Are Cars Too Safe? Are User Manuals Necessary?

Design guru Don Norman argues for making autos less automated and for infusing gadgets with natural, easy-to-interpret feedback signals

In his new book, The Design of Future Things (Basic Books), Don Norman isn't afraid to call himself out on statements he made in his earlier, wildly popular publications such as The Design of Everyday Things. The co-founder of corporate design consultancy Nielsen Norman Group and the former vice-president of Apple (AAPL) now says that he has changed his mind on several design strategies he has advocated over the years.

For one, he now says that consumers should try conforming to technology, rather than for engineers and designers to focus on adapting existing technologies for use. Why? Humans are actually more flexible than machines. He also has changed his tune on whether people should blame technology or themselves when a device doesn't work. In the 1990s, he argued that consumers were right to be angry at a device that didn't function with ease. Today, Norman writes, he believes that if people admit fault when using a machine, they might take the time to learn how to use it correctly.

The focus of his new book, however, is not on how he wishes to update his philosophies of design and innovation. Instead, it centers on so-called "smart," or automated, gadgets and products. Increasingly being produced and marketed, these range from talking refrigerators that scold you for not keeping to a diet to cars that are comfortable and easy to drive to the point of distracting drivers from the dangers of the road.

The Northwestern University design professor spoke with BusinessWeek Innovation Dept. Editor Reena Jana about the perils of automation and subtle but effective strategies for improving product design, such as offering sounds or visual signals that are more pleasant and instructive than electronic blips and bleeps. Below are edited excerpts from their conversation.

The book opens with two scenarios: one, driving with your spouse, who is frightened by the sharp turns on a winding road; the other, driving with a car that expresses a sense of "fear" by tightening seat belts automatically. The second seems like it might be more effective in curbing reckless driver behavior. Did you have a bad driving experience that prompted you to focus on automation in cars and other everyday machines?

For a long time, I've been interested in automation—in nuclear power plants, in the aviation industry. I've been an advisor to auto companies. The real impetus for the book was a trip I made to a Japanese carmaker. We were talking about automated lane-keeping. One executive told me how relaxed it made him feel. But a problem with automation in cars is that we can forget that driving is dangerous. How can we ask car drivers to be alert when it seems like not much is happening when they're in an automated car?

So you're saying that driving should feel dangerous, rather than safe—and that paradoxically that would make driving safer, because drivers will be more aware? And this is a design issue?

Having your car give you feedback is a good thing, yes. But only if this feels natural and nonobtrusive—and in no way like nagging or like a backseat driver.

Some design solutions could be to have seats vibrate when a car is starting to go off the road. The car doesn't say, "I'm scolding you." There could just be a subconscious feeling of danger. Something subtle. What I think is most helpful is to provide natural feedback and signals, not blips and bleeps. We need rich signals that people can interpret automatically. Wait a second…[sound of a door closing]. On the phone, you could tell what that was, right?

That's a door shutting….

Yes. Cars should do something similar—provide natural noises to alert people about what's happening. If it's raining outside, why not amplify the sound hitting the roof, rather than shut out the noise?

One major problem with the design of cars today is that you can be driving at 100 mph—which is quite dangerous—and the experience is comfy, smooth, and accompanied by nice music on the stereo system. Of course it's impractical to design a car so that driving it feels dangerous and shaky. But why not put passengers in the warm, smooth, comfy situation but have natural signals that give cues to the driver in terms of being alert?

Many designers say that one of their goals is to design devices and machines that don't require a manual, which pushes them to design products that are intuitive to use. Is this what you advocate, too?

Their heart is in the right place. Even a pencil could use a manual, though—how many years does it take to learn how to use it? It's more natural to use one as a weapon! The term "intuitive" is badly understood…. I don't use that term. Especially with complex things that take weeks of training [to learn how to use properly]—cars for instance. But even telephones require some type of training. What I argue for is understandable design. This would be something that would require telling me how to use it the first time. When you explain it, I would say, "I see…" and you'd never have to explain it again. Things designed with simplicity should be learned to use simply.

Of course, the rule of thumb for designers is that people don't read manuals. They want to go right to the device. With this in mind, why not make a device self-explanatory? Make it so that it gives tutorials and enough feedback so users understand what's going on—if you make a mistake, for example. Yes, it might take longer to learn how to use something with feedback, compared to using a manual. But people would enjoy the training more. They'd feel like they were accomplishing more from the beginning. I call it "just-in-time learning." It can be better to learn and fail. A device could be designed to offer pleasant and effective feedback just when you need it.

You end your book by calling for a "science of design." Does this mean a more formalized design process that can be replicated?

Yes, there is a need for a more formal basis for design. Designers need to learn how to collect evidence, how to transform, and how to evaluate. They can use formal tools, such as simulations and mathematical sciences. They can learn from engineering science about the services and products that fit the needs of human beings. Design is too complex to leave to art.

I serve on advisory boards at design schools [such as the Segal Design Institute]. We're training designers about ethnography and [they are] learning to use powerful computer tools. What's important is to teach designers how to make their skills translate into things that can be tested and manufactured. One way is to use iterative design [building upon successive iterations of an idea]. It's a powerful, scientific method.

Listen to a podcast with Don Norman about lessons from design failures (the Apple Newton) and design successes (the Roomba robotic vacuum cleaner).

Before it's here, it's on the Bloomberg Terminal.