Powerless Computing
nicholas.negroponte
wired 6.02

Until last year, my biggest PC power problem was the inconvenience of carrying around four to six spare batteries. At least, I told myself, it's good exercise.

Two very different circumstances made me rethink powering computers. The first was a story in the February 28, 1997, issue of The Wall Street Journal. PCs, the paper reported, are mostly used like "potted plants," yet according to the Sierra Club, this wasted resource can account for 200 pounds of carbon-dioxide pollution every year - about 2 percent of what's emitted by a car that is "actually doing something." Turns out the story was just plain wrong. A desktop computer running continuously requires less than half a percent of the energy used to power a car (ditto its carbon production). And a laptop can reduce energy consumption to less than 10 percent of what's used by a typical microcomputer.

My second encounter came in July, at a gathering hosted by the Media Lab and the 2B1 Foundation. Participants from 45 developing nations spent six days sharing ideas and experiences about introducing, against all odds, computers to Third World education. At first glance, the odds seemed stacked against them three to one: the high cost of computers; the low availability of connectivity (affordable or otherwise); and the arrested development of educational theory, practice, and politics.

Another challenge, however, proved even more basic: power. In the poorest countries, some schools and most homes don't have any. In fact, more than one-third of the world's population is without electricity. One of the 2B1 participants, Peter Patrao from India, offered a simple solution: bicycles. Vigorously riding a bike generates about 100 watts. The image conjured by Patrao's classroom is certainly a cute one - think of half the class pedaling while the rest work on PCs, redefining, among other things, "recess."

Additional solutions to powering PCs ranged from car batteries to more imaginative ways of harnessing the wind and the sun. Then, in August, inventor Trevor Baylis, following his work on windup radios, reported success with a windup computer - clear progress toward curbing the PC's high powered appetite.

Power diet
A laptop's power supply gets eaten up by the display, the disk drive, and the circuitry, in that order. The display takes the biggest bite, typically 25-40 percent (and rising, as processors go to lower voltages). For a variety of technical reasons, backlit displays have so far provided the best contrast ratio and highest brightness. The power problem is that most of the light is lost in transmission: typically less than 10 percent gets through the flat panel. The rest is dissipated as heat. Still, an LCD uses five to ten times less power than a CRT.

A reflective display, by contrast, uses almost no power, taking most of what it needs from ambient light. This is why most calculators and all wristwatches require only tiny power supplies. So far, nobody has achieved an active-display medium that can reflect light with sufficient contrast. Actually, one reflect-ive display does a pretty good job - paper. In fact, "digital ink" has made significant progress in labs. (See Wired 5.05, page 162.)

Considerable power is also consumed by a disk drive, which is why drives typically spin down then start up as needed. As it happens, there are all sorts of other reasons to get rid of moving parts, a direction the industry is already pursuing.

The rest of a laptop's power consumption comes from circuitry, which can be made very power-efficient with modest trade-offs in performance.

With the exception of the display, then, industry trends do not necessarily fly in the face of low-power computing. In fact, Intel and Toshiba have massive programs under way in so-called flash memory, which uses no power to hold onto data, a little to read it, and a little more to write it.

In short, making a very power-efficient computer - one that uses only a bit more than your wristwatch - is not quite as pie in the sky as you imagine. Some of the issue is just fire in the belly.

Nabisco and IBM
What the cookie company and the computer company have in common, other than a great CEO, is an interest in calories. Could this possibly be translated into a PC powered with chocolate chips? Said differently, can the energy in your body be used to power a computer?

Many wristwatches, of course, do just that. Take, for example, a recent line from Swatch - rather than rely on an old-fashioned mechanical spring, they use body motion to charge a battery, which then drives an electronic timepiece. These would have been perfect for notorious British publisher Bob Maxwell, who once told me that the last time he did any exercise was when he wore a watch that needed winding.

Thad Starner, an MIT PhD student, has studied human-powered computing in some detail and with considerable self-interest - he has worn his computer for more than five years. Though the human neck can generate considerable heat, Thad has concluded, locomotion is the best source of power. He estimates that 5-8 watts can be recovered from walking.

A great deal of body energy, in other words, is simply dissipated, like waves crashing onto a beach. Recovering just a bit of it could be quite important for "effectively" powerless computing.

Combustion?
At the turn of the century, steam engines provided about 80 percent of the total capacity for driving machinery. Today, most office equipment uses electricity, which in the US alone accounts for a $2.1 billion energy bill, not counting air-conditioning. That's wildly out of scale with the developing world, especially the poorest countries in, say, Africa where per capita power consumption is 5 percent of ours.

The answer may be combustion. People are making serious progress in putting a fuel-burning, microelectromechanical engine on a chip. Butane, for example, has very high energy density. With an onboard generator as a means for generating electricity, you might one day simply fill your laptop with gas if you're tired of pedaling.