Among serious Mac users, the Touch Bar seems to be one of the more controversial moves Apple has made in recent years. The most common critique I've seen is that it simply lacks utility; that there still aren't enough apps which provide compelling functionality for it. Harsher criticisms have called for it to be gotten rid of altogether, or at least for a high-end MacBook Pro which retains the row of function keys that it replaced. I think there's a lot that could be done to improve the Touch Bar for its hypothetical second generation, though, at least in terms of the hardware complaints.
I don't own a Touch Bar MacBook Pro myself, though my brother does so I've used one quite a bit. I like it, but I don't feel like I'm missing out on that much with my non-Touch Bar 2014 model, and there's a distinctly unpleasant lack of tactility when using the Touch Bar. I'd sum it up by saying it just doesn't feel like a part of the keyboard, which it seemingly aims to. Perhaps the main reason for this is that you can't really touch type on it (no pun intended...), you have to look where you're placing your finger, because actions trigger as soon as it makes contact.
So how could it be improved? Well, I think Apple already has precisely the tools for the job: 3D Touch and the Taptic Engine.
The idea is simple: the Touch Bar should do nothing when just touched, and only trigger a function when the user presses harder.1 There could be some nice haptic feedback to match: a click like the iPhone 7/8 home button, or the Force Touch Trackpad. Running a finger along the ‘3D Touch Bar’ could also trigger little haptic taps as you cross the threshold from one button into another,2 and that would go at least some way towards approximating the tactility of a real keyboard. It might even, if it were good enough, allow navigation by touch alone. For more dynamic content, like sliders, it could obviously fall back to touch-and-drag behaviours, and haptic feedback would augment those interactions just like on iPhones. Since pressure sensitivity would be there, a more vigourous tap on the 3D Touch Bar (dare I say, like tapping a key on a physical keyboard) could trigger a button immediately, and skip the potentially slower touch-then-press method I'm getting at above.
A second-gen Touch Bar like this seems relatively easy to implement; it'd require only the addition of technology Apple is already producing in enormous quantity for the iPhone. Sure, it would be a change from the current interaction model, but when considering the huge changes watchOS has undergone in its first couple of years, and the relatively few Mac models which include a Touch Bar, I don't see that being a stopping block. Of course, there seem to be plenty of people for whom this kind of Touch Bar would be as bad as the current one, who simply want a high-end version of the MacBook Pro with function keys. Conversely, enough people may like the Touch Bar as-is, so maybe the incentive to change it just isn't there.
Finally, it's important to note that this doesn't directly address the other complaint I referenced in the introduction: the lack of compelling functionality. I think the core value proposition of the Touch Bar - dynamic keyboard buttons - is sound, and compelling functionality is more likely to appear if the underlying hardware is equally compelling.