PART I
1
Within this context, the question arises as to whether the many proposed benefits of the new technologies will be available to ordinary citizens. Specifically, will these technologies have interfaces that are usable by the broad spectrum of people who may wish to use national information infrastructure (NII) resources? In cases where they will not, what new research is needed to make such interfaces available? This report examines these questions and their solutions as analyzed by the authoring steering committee.1
As a start for the study, the steering committee notes that good interface design can help the spread of technologies. For example, one can point to the movement of personal computers from offices to homes and the growth of electronic game systems (e.g., Nintendo and Sega) that run on special or general-purpose computers. In the last several years, we have also witnessed the explosive growth of the World Wide Web and its use by individuals, schools and universities, companies, governmental units, and nonprofit organizations. Another area of growth involves the 800- and 900-number telephone services and experiments in electronic banking, meter reading, and other specialized data services to the home, as well as the rise of "freenets" and other local public network services.
Yet problems with many interfaces have also been observed. Interfaces often frustrate or are of limited use to many users, restricting access and use.2 Problems begin with those who do use these technologies despite some apparent difficulties. They suffer from repetitive stress syndrome and from the effects of low input/output bandwidths, overly restrictive computational formats, information overload, and many other problems. Market research also shows that today's costs of owning a personal computer, in the home or office, are very high once the various support costs are factored in.3 Other problems relate to those who do not or cannot use the information infrastructure. Current interfaces are among a variety of factors that limit use today by those who have physical, sensory, cognitive, language, and learning difficulties and by those whose activities or environments impose constraints on what they can do and how. Despite an enormous number of smart people working to improve interfaces, this is an area characterized by tough problems, many of which are getting tougher as the user population and its demands grow. As Bruce Tognazzini observed at the August 1996 workshop, "While critical roadwork needs to be done in building the nation's information superhighway, we cannot afford any longer to ignore the cars. Our 1960s rattletrap hardware and 1970s rattletrap interfaces and software are not up to the task of every-citizen access to this nation's information infrastructure."
Based on its study of such problems, the steering committee recommends an aggressive research program, funded by government and private sources, that examines both the human performance side of interfaces and the interface technologies, current and potential. Certainly such funding has played a major role historically in breaking new ground in interface design, and even greater reasons exist now for its continuance. One need only look at the roots of current graphical interfaces, notably the Apple Computer Macintosh operating system, which built on the earlier Xerox PARC SmallTalk and Alto systems and yet earlier work at SRI and RAND.4 Another example is the Internet, which can be traced to federally funded network research and deployment (CSTB, 1994a, 1995). These developments show that major progress can be made in spreading the use of technologies. They also illustrate a fruitful combination of public and private investment in research and development. Box 1.1 outlines the influence of commercial conditions on interface availability.
A significant research opportunity (and challenge!) is presented
by the need to narrow the gap between the capabilities provided in
today's interface products and the capabilities that computer science and
interface design researchers believe are
possible.5 Research that can
contribute to advancing innovative concepts and that can promote better
understanding of what technology works well to make interfaces more
usable, useful, and accessible is the focus of this report.
DEFINITIONS, KEY CONCEPTS, AND PARAMETERS
The steering committee established definitions for the major terms that appear in the subtitle of this report: Toward Every-Citizen Interfaces to the Nation's Information Infrastructure. The term "every citizen" represents the steering committee's desire to consider the needs of people of all ages and physical abilities, races and ethnicities, education and ability levels, economic backgrounds, cognitive styles, and personal inclinations. It overcomes the homogenization suggested by a term used early in the project, "ordinary citizen," but also includes people who are ordinary in the sense that they are not distinguished by special needs, such as accommodation of disabilities. The term opens up consideration not only of conventional office and household equipment and services but also of systems in production manufacturing and service environments, systems embedded in equipment designed for purposes other than computing and communications, and public access systems such as public kiosks.
Associated with this definition are several principles:6
The steering committee defines the "interface" to a system as encompassing the various means by which people communicate and interact with the system to engage and guide what it does, the nature of the dialogue, and the means by which the system communicates its responses. The interface includes input/output hardware and software—the actuators, keyboard, speech recognition, and so forth—that people may employ for input and the visual, audio, and other representations that the system returns as its output. It also includes application characteristics such as information management and presentation, collaborative and other group interaction dimensions, and sharing of labor and responsibility between people and systems (e.g., by means of autonomous agents). The technical facilities (e.g., communications, computation) supporting people's interaction with systems are also part of the interface because their performance directly influences the experiences people have using a system—the interface in its broadest sense. For example, an important dimension of an interface is the amount of time delay encountered when using a particular system; this depends on the available communications bandwidth, processor speed, and memory capacity as well as the design and function of the input/output components. An interface may connect a person to a machine, an application program of some kind, other people, or a combination of such entities. The definition used in this report is more comprehensive than may be traditional; the steering committee's goal is to avoid the slippage that comes when the interface and application are narrowly defined and important issues are lost in an unfortunate gap that falls between them.
Because the NII and the population are inherently diverse and heterogeneous, it is unreasonable to expect to have a single every-citizen interface to the NII. The variety of users, applications, and vendors mandates a corresponding variety of interfaces. These will include present-day telephones and computers (including both portable and full-featured desktop or room systems), inexpensive network computers and various portable computing and communications systems, a variety of devices that can be worn on one's body or installed in a vehicle, and systems embedded in equipment and services with other primary purposes.7
The term "national information infrastructure" as used here means the collection of communications systems in the United States and the set of computers and information stores and services that may be accessible through them. It includes the telephone system, the radio and television networks, all of the libraries and computers in the country, and a long list of other communications and storage facilities and services (CSTB, 1996). Of course, elements of the information infrastructure have always existed, but the term "NII" was coined recently because of the possibility of tying together a great many (if not all) of these elements into an integrated network complex that will be accessible (with some limitations) to essentially everyone. The Internet is a significant part of the NII, but it is important not to equate it with the much richer and more complex NII as a whole. The NII concept implies that the paradigm for the coming century is one of networked machines, collaborative computing, ubiquitous and possibly continuous access, and group interactions over networks of all kinds.8
Of course, interface improvement is not sufficient for maximizing
the utility of the information infrastructure. Other factors outside the
scope of this report include the appeal of the content and activities made
available (do people want to use the information infrastructure more, or
are advocates projecting their own tastes?), regulatory and legal
conditions that affect the nature and pricing of communications products that
compose and depend on the infrastructure, free speech protections and
intellectual property controls that shape what information is available
through different components of the infrastructure, and political processes
based on value judgments about how technology can be leveraged to meet
societal objectives (themselves subjective) and decisions about the
allocation of public resources to technology relative to other uses. All of
these factors combine to shape who uses the information infrastructure
and how, when, where, why, and how well. They indicate why public
policy is an important element of context. In this context, the Americans
with Disabilities Act and the Telecommunications Reform Act are recent
examples of public policy interventions to promote more and easier use
of the information infrastructure (and other facilities) by every citizen;
the Telecommunications Reform Act, for example, contains language
requiring that product design incorporate features enabling access by
people with disabilities.9
ADDITIONAL COMMENTS
There is a vital interaction between the shape of the technology and the public and private objectives for its use, drawing implicitly on ideas associated with universal service in telecommunications or equitable access more broadly viewed. For example, the expression "every-citizen interface" led some contributors to this report to suggest framing research relative to objectives concerning how much of the population can undertake certain activities associated with exercising the rights and responsibilities of citizenship (e.g., sending and receiving e-mail, querying a government agency, or participating via the network in a multiweek organized educational experience). Others pointed to problems in framing such objectives—how much of the population constitutes critical mass or social equity? What activities are most essential to enable? How can their achievement be paid for if they are not likely to emerge from the marketplace?10 Contributors to this project noted, therefore, that given what people can and do do with today's information infrastructure (including activities that substitute for use of other technologies), prematurely promoting broader access and use might result in either underused resources or a diversion of personal and institutional resources from less to more expensive technologies currently used in somewhat similar ways. Overall, the difficulty of getting many things right, at the same time, suggests that serving every citizen will take time—and that research can help accelerate the progress.
Questions of timing and the incidence of costs and benefits are
familiar ones in the evolution and diffusion of new technologies. What
stands out in the context of the information infrastructure, however, is the
strong belief by many that the associated technologies, including their
interfaces, have public-interest ramifications that should be considered in
decision making about associated computer and social science research
agendas.11 The steering committee shares that belief but reports that this
project provides evidence of divergent views about specific opportunities
and needs: consensus on social objectives and how trade-offs should be
made is difficult to attain or define. Discussions and disagreements
among workshop participants point to the value of developing applications
and access technologies, including interfaces, concurrently. They also
explain why some issues are identifiable only through actual use of an interface.
SYNTHESIZING A RESEARCH AGENDA
The starting point for user interface improvement comes in multiple forms—there is no single approach to recommend. The visual approach is epitomized in commercial graphical user interfaces. Commercial speech synthesis and recognition systems exist, but they are restricted in applicability and bandwidth compared to visual interfaces. Menus are common in many kinds of systems, yet today's systems point to limitations in how they are implemented. But their success suggests that, for the near, medium, and even longer terms, progress will come through introducing more options that meet more needs rather than in eliminating the incumbents.
The next milestone depends on one's vision of the future. From almost any angle—technical, social, business, or policy—disagreements exist. Overall, the future may be reached incrementally, by extrapolating from and building directly on earlier successes. A large variety of new technologies are being pioneered that could potentially address people's needs. They include speech recognition and generation, virtual reality and advanced graphics systems, haptic devices, advanced database query mechanisms, and intelligent agents, all aided by much faster processing and greater memory than were available before. In most cases, substantial research is needed before the functionality, reliability, and appropriateness of new technologies for use by different people will be understood to the point that such technologies can be widely used.
Incremental improvements will take place; they should be considered part of the (moving) baseline. For example, more conservative members of the community point to the successes of direct manipulation interfaces and suggest that development efforts should build on them. The future may also be radically different as a result of technical breakthroughs or transformations arising from cumulative incremental developments. Some of the more visionary objectives come out of the (sub)discipline of artificial intelligence and involve credible and useful realization of natural language and intelligent systems. But to experts from other subdisciplines, those visions will not work or may be undesirable because of what they imply for complexity, control, and time investment to develop. Other tensions relate to emphasis on support for real-world communities and interactions versus support for artificial worlds and virtual environments. These are not inconsistent, but reflect different preferences and assumptions regarding people, activities, and allocation of resources. Some of the difference in perspective reflects differing views of the desirability of automating different functions or of different approaches to simplifying technology (e.g., whether functions should be hidden). What is simple, easy to use or learn, or even helpful is a matter of opinion. Some reflect attention to different segments of society: attention to information components of work and other activity tends to emphasize white-collar activities, raising questions about support for other kinds of activities, including those that may have little connection to information finding and manipulation (such as the purely recreational). Yet other differences arise in contemplating the larger architecture of the evolving information infrastructure: What is likely or preferable as the locus of intelligence, processing, various input/output functions, and so on? What capabilities belong in what kind of user device, and what capabilities belong in the network? Will one approach dominate, or will multiple solutions be sustainable technically and economically? How these issues are sorted out bears on end-user device and service options (technical features and costs), the cost structure of information and communications service providers, and the features and qualities desired in interfaces. The range of issues and their interactions underscore the value of joining social and computer science perspectives because technical capability is only a piece of the puzzle.
This project has shown that experts differ strongly in their views about what visions should shape the agenda for future user interface research. Those differences reflect the inherent biases of personal concentration and investment in a given subdiscipline, as well as differential understanding and evaluation of what has and has not worked in the past, varying orientations to medium- and long-term time horizons, relative openness to new or synthetic ideas/approaches, diverging values and frameworks, and myriad other factors. The increasing scale and diversity of the research effort make it harder for people to know about and understand progress outside their own niche; uneven understanding can be constraining in an inherently multifaceted arena. This report tries to be catholic in its approach, accepting the value of incremental and radical, as well as foreseeable and speculative, approaches and embracing the promise offered by multiple technologies and the interaction of technical and social science perspectives.
Interface research must proceed along many dimensions to accommodate every citizen and to sustain the marketplace of ideas and experimentation that composes a healthy research enterprise. It is already proceeding along multiple paths because of the fragmentation and multidisciplinary nature of the research community. Many different and often disagreeing communities exist: those oriented toward speech and those toward visual systems, those oriented toward artificial intelligence and those toward direct manipulation, those oriented toward meeting the needs of people who have disabilities or are disadvantaged, and those oriented toward the high-end, fully enabled, resource-rich users (these are characteristics often found among the early adopters of any new technology). These and other differences characterize the technical community; they are complemented by differing approaches among a variety of social sciences—psychology (e.g., cognitive, perception, social, industrial), sociology, anthropology, and economics—that offer valuable insights into how and why people use the information infrastructure and how those uses can be better in one or more ways. Diversity in research does not imply that all ideas are equivalent in merit or priority, but rather that there are risks in focusing too soon and too narrowly and that there is value in reassessing the prospects of certain technologies as conditions change. The diversity in research outlooks evidenced in this project underscores both the value of fostering interdisciplinary research and the challenge of undertaking such collaborations.
Progress toward achieving ECIs will involve basic research in
theory, modeling, and conceptualization; experimental research involving
evaluating, testing, and implementing artifacts; and empirical social
science research assessing segments of the population and how people
actually work with different systems. In all cases, data, methodology, and
tools are themselves targets for research or research support.
ORGANIZATION OF THIS REPORT
Part I of this report represents the steering committee's synthesis
of the factors shaping goals for ECI design (Chapter 2) and issues and
directions for research in relevant and promising areas (Chapters 3 through 6).
The cross-cutting issue of design and evaluation is covered in Chapter 4.
Chapter 7 presents overarching conclusions and recommendations
developed by the steering committee. Part II includes selected position
papers prepared for the project's August 1996 workshop. These papers
contain additional details on overall and specific ECI issues and the authors'
personal recommendations for further research. Additional position
papers are posted on the World Wide Web at http://www2.nas.edu/CSTBWEB.
NOTES
1.Use of the NII is, of course, a matter of individual choice. Although many factors affect who chooses to use what technology, when, where, how, and how well, contemporary experience shows that many people seem to want to use infrastructure technologies. Further, economists have shown that individuals benefit from widespread use of network technologies, suggesting that enabling use by more of the population will benefit more than just those newly accommodated. Broader deployment, if not ubiquity, and broader use allow for the economic and social benefits of what economists call network externalities—individuals gain value and appreciate networked systems more as more people become connected users. Even if advocacy based on the social or societal value of enhanced information infrastructures is premature, as some skeptics argue, the evidence for the promise of the technologies is great enough to make continued efforts to improve interfaces a wise national investment.
2. Some of the difficulty reflects the explosion in numbers of features and capabilities in the software and information infrastructure, a result of which is to make the older WIMP (windows, icons, menus, and pointers) interfaces more clumsy to use. Whereas an early-1980s desktop-style interface might be used with perhaps 50 documents, orders-of-magnitude more documents might be used in late-1990s applications.
3. Many of these support costs reflect limitations of today's interfaces, inasmuch as they involve ongoing training, consultation, and third-party adaptation of how a system is configured to meet the user's needs; these costs can be compounded by significant operating costs for telephone and cable television service. Note that even mouse pointing devices confound people initially, as recognized by Microsoft's inclusion of a solitaire game with Windows software packages.
4. Others building on that research legacy include Motif, the evolution of the Microsoft Windows line through Windows 95, and Mosaic and other Web browsers.
5. Context and motivation are provided by changing needs, an example of which is the contrast between searching a single document and searching vast repositories of documents (with the Library of Congress providing the canonical example). Research can enable a competitive marketplace to produce a wider variety of interfaces and applications, supporting citizen choice based on individual perceptions of needs and wants.
6. Other principles or goals can be framed as a function of value judgments about what people should be able to do and under what circumstances. Goals related to specific applications such as health care, education, or other domains can be very helpful in suggesting attributes for interfaces, and these are the focus of separate literatures, references to which are scattered throughout this report. For example, as Sproull and Faraj (1995) point out, policy discussions of the Internet and other electronic networks tend to assume that these media are mainly informational in nature and that users chiefly want better ways to browse and find the information they desire or to send information to others. Empirical social science studies suggest a contrasting view of users as complex social beings whose information needs are inextricably bound up with a collection of other ends that are communicative, participatory, and social in nature.
7. The expectation that computer and communications systems will take many forms and be used in many contexts underscores a caution voiced by some project contributors against "personal computer centrism": many devices and systems present many interface needs. The proliferation of technology promises ubiquity—eventually—and calls attention to the context of use: there is an evolution of what Mark Weiser, a workshop participant from Xerox PARC, has called "personal information infrastructures," and systems can be differentiated or used in common among personal, household, organizational, and public facility or public space environments.
8. Taking advantage of improvements in the cost, performance, and capabilities of the underlying hardware technologies, as well as the basic and applied interface research that has made possible today's successful commercial implementations, today's successful infrastructure products (epitomized by Internet services now extending to audio and video delivery) are often creatures of software. User interfaces are also driven increasingly by software, which can increase (or decrease) the accessibility of computer-based products to meet varying needs. Hardware improvements have enabled faster computing speed, one of the key sources of advances found in commercial graphical and, to a lesser degree, speech recognition and synthesis user interfaces.
As Gregg Vanderheiden (1996) explained in testimony to the Federal Communications Commission:
Software determines the user interface more than hardware: In today's computers and most telecommunication and enhanced telecommunication devices, the user interface is almost entirely defined by the software rather than the hardware. Although the hardware provides some limits on what the software can do, the bulk of the user interface is determined by the software. Work with Apple Computer, IBM, Microsoft, and others in computer operating systems has shown how much disability access can be achieved without making any hardware changes. Software can be used to make mouse functions operable from the keyboard for those who cannot operate a mouse. Software can allow screen displays to be made accessible to individuals with low vision or blindness, and information emitted by speakers to be displayed visually for individuals who are blind—all without any changes in the hardware. In fact, the accessibility of almost any product can be tremendously enhanced by modifying nothing more than the instructions (the software) which govern its behavior. On the other hand, relatively little can be done to make a product more cross-disability accessibility without addressing the software issue.
9. Section 255: Access by Individuals with Disabilities.
(a)DEFINITIONS—As used in this section—
2. READILY ACHIEVABLE—The term "readily achievable" has the meaning given to it by section 301 (9) of that Act (42 U.S.C. 12181(9)).
(b) MANUFACTURING—A manufacturer of telecommunications equipment or customer premises equipment shall ensure that the equipment is designed, developed, and fabricated to be accessible to and usable by individuals with disabilities, if readily achievable.
(c) TELECOMMUNICATIONS SERVICES—A provider of telecommunications services shall ensure that the service is accessible to and usable by individuals with disabilities, if readily achievable.
(d) COMPATIBILITY—Whenever the requirements of subsections (b) and (c) are not readily achievable, such a manufacturer or provider shall ensure that the equipment or service is compatible with existing peripheral devices or specialized customer premises equipment commonly used by individuals with disabilities to achieve access, if readily achievable.
(e) GUIDELINES—Within 18 months after the date of enactment of the Telecommunications Act of 1996, the Architectural and Transportation Barriers Compliance Board shall develop guidelines for accessibility of telecommunications equipment and customer premises equipment in conjunction with the Commission. The Board shall review and update the guidelines periodically.
10. Lee Sproull, Boston University, personal communication.
11. This is central to the linkage with universal telecommunications service, a concept many seek to expand from telephony, and to the discussion of public access points, from network access computers in libraries to kiosks in shopping malls.