The Design of Everyday Things

Don Norman starts with doors. Almost everyone has walked up to a glass door, pushed it, found that it needed to be pulled, and felt faintly foolish. Norman’s argument, which he has been making since the first edition of this book in 1988, is that the foolishness is in the wrong place. When ordinary, competent people keep making the same mistake with an object, the problem is not the people. It is the object. Doors that confuse their users are now widely known, in his honour, as Norman doors, and the fact that the term caught on says something about how familiar the experience is and how rarely anyone had named it.

The book first appeared as The Psychology of Everyday Things, a title Norman liked because its initials spelled POET, and which his publisher changed for the paperback because booksellers kept shelving it under psychology. A revised and expanded edition came out in 2013, updated for a world of touch screens and software. Norman was a cognitive scientist before he became known as a designer, and the book’s great strength is that it explains design failures in terms of how human perception, memory and attention actually work. It is not a style guide. It is an account of what goes on in someone’s head when they try to use a thing, and of what the thing can do to help.

The central concept, and the most misused, is the affordance. Norman borrowed the word from the psychologist James J. Gibson, who used it for the possibilities for action that an environment offers to an animal: a flat surface affords standing, a gap affords passing through. In design, an affordance is a relationship between an object and a person: a chair affords sitting, a button affords pressing, a flat metal plate on a door affords pushing. Over the years Norman noticed designers saying “affordance” when they really meant something else, the visible cue that tells a person where and how to act, and in the 2013 edition he introduced a separate term for that: the signifier. A handle signals pulling; a label says “push”; an underlined word on a screen signals a link. Many usability problems turn out to be signifier problems. The action is possible, but nothing shows that it is.

Around these two ideas Norman builds a small, practical vocabulary. Mapping is the relationship between controls and what they control, and his favourite example is the kitchen stove. When four burners are arranged in a square and their four knobs sit in a row, users have to guess or read tiny labels; when the knobs are laid out in the same pattern as the burners, no labels are needed. Feedback is the information that tells you an action has had an effect, quickly and clearly; a lift button that does not light up gets pressed again and again. Constraints limit the actions that are possible, physically, culturally or logically, so that the wrong action is hard to take: a plug that only fits one way, a form that will not submit until the required fields are filled. Underneath all of these is the conceptual model, the simplified picture a person builds of how a system works. A good design communicates a model that is accurate enough to guide action, even if it is not technically complete.

Norman also offers a way of locating where things go wrong. Using anything means crossing two gaps. The gulf of execution lies between what a person wants to do and figuring out how to do it with the controls in front of them. The gulf of evaluation lies between what the system does and working out whether it did what was wanted. Most frustrations sit in one of these gulfs, and design closes them from both sides: clear signifiers and natural mappings help with execution, and clear, immediate feedback helps with evaluation. I find this one of the most useful ideas in the book, because it turns a vague feeling of annoyance into a precise question. Do people not know what to do, or do they not know what happened?

The chapter on human error is the one that has stayed with me most. Norman distinguishes slips, where a person intends the right thing but carries out the wrong action, such as pouring orange juice into a coffee cup out of habit, from mistakes, where the goal or plan itself is wrong, often because of a misleading conceptual model. Both are normal features of human cognition, and both are made more likely by poor design. Blaming the person, the usual response in accident reports, rarely helps, because the next person will make the same error in the same situation. What helps is changing the situation: making errors harder to commit, easier to notice and easier to undo. Norman draws examples from aviation, nuclear power and medicine, where the consequences are serious, and the argument carries over directly to the screens most of us use every day. An undo button is a moral position as much as a feature.

The book is not without flaws. Some of the original examples, such as the controls of 1980s telephones and video recorders, now read as period pieces, though the revised edition replaces many of them. Norman can be repetitive, and occasionally his confidence that good design is simply a matter of applying principles underplays the real trade-offs designers face between cost, appearance, safety and ease of use. But these are minor complaints about a book whose main argument has aged remarkably well.

Since reading it, when something on a screen or in a building confuses me, I ask which of Norman’s principles is missing. Usually it is a signifier or feedback: something that should have told me what to do, or what had just happened. The book changed how I build interfaces, including this site, and more broadly how I think about blame. If people keep getting something wrong, the first suspect should be the design, not the people.