The Design of Everyday Things

by Don Norman

  As we move away from many physical aids, such as printed books and magazines, paper notes, and calendars, much of what we use today as knowledge in the world will become invisible. Yes, it will all be available on display screens, but unless the screens always show this material, we will have added to the burden of memory in the head. We may not have to remember all the details of the information stored away for us, but we will have to remember that it is there, that it needs to be redisplayed at the appropriate time for use or for reminding.

  Memory in Multiple Heads, Multiple Devices

  If knowledge and structure in the world can combine with knowledge in the head to enhance memory performance, why not use the knowledge in multiple heads, or in multiple devices?

  Most of us have experienced the power of multiple minds in remembering things. You are with a group of friends trying to remember the name of a movie, or perhaps a restaurant, and failing. But others try to help. The conversation goes something like this:

  “That new place where they grill meat”

  “Oh, the Korean barbecue on Fifth Street?”

  “No, not Korean, South American, um,“

  “Oh, yeah, Brazilian, it’s what’s its name?”

  “Yes, that’s the one!”

  “Pampas something.”

  “Yes, Pampas Chewy. Um, Churry, um,”

  “Churrascaria. Pampas Churrascaria.”

  How many people are involved? It could be any number, but the point is that each adds their bit of knowledge, slowly constraining the choices, recalling something that no single one of them could have done alone. Daniel Wegner, a Harvard professor of psychology, has called this “transactive memory.”

  Of course, we often turn to technological aids to answer our questions, reaching for our smart devices to search our electronic resources and the Internet. When we expand from seeking aids from other people to seeking aids from our technologies, which Wegner labels as “cybermind,” the principle is basically the same. The cybermind doesn’t always produce the answer, but it can produce sufficient clues so that we can generate the answer. Even where the technology produces the answer, it is often buried in a list of potential answers, so we have to use our own knowledge— or the knowledge of our friends—to determine which of the potential items is the correct one.

  What happens when we rely too much upon external knowledge, be it knowledge in the world, knowledge of friends, or knowledge provided by our technology? On the one hand, there no such thing as “too much.” The more we learn to use these resources, the better our performance. External knowledge is a powerful tool for enhanced intelligence. On the other hand, external knowledge is often erroneous: witness the difficulties of trusting online sources and the controversies that arise over Wikipedia entries. It doesn’t matter where our knowledge comes from. What matters is the quality of the end result.

  In an earlier book, Things That Make Us Smart, I argued that it is this combination of technology and people that creates super-powerful beings. Technology does not make us smarter. People do not make technology smart. It is the combination of the two, the person plus the artifact, that is smart. Together, with our tools, we are a powerful combination. On the other hand, if we are suddenly without these external devices, then we don’t do very well. In many ways, we do become less smart.

  Take away their calculator, and many people cannot do arithmetic. Take away a navigation system, and people can no longer get around, even in their own cities. Take away a phone’s or computer’s address book, and people can no longer reach their friends (in my case, I can no longer remember my own phone number). Without a keyboard, I can’t write. Without a spelling corrector, I can’t spell.

  What does all of this mean? Is this bad or good? It is not a new phenomenon. Take away our gas supply and electrical service and we might starve. Take away our housing and clothes and we might freeze. We rely on commercial stores, transportation, and government services to provide us with the essentials for living. Is this bad?

  The partnership of technology and people makes us smarter, stronger, and better able to live in the modern world. We have become reliant on the technology and we can no longer function without it. The dependence is even stronger today than ever before, including mechanical, physical things such as housing, clothing, heating, food preparation and storage, and transportation. Now this range of dependencies is extended to information services as well: communication, news, entertainment, education, and social interaction. When things work, we are informed, comfortable, and effective. When things break, we may no longer be able to function. This dependence upon technology is very old, but every decade, the impact covers more and more activities.

  Natural Mapping

  Mapping, a topic from Chapter 1, provides a good example of the power of combining knowledge in the world with that in the head. Did you ever turn the wrong burner of a stove on or off? You would think that doing it correctly would be an easy task. A simple control turns the burner on, controls the temperature, and allows the burner to be turned off. In fact, the task appears to be so simple that when people do it wrong, which happens more frequently than you might have thought, they blame themselves: “How could I be so stupid as to do this simple task wrong?” they think to themselves. Well, it isn’t so simple, and it is not their fault: even as simple a device as the everyday kitchen stove is frequently badly designed, in a way that guarantees the errors.

  Most stoves have only four burners and four controls in one-to-one correspondence. Why is it so hard to remember four things? In principle, it should be easy to remember the relationship between the controls and the burners. In practice, however, it is almost impossible. Why? Because of the poor mappings between the controls and the burners. Look at Figure 3.2, which depicts four possible mappings between the four burners and controls. Figures 3.2A and B show how not to map one dimension onto two. Figures 3.2C and D show two ways of doing it properly: arrange the controls in two dimensions (C) or stagger the burners (D) so they can be ordered left to right.

  FIGURE 3.2.Mappings of Stove Controls with Burners. With the traditional arrangement of stove burners shown in Figures A and B, the burners are arranged in a rectangle and the controls in a linear line. Usually there is a partial natural mapping, with the left two controls operating the left burners and the right two controls operating the right burners. Even so, there are four possible mappings of controls to burners, all four of which are used on commercial stoves. The only way to know which control works which burner is to read the labels. But if the controls were also in a rectangle (Figure C) or the burners staggered (Figure D), no labels would be needed. Learning would be easy; errors would be reduced.

  To make matters worse, stove manufacturers cannot agree upon what the mapping should be. If all stoves used the same arrangement of controls, even if it is unnatural, everyone could learn it once and forever after get things right. As the legend of Figure 3.2 points out, even if the stove manufacturer is nice enough to ensure that each pair of controls operates the pair of burners on its side, there are still four possible mappings. All four are in common use. Some stoves arrange the controls in a vertical line, giving even more possible mappings. Every stove seems to be different. Even different stoves from the same manufacturer differ. No wonder people have trouble, leading their food to go uncooked, and in the worst cases, leading to fire.

  Natural mappings are those where the relationship between the controls and the object to be controlled (the burners, in this case) is obvious. Depending upon circumstances, natural mappings will employ spatial cues. Here are three levels of mapping, arranged in decreasing effectiveness as memory aids:

  •Best mapping: Controls are mounted directly on the item to be controlled.

  •Second-best mapping: Controls are as close as possible to the object to be controlled.

  •Third-best mapping: Controls are arranged in the same spatial configuration as the objects to be controlled.

  In the ideal and second
-best cases, the mappings are indeed clear and unambiguous.

  Want excellent examples of natural mapping? Consider gesture-controlled faucets, soap dispensers, and hand dryers. Put your hands under the faucet or soap dispenser and the water or soap appears. Wave your hand in front of the paper towel dispenser and out pops a new towel, or in the case of blower-controlled hand dryers, simply put your hands beneath or into the dryer and the drying air turns on. Mind you, although the mappings of these devices are appropriate, they do have problems. First, they often lack signifiers, hence they lack discoverability. The controls are often invisible, so we sometimes put our hands under faucets expecting to receive water, but wait in vain: these are mechanical faucets that require handle turning. Or the water turns on and then stops, so we wave our hands up and down, hoping to find the precise location where the water turns on. When I wave my hand in front of the towel dispenser but get no towel, I do not know whether this means the dispenser is broken or out of towels; or that I did the waving wrong, or in the wrong place; or that maybe this doesn’t work by gesture, but I must push, pull, or turn something. The lack of signifiers is a real drawback. These devices aren’t perfect, but at least they got the mapping right.

  In the case of stove controls, it is obviously not possible to put the controls directly on the burners. In most cases, it is also dangerous to put the controls adjacent to the burners, not only for fear of burning the person using the stove, but also because it would interfere with the placement of cooking utensils. Stove controls are usually situated on the side, back, or front panel of the stove, in which case they ought to be arranged in spatial harmony with the burners, as in Figures 3.2 C and D.

  With a good natural mapping, the relationship of the controls to the burner is completely contained in the world; the load on human memory is much reduced. With a bad mapping, however, a burden is placed upon memory, leading to more mental effort and a higher chance of error. Without a good mapping, people new to the stove cannot readily determine which burner goes with which control and even frequent users will still occasionally err.

  Why do stove designers insist on arranging the burners in a two-dimensional rectangular pattern, and the controls in a one-dimensional row? We have known for roughly a century just how bad such an arrangement is. Sometimes the stove comes with clever little diagrams to indicate which control works which burner. Sometimes there are labels. But the proper natural mapping requires no diagrams, no labels, and no instructions.

  The irony about stove design is that it isn’t hard to do right. Textbooks of ergonomics, human factors, psychology, and industrial engineering have been demonstrating both the problems and the solutions for over fifty years. Some stove manufacturers do use good designs. Oddly, sometimes the best and the worst designs are manufactured by the same companies and are illustrated side by side in their catalogs. Why do users still purchase stoves that cause so much trouble? Why not revolt and refuse to buy them unless the controls have an intelligent relationship to the burners?

  The problem of the stovetop may seem trivial, but similar mapping problems exist in many situations, including commercial and industrial settings, where selecting the wrong button, dial, or lever can lead to major economic impact or even fatalities.

  In industrial settings good mapping is of special importance, whether it is a remotely piloted airplane, a large building crane where the operator is at a distance from the objects being manipulated, or even in an automobile where the driver might wish to control temperature or windows while driving at high speeds or in crowded streets. In these cases, the best controls usually are spatial mappings of the controls to the items being controlled. We see this done properly in most automobiles where the driver can operate the windows through switches that are arranged in spatial correspondence to the windows.

  Usability is not often thought about during the purchasing process. Unless you actually test a number of units in a realistic environment, doing typical tasks, you are not likely to notice the ease or difficulty of use. If you just look at something, it appears straightforward enough, and the array of wonderful features seems to be a virtue. You may not realize that you won’t be able to figure out how to use those features. I urge you to test products before you buy them. Before purchasing a new stovetop, pretend you are cooking a meal. Do it right there in the store. Do not be afraid to make mistakes or ask stupid questions. Remember, any problems you have are probably the design’s fault, not yours.

  A major obstacle is that often the purchaser is not the user. Appliances may be in a home when people move in. In the office, the purchasing department orders equipment based upon such factors as price, relationships with the supplier, and perhaps reliability: usability is seldom considered. Finally, even when the purchaser is the end user, it is sometimes necessary to trade off one desirable feature for an undesirable one. In the case of my family’s stove, we did not like the arrangement of controls, but we bought the stove anyway: we traded off the layout of the burner controls for another design feature that was more important to us and available only from one manufacturer. But why should we have to make a tradeoff? It wouldn’t be hard for all stove manufacturers to use natural mappings, or at the least, to standardize their mappings.

  Culture and Design: Natural Mappings Can Vary with Culture

  I was in Asia, giving a talk. My computer was connected to a projector and I was given a remote controller for advancing through the illustrations for my talk. This one had two buttons, one above the other. The title was already displayed on the screen, so when I started, all I had to do was to advance to the first photograph in my presentation, but when I pushed the upper button, to my amazement I went backward through my illustrations, not forward.

  “How could this happen?” I wondered. To me, top means forward; bottom, backward. The mapping is clear and obvious. If the buttons had been side by side, then the control would have been ambiguous: which comes first, right or left? This controller appeared to use an appropriate mapping of top and bottom. Why was it working backward? Was this yet another example of poor design?

  I decided to ask the audience. I showed them the controller and asked: “To get to my next picture, which button should I push, the top or the bottom?” To my great surprise, the audience was split in their responses. Many thought that it should be the top button, just as I had thought. But a large number thought it should be the bottom.

  What’s the correct answer? I decided to ask this question to my audiences around the world. I discovered that they, too, were split in their opinions: some people firmly believe that it is the top button and some, just as firmly, believe it is the bottom button. Everyone is surprised to learn that someone else might think differently.

  I was puzzled until I realized that this was a point-of-view problem, very similar to the way different cultures view time. In some cultures, time is represented mentally as if it were a road stretching out ahead of the person. As a person moves through time, the person moves forward along the time line. Other cultures use the same representation, except now it is the person who is fixed and it is time that moves: an event in the future moves toward the person.

  This is precisely what was happening with the controller. Yes, the top button does cause something to move forward, but the question is, what is moving? Some people thought that the person would move through the images, other people thought the images would move. People who thought that they moved through the images wanted the top button to indicate the next one. People who thought it was the illustrations that moved would get to the next image by pushing the bottom button, causing the images to move toward them.

  Some cultures represent the time line vertically: up for the future, down for the past. Other cultures have rather different views. For example, does the future lie ahead or behind? To most of us, the question makes no sense: of course, the future lies ahead—the past is behind us. We speak this way, discussing the “arrival” of the future; we are pleased that many unfortunate events
of the past have been “left behind.”

  But why couldn’t the past be in front of us and the future behind? Does that sound strange? Why? We can see what is in front of us, but not what is behind, just as we can remember what happened in the past, but we can’t remember the future. Not only that, but we can remember recent events much more clearly than long-past events, captured neatly by the visual metaphor in which the past lines up before us, the most recent events being the closest so that they are clearly perceived (remembered), with long-past events far in the distance, remembered and perceived with difficulty. Still sound weird? This is how the South American Indian group, the Aymara, represent time. When they speak of the future, they use the phrase back days and often gesture behind them. Think about it: it is a perfectly logical way to view the world.

  If time is displayed along a horizontal line, does it go from left to right or right to left? Either answer is correct because the choice is arbitrary, just as the choice of whether text should be strung along the page from left to right or right to left is arbitrary. The choice of text direction also corresponds to people’s preference for time direction. People whose native language is Arabic or Hebrew prefer time to flow from right to left (the future being toward the left), whereas those who use a left-to-right writing system have time flowing in the same direction, so the future is to the right.