Recent (?) fascination with touch-controlled interfaces is perhaps good for their development, but in my opinion they are not necessarily the future of device manipulation.
One of the big mixed blessings is that you have to rely on the visual feedback to operate such an interface. While it is perhaps a tad faster to manipulate directly items that you’d always want to look at—like photos—or that wide-sweeping gestures are faster than looking for “next” or “previous” buttons, it is not necessarily so with interfaces that rely on mixed tactile/visual, and sometimes even auditory feedback.
An excellent example is a keyboard. A keyboard gives you at least three kinds of input: tactile (feeling of pressing the key and its shape), auditory (click of pressing the key), and finally visual – letters appear (or not 🙂 ) on the screen. Many people do not appreciate the first two, mostly because they were not trained in typing without looking at the keyboard to find each letter that they type, or to use all their fingers while typing. Personally I believe that classes of quick typing should be obligatory in primary or high schools and would be more useful in daily life. For example, when visiting a doctor, I often find that he takes more time to use his computer to type the diagnosis with two fingers than to actually examine his patient. What a terrible waste of time.
Anyway, the reason that mixed input is important is about efficiency. Once you learn to stop looking at the keyboard while you type, you reach a new level of efficiency. You start relying on your tactile and auditory input to feel and hear if you have pressed your key, and to an extent to know which key you pressed, only using visual feedback for confirmation, not estimation. For those who wonder why there are small embossed dashes on the F and J keys – they are places where your index fingers will be when you use proper typing technique.
Touch screen does not give you this advantage. You use visual cues to find the proper key, robbing yourself of the input by covering the key you want to press at the same time, and then for verification. You use a single channel to process the information. It is slower not only because tactile information reaches your brain and is processed 10 times faster, but also because you use serial processing instead of parallel one. While typing on a classical keyboard I know I have pressed a wrong or a good key even before I get the confirmation on the screen. Therefore it is much easier for me to switch from typing to correcting mode (and yes, there is a noticeable switch going on), than I would when I am typing on a touch-screen keyboard. My impression also is that the responsiveness and robustness of touch screen interfaces is still not at the level of keyboards, but I might be wrong, since this field evolves very quickly.
Another example where tactile input is vital, are the devices which one would be able to operate without having to look at them. One that comes to my mind is an mp3 player. Usually this device sits in my pocket or in a place that I do not have an easy visual access to, and for a good reason. Therefore if I want to increase the volume, lock the controls or change/rewind the track, I would prefer not to need to put the device in the center of my visual attention. Running, cycling, driving — these are the activities that do not lend themselves well to visual distractions. Admittedly, using any device while driving will lessen one’s concentration and might result in an accident, but this is precisely why most of the car interiors are built in such a way that you can rely on tactile input to turn on radio, heating, conditioner and all else.
Therefore, it makes little sense to design an mp3 player with touch screen input. When the buttons are present, you can learn their layout, and operate the device without the need to look at it. You will get immediate auditory input — increase in volume, next track will start playing etc. And you can easily lock/unlock controls, which perhaps is the biggest advantage of all.
There is also another issue. While using touch-screen to manipulate photos you often cover the part that you’re interested in manipulating, therefore robbing yourself of a visual feedback that the touch-screen is supposed to give you. This is not necessarily an optimal way to work. I would agree that it is the way that we manually paint or write, but it only shows the limitation of our tools (limbs). Personally, when faced with a choice of a touch-screen tablet, and a standard screen with a cursor, and a traditional tablet, I prefer the latter, simply because my field of view is wider. Motion estimation is similar in both cases, even if the second way takes more time to learn, and to get used to, like learning to use any tool or device.
All these examples show that if touch-screen interfaces want to become more useful, they will have to evolve additional feedback mechanisms. As of now, there are too many applications where they are detrimental to efficiency, and when we consider them setting their “coolness” factor aside, their application is still limited in scope.