Design For The Digital Revolution As computers change the world, we need to make sure the new world works for humans.
By Bill Joy

(FORTUNE Magazine) – Over the next three decades computing power will grow astonishingly, by a factor of about a million. Computers will become so powerful, easy to use, and ubiquitous that they will change our lives utterly. In this era of amazing change we will face a huge challenge of design: how to humanize our digital devices, our homes and offices, and our public places; how we will make them serve our needs; and how we will make the digitally enhanced places beautiful.

Our emerging digital universe is still primitive. While we have incredible technology in hand and the arc of progress is easy to see, we do not yet have the design skills to make that universe a pleasant place. For example, if you don't know where you are or what time it is, it's hard to behave properly. Young children often have this unawareness, not knowing when to talk, to sit still, to be quiet. The devices that we carry today--our pagers, cell phones, and the like--annoy us like other people's misbehaving children, because they too are unaware of their context.

Digital networks and miniaturized electronics are already altering our perception of space and time. It is less and less often the case that we have to be anywhere in particular at any specific time--we can work, shop, and entertain ourselves in virtual and physical space 24 hours a day. Cellular phones are changing our notion of space by enabling us to call a person, not a location; if you are traveling, my call may easily come at a ridiculous time.

Until recently we used our computers primarily to edit files and work with information that resided on our desktops, in the machines' memory or disks or on CD-ROMs. With the emergence of the Internet, the computer's primary use is now as a communications tool, looking outward, not inward. The link to the network has become essential. The e-mail I get now more often than not invites or even requires me to look at other information on the network before I can respond. Disconnected from the network, my computer is nearly as useless as a cell phone in an area with no cellular service.

The ongoing miniaturization of electronics is expanding computing beyond the personal computer. It is now possible to put an entire computing system on a single chip and then embed this chip in many different kinds of devices, most of which don't look like a computer at all. These "appliances" will be easy to use, with shapes and knobs and buttons that suggest directly what they do and how they do it, and will have built-in Internet access so they may increase their power by working together.

We are digitizing all forms of media and even our commercial interactions, and unbottling information from its conventional containers. We no longer need CDs to listen to music, tapes or DVDs to view videos, newsprint to read the newspaper, or cars to go to a store to buy presents. This unbottling has big implications for the way we live our lives. George Carlin observed humorously that our houses are the places where we keep all our stuff, and that if we didn't have so much stuff we could just wander around. As digitization drives the weight of things to zero, we are becoming much more nomadic.

The strongest technical trend for the next decade is pervasiveness--which will bring about amazing increases in our ability to connect and communicate. We will have pervasive computing, where everything that has electricity will have software in it, and pervasive networking--everything will be hooked together.

We will have seamless worldwide wireless data-networking coverage, so that all kinds of wireless devices have access to digital information. The PC of the next decade will be a personal communicator. It will fit in your pocket, almost always be with you, have a persistent connection to the Internet, and resemble your cellular phone in its ease of use and usefulness.

We will have far more cellular devices than people--I fully expect my dog to have a cellular phone in her collar, so I can call her and tell her to come home. (There will probably be a cellular device in my cat's collar as well, but I'll need a GPS locator device on her, because I'll have to go get her; she'll never come just because I ask.) Many other objects will have wireless data connections so I can find them and control them remotely.

Just over the horizon are incredible wireless networking technologies, such as ultra-wide-band radio (UWB), which will give us radios that can send huge amounts of information using so little power that the batteries will last for many years. Huge advances will also be made in wired networking, especially in the speeds achievable with optical networks. And anything that is plugged in to AC power can have two-way data connectivity through power-line networking.

Another strong trend affecting design is that the network is becoming dumb and the end nodes smart. The internally simple Internet is replacing the internally complex public phone system. This is part of a general trend toward disintermediation: Physical retail spaces are getting replaced with virtual e-commerce places, eliminating the physical stores between the warehouses and the customer; college campuses for continuing education are being replaced with distance learning, eliminating the classrooms; buyers and sellers can find each other in online auctions, providing much more efficient markets. The encompassing trend is that most complex systems with centralized control are giving way to simple distributed systems.

Such changes are likely to continue for a long time, and the rate of change is likely to accelerate. Moore's law, which predicts a doubling of chip capabilities every 18 months or so, has held since the mid-'60s, and recent developments in so-called molecular electronics--arranging individual molecules in electronic circuits--suggest that the improvements are likely to continue for another 30 years. That would bring about another millionfold improvement. That big a shift cannot be understood quantitatively--it is so large that it will change our perceptions. Consider: A factor of a million shrinks a millennium to nine hours, a lifetime to 40 minutes, a year to 30 seconds. A jet plane is less than 1,000 times faster than walking, yet even that drastically reshapes our perception of the world.

We can anticipate both good and bad results from these advances.

Networking all our appliances and giving them smart software will enable them to report in when they have problems, so that the serviceman can show up to replace failing parts, often even before they break. A network of smart devices, properly designed, should be cheaper to own and maintain. And if we network most devices wirelessly, there will be fewer wires to break. We'll be able to arrange our systems to reconfigure themselves automatically when changes occur.

As the wireless network makes us more nomadic, we clearly will have the power to remove bothersome space and time constraints from our lives. Yet the same technologies that make possible wireless ubiquity and nomadism threaten our privacy. UWB radios are essentially undetectable, so a hidden wireless microphone incorporating this technology might also be undetectable. We are likely to find ourselves living in a world where every action will be watched the way the actions of celebrities are scrutinized today. The vulnerability and anxiety that we feel as our lives become electronic is already the stuff of Hollywood movies like The Net.

And there is a clear danger that if we put software and networking in many different devices without careful design, we may find ourselves adrift in a world where, no matter where we look, we see the practical equivalent of a thousand VCRs brightly flashing "12:00."

Moore's law provides us with the chance to do much better. We have now, or will soon have, such great computing power in any chosen device that we can start focusing our software efforts on the human factors in the design of our digital appliances. We can design the software and interfaces to our wants and needs.

The tendency will be for the new digital landscape to leave its inhabitants too exposed--cell phones ring in the theater, miniaturized Web cameras and microphones observe us clandestinely, data-mining reveals our habits and predilections. Will digital design save us by restoring some sense of privacy and anonymity? History suggests we can botch it badly. The last major man-made overlay on civilization's landscape was the automobile. Even though we love our cars, we did a horrible job with them--they pollute, they isolate, and they have changed the landscape in ways that make them both incredibly frustrating and absolutely essential. If we are to build a new digital overlay on our world, we must do better this time.

Most of my work is as a computer architect. For the past ten years I've focused on building software and hardware systems that will make it easy to model and implement the new kinds of devices and services that will populate the digital landscape.

Our understanding of the natural world was greatly aided by classifying the plants and animals that populate this planet into a hierarchy that captures their common features and ancestry. I believe that a similar approach, known as object-oriented programming, is useful in modeling the devices and services of the man-made digital world. Put aside the technical details--all you need to know is that by associating a chunk of software called an object with every device or service, we get a way to structure the digital world so that it connects with reality in useful and pleasing ways.

Looked at from this standpoint, for example, a speaker is a kind of object that picks up a stream of audio information and converts it into sound. You probably have such an object in your pocket--it's part of your cellular phone. You also have such objects in your car--they're part of the car's stereo system. Why can't the phone in my pocket detect and make use of the speakers in my car, so I can carry on a phone conversation while my hands are on the wheel? What's missing is the software that lets us represent what things do not as protocols, and not as applications running on a complex operating system, but as objects with behavior. By using software objects to choreograph dynamically the behavior of devices in my vicinity, I gain control--my personal communicator can find an appropriate microphone and speaker, and lets me make a call in my car without my having to fumble to pull a device out of my pocket. In the future I also expect to have microphones in my clothing (this will require washable chips, and long-lived batteries, but these are near-term or at worst intermediate-term technologies).

The design style I am advocating is one in which each device and service is simple and directly reflects its use. This is certainly not the style that we see in personal computers today. Sadly, they are jacks-of-all-trades but masters of none--they can do many things, but almost all of them with more difficulty and complexity than a more simple appropriate tool. They are like Swiss Army knives with 50 or 100 blades and tools, all too small and weak to do the job properly and conveniently.

To implement this simpler style we need other tools besides software objects. We need standardized data and forms for the network--the computer industry is working to create these standards with a technology called XML. We need software "agents"--processes that can move around the network and then act after having moved. Consider the difference when traveling in the physical world between carrying a printed set of directions and having a guide with local knowledge--the printed directions are brittle, and if I miss a turn they are no longer useful, because they have no ability to adapt. (You know the feeling: You get off at the wrong freeway exit, and then turn around, but there isn't an on ramp and then, oh well.) If, however, you give me an agent or guide, which is aware of the goal and can work toward the goal in context, then I won't get hopelessly lost simply because I missed a turn. Agents are a general solution to this problem and greatly increase the power of the network.

We will need more layers of abstraction to do great digital design. We need the devices and services to become aware of their location, both in space and time, especially if they are mobile. And once we give them this awareness of context, the natural next thing to want is to capture some of the user's preferences in the form of rules for action in certain situations. These rules are difficult to program in traditional computer languages. Fortunately, technologies from the artificial intelligence community--rule-based systems and the like--are quite well suited to writing situation-aware software.

We don't need to attempt to make individual devices "conscious." Instead, we can build up layers of behavior, so that the devices and services react in a much more direct way to their environment. For example, rather than implement a complex central management scheme for the digital devices in a car--cell phone, radio, GPS system--we can connect them to sensors that let them respond to one another's input and to the driver, and turn off when they detect that no one is around.

There are three emerging digital landscapes that need great design. The first is the virtual worlds that are emerging behind our screens. These appear today in nascent form as graphical interfaces to mixed text-and-graphic databases, at sites such as Yahoo, AOL, and These sites will evolve to include virtual world simulations, as the portals become virtual worlds. The kind of hardware needed to create new and immersive experiences will be widely available soon, in devices like the Sony PlayStation 2, and entertainment provided by such devices will be a primary use of the wide-band Internet.

The second digital landscape will emerge in a redesign of the places where we live and work--in new digital rooms and in the interiors of cars and planes. This redesign begins by designing individual objects to be part of these spaces, without trying to dictate in advance exactly how the people who use the spaces will ultimately put them to use; as engineers we call this style "bottom-up design." To return to the example of the car: I happen to be sitting in one as I write this. This car has a radio, designed in the usual style, including knobs, FM and AM tuners, amplifiers, speakers, and a display. The car also has a cellular hands-free kit with its own speaker and microphone. Why do these devices have separate speakers? Because the radio is integrated rather than modular: The individual functions of the radio's components are unavailable to other devices in the car.

In the new object-oriented design style, the radio would be disaggregated into simpler components, each of which becomes a device or service available over a network. This would let the phone use the speaker as a network service, since the speaker is no longer "hidden" inside an overly complex radio. Other devices that I bring with me into the car would work with the devices that are already there as naturally as if they had been built in.

This design style is fluid and flexible: If there are several devices that can do the same thing, I can survive a failure of one; the other devices and services can naturally switch to using ones that still work. And I don't have to anticipate all the possible uses when designing the individual devices.

What I want to do in such a space is to choreograph the behavior of the devices. As I get into the car, my cell phone or PDA can locate, through a wireless network, the resources that are available locally and then coordinate their behavior in response to my desires.

But it isn't enough just to stick devices in the places we inhabit; we also need to design the places. We want the devices to be part of the room itself. For example, the displays, speakers, microphones, and other objects and sensors need to know how they relate to each other, so information can be presented in appropriate ways and on appropriate displays.

Redesigning the world to include embedded functions particular to specific tasks is a trillion-dollar industry. The issues are large; Don Norman's books, such as The Design of Everyday Things, show how hard it is to design even simple appliances. The kind of design needed to have great digital places is even more difficult than architecture, since architectural design is just a part of this larger problem.

The third digital landscape we need to design is the way in which the new devices affect our public places, our cities, and our outdoor spaces. The last major technology we overlaid on these spaces--the automobile--was a relative disaster. As Jerry Mander points out in his cautionary book In the Absence of the Sacred: The Failure of Technology and the Survival of the Indian Nations, we might at least have flinched if we'd seen how this was going to come out, with our cities choked with cars, us forced to spend large amounts of time in them, and our communities in decline as a result.

Such unintended consequences should give us pause. We should try to do better as we put wired and wireless networking, sensors, and other devices in our cities. It's hard to avoid unintended consequences in making massive changes but especially hard to avoid them if we don't try.

A vision for this electronic overlay was a large part of David Gelernter's groundbreaking 1991 book, Mirror Worlds. It anticipated not only the Web but also how virtual worlds could be implemented. Gelernter envisioned the city as live, running a simulation of itself. Imagine that when you are in town, the town is aware of your presence, and of who your friends are and where they are. You may be driving along and be told that someone you haven't seen for a while is having coffee just over there (and there is a parking space out front, too!). Simple things like this may begin to restore the human scale to our now-so-large cities.

We face at least two challenges in designing new digital landscapes. The first is to make the landscapes human. A person brought up in a world without cars would be hard-pressed to agree that the landscape created by the automobile represents human values. Cars spread out things to distances appropriate to cars, and intrude on our space and our ability to appreciate the world at a scale appropriate to pedestrians. We tolerate them and the consequences of their intrusions largely because we are desensitized.

We need to design the new digital landscapes to be for us, instead of simply adapting to their values. They need to reflect our ways of perceiving and our emotions. This will not be easy to accomplish; it is not the by-default result.

The second challenge is to make the new landscapes beautiful. To do this, we will need great design more than ever. Christopher Alexander's ideas, introduced in his seminal work A Pattern Language, are important and hopeful. Pattern languages emerge from a study of the small elements that work in our built world, and the relationships between them. For example, rooms work better if they have light coming in from two sides, so there is glare-free natural light; spaces with lower ceilings feel more intimate; entering a comfortable building, we proceed through a sequence of spaces, from the most public to the most intimate. Such patterns aren't laws, yet designs that are "beautiful" tend to respect them.

Since writing A Pattern Language more than 20 years ago, Alexander has been working to understand what it is that makes things beautiful, to uncover a natural order that underlies beautiful things. I certainly do not believe that all beauty can be captured mathematically, but I do believe there are usually organizing principles shared by the things that we like and find beautiful. For example, beautiful things tend to have a strong center toward which we gravitate, whether it be a warm room at the center of a house, candles in the center of a table, or a central element in a beautiful painting. We can imagine organizing a new kind of design software around this principle and other generative design elements.

A century ago, the world was remade: We synchronized the clocks for the sake of railroad schedules; we discovered, through art and literature, new ways to perceive time; the automobile reshaped our notions of distance and adjacency.

Now our world is being reshaped again, as a result of computing and communications technology. It is for this new digital world, and for its new landscapes, that we need great digital design. These will be electric places with manifold possibilities for designers and explorers, places where we can express and experience all the beauty that we can imagine.