This article was originally published in the May 2009 issue (Vol 14, No. 2)

About the Author

Ben Shneiderman www.cs.umd.edu/~ben) is a Professor in the Department of Computer Science, Founding Director (1983-2000) of the Human-Computer Interaction Laboratory, and Member of the Institute for Advanced Computer Studies at the University of Maryland at College Park. He works on information visualization for networks and electronic health records.

STC UUX Community Newsletter
Usability Interface

Lively Controversies over Usability Design

By Ben Shneiderman

The following is an excerpt from the 5th edition of Designing the User Interface reprinted with author Ben Shneiderman's permission.

Many of the rapid advances in user interfaces have been shaped by the contrasting visions of researchers and designers, which have sometimes broken out into heated controversies. There have been continuing debates over which directions would be most fruitful. In each case, there is room for victory claims by all parties and reasons to believe that with more research funding the future will favor their positions. Informed discussion can lead to agreement, or at least ways to balance the dangers with greater protection. Issues at the center of the debates include:

• Machine automation versus user control. This fundamental issue remains a lively source of controversy, spawning many sub-debates about issues such as the degree of automation in cockpits, the utility of automatic indentation in word processors, and the dangers of autonomous agents in financial markets. While designers are often proud to enhance automation, users-some more than others-often want to be in control. Users' desire for mastery and a sense of accomplishment can be undermined by an overly enthusiastic interface that does more than they want. Making automation understandable, predictable, and controllable helps in many situations, especially if designers have built-in highlighting and other informative feedback to make the machine's state clear to users. Some examples may help clarify why users may desire control. Doctors do not want machines that make medical diagnoses; rather, they want machines that enable them to make more accurate, reliable diagnoses, to obtain relevant references to scientific papers or clinical trials, to gather consultative support rapidly, and to record that support accurately. Similarly, air-traffic or manufacturing controllers do not want machines that automatically do their jobs; rather they want machines that increase their productivity, reduce their error rates, and enable them to handle special cases or emergencies effectively. The argument for user control is tied to the belief that an increase in personal responsibility will result in improved service. Advocates of increased automation argue that in some complex fast-moving situations (such as NASA shuttle launchings), only a machine can make adequately fast and accurate decisions. In such cases, careful design and thorough testing are much needed, but launch failures and the long history of computer glitches should be cautionary tales for those who believe that automation can be made to be flawless.
• Speech recognition versus visual interaction. Early dreamers believed that speech was the "natural" way for humans to interact and therefore that speech recognition would be the "natural" way for users to operate computers. While speech-recognition technology has matured, it has proven to be only modestly effective in interfaces. Dictation and limited phone-based systems have shown steady improvements, but the huge success story has been visual interaction. Computers are not like people; they have large displays that can rapidly offer visual overviews, present large forms to fill in, and offer multiple choices in menus. Since the cognitive load of speaking is high, it makes good design sense to instead allow users to point and click (activities the human brain can process in parallel with planning). In addition to visual displays, well-designed interfaces allow operators of digital cameras, game computers, and fighter planes to rapidly slide switches, press buttons, and move joysticks.
• Natural-language interaction versus direct manipulation. The early fantasy of typing commands or speaking to computers in natural language to get answers to questions has thrived in Hollywood but languished on Wall Street and Main Street. The idea remains alive in scenarios of typing natural-language questions to web-based agents or speaking simple commands to household appliances, but the commercial markets have grown more rapidly for direct manipulation and graphical user interfaces. Natural-language interaction, whether by typing or talking, has usually proven to be more cumbersome and slower than pointing, dragging, and clicking on graphical representations. Devotees of natural-language interaction still believe that with further research improvements their approach could become attractive, but direct manipulation strategies also continue to improve.
• Anthropomorphic partners versus human operation. The metaphors, images, and names chosen for user interfaces play a key role in the designers' and users' perceptions. It is not surprising that many user interface designers still mimic human or animal forms: Our first attempts at flight imitated birds, and the first designs for microphones followed the shape of the human ear. Such primitive visions may be useful starting points, but success comes most rapidly to people who move beyond these simple concepts. Except for purposes of amusement or for crash test dummies, the goal is rarely to accurately mimic the human form, but rather is to provide effective service to the users in accomplishing their tasks. Lewis Mumford, in his classic book Technics and Civilization (1934), characterized the problem of "dissociation of the animate and the mechanical" as the "obstacle of animism." He described Leonardo da Vinci's attempt to reproduce the motion of birds' wings, then Ader's batlike airplane (as late as 1897), and Branca's steam engine in the form of a human head and torso. Mumford wrote: "The most ineffective kind of machine is the realistic mechanical imitation of a man or another animal . . . for thousands of years animism has stood in the way of . . . development." Choosing human or animal forms as the inspiration for some projects is understandable, but significant advances will come more quickly if designers recognize the goals that serve human needs and the inherent attributes of the technology that is employed. Hand-held calculators do not follow human forms, but they serve effectively for doing arithmetic. Designers of championship chess-playing programs no longer imitate human strategies, but use hardware accelerators to explore billions of alternatives. Vision-systems researchers have realized the advantages of radar or sonar range finders and retreated from using human-like stereo depth-perception cues. Mature technologies such as industrial robots, clothes-washing machines, or robotic vacuum cleaners are not based on anthropomorphic designs. Still, in recent years, the belief in human-inspired robot design has been carried forward by a community of researchers who have shown some successes in helping autistic children, providing training, and offering robotic assistance for older adults.
• Adaptive versus adaptable interfaces. Designers who believe in their capacity to model and thereby anticipate user needs have proposed adaptive interfaces in which the layout and content change based on past user performance. Their goal of helping users by offering only relevant interface controls and content is admirable, but there are two problems: (1) users are not always predictable, so the changes based on past performance may not be helpful, and (2) changing interfaces can be surprising and disruptive to users who become familiar with a stable choice set. A successful compromise is to keep a stable display and then add one element-for example, a toolbar-that offers varying choices. Another useful adaptation might be in display areas that are already changing, such as a newspaper web site that has a box filled with topics predicted to be of interest to a specific user.
• Media richness versus lean design. Some communications theorists have argued that users will prefer and perform more effectively with richer media. They believe that videoconferencing would win over telephone conferencing, and that phone conversations are inherently more effective than typewritten messages. There are times when these beliefs are valid, but the surprising successes are often with leaner designs. Videoconferencing has the extra burden of requiring participants to give more attention to how they look and to show their interest in other speakers. By contrast, phone conferences allow participants to check e-mail or do something else during less interesting moments in the conversation. Similarly, text messaging and Twitter have become huge success stories because the messages exchanged are lightweight, take less time to read, are easily searched, saved, and re-sent. Both sides in this controversy have good claims for success, but over-generalizing leads to flawed predictions. While users often appreciate rich media with high-resolution video and high-quality audio, the high payoffs in rapid usage and low cognitive load for lean designs are also strong.
• 3D versus 2D interfaces. The compelling success of Hollywood filmmakers in telling stories and showing the world through moving images is apparent. The three-dimensional video games from Electronic Arts®, Sony, Microsoft, Nintendo and others, as well as the animated films from Pixar®, Industrial Light & Magic®, or Disney®, are stunning accomplishments for the graphics technology community. They offer users and viewers satisfying experiences that would be difficult to achieve in two-dimensional flat representations. However, in showing information, 2D strategies are almost always more effective: Users tend to initially favor 3D, but with regular use, well-designed 2D interfaces are perceived as more effective and preferred. Despite the attraction of 3D, immersive 3D interfaces and even 3D glasses to give stereo effects have not yet proven to be widely effective or popular. Here again, the controversy has led to refined understanding that higher dimensions and more immersive environments are not always better. Lower cognitive load, simplified navigation, less occlusion, and powerful actions are attractive goals.
• Data gathering versus privacy. Technology advances have given corporations and governments the capacity to collect vast amounts of data about individuals to advance their commercial or security goals. However well intended the goals are, the loss of individual privacy because of rapid search capabilities is a dramatic change that is disturbing to many people. While radio-frequency identification (RFID) tags can speed drivers through toll booths or commuters through train stations, the detailed tracking of personal behaviors undermines traditional expectations. Credit databases facilitate loans, but they are a centralized collection of personal information that could be used by criminals or oppressive political groups. Terrorist-detection schemes may promote security, but data-mining strategies have been challenged as ineffective and potentially invasive of personal privacy, while other approaches might simultaneously increase security as well as privacy. Even social-networking and user-generated- content web sites raise concerns as individuals put personal information and photos in public spaces where they could be maliciously misused.

Controversies are an indicator of lively interest and emergent technologies. New controversies are arising about the benefits of ambient displays, strategies for motivating participation in social media participation, technologies to protect privacy, and much more. If controversies signal strength, our discipline is thriving.