Donald A. Norman s The Design of Everyday Things

Norman, Donald A. The Design of Everyday Things. New York, NY: Currency/Doubleday, 1990.

This is an excellent user interface design book, but it does not directly discuss software user interfaces. By "everyday things," Norman means very common objects such as doors, telephones, refrigerators, cars, VCRs, and faucets. He only occasionally discusses computers. While at first this book might seem irrelevant to software user interface design, in fact it is nothing but relevant. Nearly every key idea applies directly to software user interface design with surprisingly little effort. One could argue that since the time this book was written, the computer has become an everyday thing. With today's sub-$1000 computers and with the Internet seemingly everywhere, formerly "nontechnical" people are using computers in their daily routine, and software user interface design needs to reflect this fact. Even my two-year-old son Philippe is able to turn his computer on and off, insert CD-ROMs (using caddies!), and start and use programs all on his own.

You need to read this book for two reasons: first, to understand basic user interface design concepts, which are explained extraordinarily well in this book; and second, to understand how to make complex software work as simply as an "everyday thing" by eliminating unnecessary complexity. Whereas Howlett's book helps you understand how a Windows program should look, this book helps you understand how a Windows program can be made easy to learn and use. This book has incredible insight.

Originally titled The Psychology of Everyday Things—the title was changed to make the book more appealing to a broader audience—Norman's book could be accurately titled How Users Learn to Use Everyday Things. For software user interfaces, human psychology forms the vital link between what the user sees on the screen and how the user decides what to do (ultimately determining how the user learns to interact with the user interface). It is therefore valuable to understand this psychology—how users understand visual elements, explain them, remember them, form relationships between them, become confused, and make mistakes. Knowing this information is fundamental to good user interface design. Since the book focuses on simple, practical examples of everyday objects and avoids academic psychobabble, it is enjoyable reading for anyone.

Many design principles are examined throughout the book, the most important being visibility, affordance, natural mapping, constraints, conceptual models, and feedback. User interfaces that implement these principles well are easy for the user to learn, understand, and use. Visibility gives the user the ability to figure out how to use something just by looking at it. Affordance relates to the perceived and actual properties of an object that suggest how the object is to be used. Natural mapping creates a clear relationship between what the user wants to do and the mechanism for doing it. Constraints reduce the number of ways to perform a task and the amount of knowledge necessary to perform a task, making it easier to figure out. A good conceptual model is one in which the user's understanding of how something works corresponds to the way it actually works. This way the user can confidently predict the effect of his actions. Lastly, feedback indicates to the user that a task is being done and being done correctly.

Here's another interesting principle that Norman discusses: If a control needs a label, the design of the control has failed. While labels are normally fine in software user interfaces, clearly a control that makes sense without a label is better than a control that requires a label. Specifically, edit boxes, which usually require labels to make sense, are not as effective as more constrained controls, such as combo boxes and slider bars, which are easier to understand without labels than edits boxes are. In terms of everyday things, a door handle is an excellent example. If a door looks like it should be pushed, people are going to push it. Only after a door fails to open after being pushed a couple of times do users bother to read the label that instructs them to pull.

I had many revelations while reading this book; I'll share just one. Many products are advertised as not requiring programming. What exactly does this mean? This distinction is somewhat arbitrary since you could argue that just about everything is "programmed" in some way. Doors are effectively programmed to open and close. Washing machines are effectively programmed to handle differing load sizes, colors, and fabrics. Certainly, using any sort of computer software is a form of programming. However, a task can be said to be performed without programming if there is a simple, visible, and direct mechanism that maps from what the user wants to do to how the object does it. Each control on the object has a single obvious function and provides some sort of feedback that the task is being accomplished correctly. On the other hand, a task requires programming when there is no direct relationship between what the user wants to do and how the object does it. For example, a digital watch requires programming and an analog watch does not. While it is possible to design a digital watch that doesn't require programming, its controls would support only a few functions. The reason digital watches require programming is to provide many features with a small number of controls. In the end, if given a choice, most users prefer to perform their work without programming.

This is a must-have book.