Well,
You'll also note that those are single-purpose interfaces, similar to the function-specific controls for today's power plants or an aircraft carriers. Those consoles and control panels show specific controls only, ones dedicated to controlling generators or or navigating the ship. It's easier to present the information in advance when you only have to present a limited set of data - amps created, weapons status, or location, direction, and speed. You wouldn't have to wait on the Torpedo Firing Console, since it would be the only one displayed to the weapons operator. Even the newest cars with built-in Internet connectivity don't use it to operate the vehicle; the dedicated steering wheel and pedal retain their functions.
You don't see Geordi updating his SpaceBook page on the main Engineering console, or Worf using the weapons controls for multi-player games. In sci-fi, most of what would be termed 'general computing' is done via a voice interface, like interacting with the main computer or the holodeck. Unfortunately, we're a long way from effective voice interaction for general computing. I haven't tried Apple's Siri yet, but I'll bet it will be several years before you see them deploy it to their laptops. It can handle the functions specific to the phone, but will it interact with other applications, including web-based ones? That's critical to the mobility you mentioned as more applications move off the local processor.
I see limited consoles as a great place to deploy touch systems. Look at POS registers, ATMs, or tourist information displays. They have limited capabilities, shallow menu options, and are often used for short periods before moving to another task. None of these are running general purpose GUIs like a traditional desktop; that would definitely be out of place.
I agree about putting your eggs in multiple baskets.
