.
It could also be really helpful for keyboard commands that have a lot of modifiers, and which you don't use much. They can be real handy at times, but the trouble of looking up the key binding in the heat of the moment prevents the habit from getting built up.
