NextFest Thin, bendable computers are coming - and they present a unique challenge for human interface designers. Flexible components have already been demonstrated, such as OLED displays that could be produced as thin as 0.2 mm. But the tricky part is finding a new interaction model that makes the most of the malleable machines. Clearly, when you can twist a computer, then a mouse or pen aren't necessarily the most optimal ways of communicating. So then, how?
At San Francisco's NextFest expo, Sony's Shinagawa-ku Interaction Labs shared what they'd learned from their Gummi project, which formed a presentation at the CHI conference in Vienna last month. Credit card-sized bendables are perhaps only five years away, researcher Carsten Schwesig told us. Gummi itself is a testbed using a TFT display, with no buttons: the user holds it with both hands and twists. Like a WIMP interface, it's more complicated to explain than do. In place of clicks and drags are five states: a neutral position, up, down and two transition states. Gummi has been a platform for a web browser, but we used a navigational map. This works surprisingly well, it's perhaps the easiest form of navigating an electronic representation of a map we've tried.
In Gummi, the designers mapped target down motions to selecting an object, or following a hyperlink, or starting playback, with a transition down mapped to a zoom. How did the team make choices? The answer was another surprise.
"It's like scroll bars on a 2D human interface," said Schwesig. "About fifty per cent of say up and the other say down." It depends on your point of view, it seems. Either you think the map is in the machine, in which case bending the screen towards you zooms out, or you think the machine is a map, in which case an upward manipulation brings the detail closer.
The Gummi team found navigating large hyperlinked documents faster too, and because the zoom was so natural, selecting objects didn't require the precision that most of us take for granted. Clever use of opacity - an eye candy gimmick on today's GUIs - also yielded some interesting results. Gummi allowed users to blend different maps together, or leaf through discrete but overlaid web pages. (Think how long it takes you to switch between overlaid windows on a GUI desktop). The team reported that it took a little longer for testers to grasp - er, literally - the value of continuous gestures for these - but once discovered these paid dividends, particularly in for blending maps. Despite its obvious early appeal, however, Carsten doesn't envisage flexible interfaces being suitable for all purposes. The most likely scenario is in a specific hotel or public location.