I expect we'll have direct neural interfaces (DNI) (and thus a telepathic-like ability to communicate with our gadgets, wired and/or wireless) before we have real telepathy. That will help these devices, more than hurt them. No more keyboards at all -- just dictate via thought. Then the only things you'll worry/care about in device formfactor are:

screen size/resolution (assuming it doesn't have the ability to display via DNI as well)
battery life
CPU power/application compatibilty
WAN (type, affect on battery, etc.)

oh, and, probably security ... if not security of the device then security of the DNI.

Imagine a set of (sun)glasses, where the frame, arms, and strap behind the arms function as a non-invasive DNI (ie. a sensor grid, not a plug), and the glasses function as an augmented reality/HUD display, and there are ear-buds or earphones that connect to the arms. There would also be microphones in the frames, both for conversations and recordings, as well as noise cancellation and augmented reality/enhanced-hearing. There may also be 360 motion/optical sensors to help with proximity sensors around your body. Then you connect it (via wire or wireless USB or bluetooth) to a WWAN/PDA/phone gadget that you keep in your pocket. You'd control it through a combination of head tilt/orientation, voice, and brain activity. And, with the MIT finger-tip sensor UI, you could possibly add that to the control mechanisms.