As digital technologies go, the personal computer is perhaps the most frustrating technology ever. Even the vocabulary boggles the mind. Boot, RAM, DRAM, ROM, floppy disk, hard disk, megahertz, gigabyte. Why do we need to know any of these terms? Answer: We don't. We are told we need to know because we are driven by technology and technologists.

The computer should be thought of as infrastructure. It should be quiet, unobtrusive, but it is too visible, too demanding. Its complexities and frustrations are largely the result of an attempt to cram far too many functions into a single box. The personal computer tries to be all things to all people. It casts all the activities of a person onto the same bland, homogeneous structure of the computer: a display screen, a keyboard, and some sort of pointing device. This is a certain guarantee of trouble. Any single set of tools is a compromise when faced with a wide range of tasks.

Take another look at the Swiss Army knife with those umpteen blades. Sure, it is fun to look at. Sure, it is handy if you are off in the wilderness and it is the only tool you have. But of all the things it does, it does none of them particularly well. Yes, my Swiss Army knife has a screwdriver and scissors and corkscrew—it even has a knife blade—but when I am home, I much prefer to use a real screwdriver, real scissors, a real corkscrew, and even a real knife. Not only are the simpler devices superior, but they are easier to use. With the Swiss Army knife, I invariably pry up the wrong blade until I find the one I am seeking.

Now take another look at the PC. It does everything, serves all masters, works around the world. The result is that it comes with large instruction manuals, plus multiple layers of menus and screen icons and toolbars. Each item presents me with options that I neither understand nor care about. Today, the PC has become more complex than the old mainframe computer it was intended to replace.

Some of the PC's overbearing complexity stems from the ubiquitous graphical user interface. It was right for its time, but it is wrong for today. It has outgrown its usefulness. The basic interface design was developed back in the days when personal computer software was relatively simple. The essential design principle was to make everything visible. Instead of memorizing archaic commands, one could see the entire array of possible commands, file names, and directory names. But, today, machines have expanded in power thousands of times. And the graphical user interface solution doesn't scale. Making everything visible is great when you have only 20 things. When you have 20,000, it only adds to the confusion. Show everything at once, and the result is chaos. Don't show everything, and stuff gets lost.

Some will say that computers are only difficult for those of us in the older generation. Young children who grow up with computers, I am constantly told, have no problems. Nonsense. My children call on me for help. Sometimes when I had trouble with a computer, I called in the very people who designed them. I have watched the designers of these systems grow befuddled. "Gee," they would say, scratching their heads, "I've never seen it act this way before."

Not only were these the world experts on these computers (after all, they had designed them), but this was for the Macintosh by Apple, a company that took pride in having a computer that was simple, easy, and well-designed. This "old folks don't get it, but kids always do" argument is annoying. We got it. I get it. But we can't always make it work. It is time for a new solution.

The computer industry itself stands in the way. All industries have a problem of ensuring a continual revenue stream from their customers. In the computer industry, the answer is to convince users that whatever hardware and software they are now perfectly happy with is, in actuality, unsatisfactory. This is done through the introduction of new features. The strange thing is that this strategy appears to work. Every six months, the hardware side of the computer industry comes out with new models, each better and faster, with more capacity and lower cost than the previous model. Every year, the software industry comes out with new products and new versions of existing products, each with brilliant new features that are essential for the health, safety, and well-being of the planet.

But instead of one complex PC, what we need is a series of specialized devices, each dedicated to a single task, that work invisibly together. What we need are information appliances with computers embedded so seamlessly into the tools that we don't even know they are there. Design the tool so well that the tool becomes part of the task, feeling like a natural extension of the work, a natural extension of the person.

Such embedded computers are already at work in automobiles. They control the fuel injection, ignition timing, antilock braking, and other systems. They can be applied to the fuel gauge to improve upon the way most gauges work today. You don't really want to know how much fuel you have left (no, honest, you don't). What you really want to know is how far you can drive. The normal fuel gauge can't do this because it is a simple float that rides up and down on the surface of the fuel, allowing its level to be translated into how much fuel is left in the gas tank. To translate fuel level into how many more miles of driving is possible requires computation. The fuel level has to be converted to the amount of fuel. Then, an estimate of the efficiency has to be made: How much gasoline have you been using per mile recently? Multiply the efficiency by the amount of fuel and you have the predicted range. These computations require a computer; hence the moral of this story.

Computers need to be like that fuel gauge: invisible, automatic, and useful.

We have been told that "being digital" is a virtue. It isn't. People are analog, not digital; biological, not mechanical. It is time for human-centered technology, humane technology.