Thinking about the interface, right now we have a narrow, linear keyboard interface (okay, slightly 'chorded' when you consider simultaneous keyboard and mouse activities (shift-clicks, etc))

Lots of directions are being taken to make this pipe wider and deeper at the same time (voice control, intelligent appliances, etc.). Eventually, the best interfaces aren't going to be little command pipes connecting the human to the device. It's like a jacket. It works because of what it is. It stays with your body because of its construction. You don't have to tell it to keep you warm. You don't have to tell it to move its arm when you move yours.

The jacket is not smart.

But it does its job very well. So many 'ubiquitous computing' initiatives are about making very smart things, then hiding that intelligence so that it seems to just 'act naturally'. Maybe going the other direction isn't so bad: Making devices that are so dumb that we know how they work the way we know how a towel works. Give them small, yet powerful functionalities that don't currently exist, and manipulate them, combining their capabilities in ways people understand, not via APIs, interfaces, and coding linkages, but by stacking them, putting the jacket over your sweater, putting the 'heatball' in the water to cook it, or point the remote control at the TV and hitting the 'ON' button.

There's a good one right there: One remote with the basic cognitive tasks (play, channel up/down, volume, power, blah) that's directional enough that you point it at the unit you want to control. Point it at the TV and the TV's little red light glows a little brighter to say "I'm the one the Remote sees now" then you press the button you want. One remote, no modal interface.

This is obviously a rough cut, but I'm so busy at the moment (and for so many moments previously and henceforth) that I just wanted to get it out in any form, as a topic for discussion...

Your thoughts?