C.7 Ease of Use

Principal Authors:

Bonnie E. John, Raj Reddy and Rod Smith

Additional Contributors:

Ruzena Bajcsy, Doyle Baker, Sara Bly, Anita Borg, Bill Buzbee, Elizabeth Coppinger, Michael Deering, James Farber, Joel Frendt, Kevin Lacobie, Robert Logan, David Marques, James H. Morris, Steve Oberlin, Eldon Patterson, William A. Rowe, David R. Schwartz, Neil Scott and Alan Tannenbaum

1. Introduction

The success of the National Information Infrastructure ultimately depends on it being usable by real people. If access to information and communications-based services is to be effective and universally available, the user interface to the network and its services must be easy and fun to use. Simply providing bandwidth and connectivity, which has served the scientific community well, will not be sufficient for the NII without ensuring that services are accessible and usable by people without computer training.

Examples of the importance of the user interface to the success of communications and information systems are easy to find. For example:

A major factor in the success of the basic telephone system is its simple and ubiquitous user interface. Even a child can dial or answer a call--anywhere in the world.
The usability and attractiveness of graphical, windowed computer interfaces are largely responsible for the recent explosion in the use of personal computer software by professionals and their families.
By providing an interface that is easy to learn and use, Mosaic is opening the Internet to a much broader audience than ever before.

2. The Technical Challenges of Universal Access

The theme of universal access interfaces focuses on the notion that all users without exception should have availability to NII services. This concept has three prongs: affordability, ubiquity and usability. The lack of any one of these can represent an impediment for potential NII users. The human-computer design principles for universal access interfaces should be clearly defined as:

Not all users have the same economic resources. The interface must provide adequate access to information on inexpensive devices.
Not all users have the same cognitive and physical capabilities. The interface must provide all users with equivalent access to information despite these differences. Additionally, the interface must address the differences in the levels of skill and experience of various users.
Not all equipment is equal in function. Different equipment requires different interaction paradigms--i.e. set-top boxes for interactive TV, personal computers, telephones, kiosks, personal digital assistants, etc. The interface should meet the needs of various classes of users and should not be forced to some lower common denominator.
An NII user interface must automatically adapt to the characteristics of the connection--e.g., different physical interface connections, resolutions and bandwidths.
The network must support the connection of a variety of access devices meeting the wide-ranging needs of the entire population.
Although a wide variety of interfaces must be supported, users should not have to wait until all have access in order for some to have access. Careful balance is needed between the conflicting goals of universal availability and rapid availability.

The approach for achieving universal access interfaces should be to focus on research issues that are at the intersection of NII needs and general ease-of-use research.

3. Research and Development Recommendations

Ensuring that the NII will be usable by the widest possible audience requires an explicit program of research, standards and development that focuses on three critical elements:

Understanding user needs at the individual, organizational and sociocultural levels.
Identifying and promoting technologies critical to ease of use.
Establishing processes and metrics to guide the development of usable interfaces.

This section provides key findings and recommendations for a research agenda in each of the above areas.

3.1 Findings and Recommendations: Individual, Social and Cultural Aspects

Developing an in-depth understanding of the individual, social and cultural aspects of NII users will be one of the most important first steps of the ease-of-use research. The wealth of knowledge developed initially by studying the areas of education, health and electronic commerce can set the groundwork for evolving NII services and applications to reach a broader audience over time. The issues include understanding the user on individual, organizational and sociocultural levels; and understanding tasks, environments and multiuser collaboration.

3.1.1 Understanding Users

Americans are extremely diverse. To achieve universal accessibility of the NII, the differences and similarities between individuals and groups must be understood. This research can be broken into multiple research areas:

1) "Walk-up-and-learn" interfaces will be the predominant force for universal access. Understanding how individuals can use their previous real-world knowledge to navigate through the NII and use its applications is the purview of psychological research. Although there is much existing research in psychology that can be of use, additional support is needed to put that knowledge into a form that developers of the NII can use to design their systems.

2) NII interfaces must "evolve gracefully." That is, consumers must be allowed a step approach to access the NII with increasing power and sophistication as their familiarity, needs and equipment evolve without relearning a core set of skills. This is sometimes called "research in gentle-slope systems."

3) The cultural context of the NII must be explored. Basic research on broadly defined aspects of American life is needed to identify opportunities that could improve quality of life both at work and at home. Analysis of current trends can point to new product and service categories. Broad domains of trend research should include entertainment, communications, work and travel. In addition, the existence of the NII itself will create new social systems in the United States that will have their own emerging needs. Sample topics include:

Electronic mail as a new human-to-human communication medium, no less significant for culture than writing or the telephone. Understanding this new medium is essential in helping to anticipate and design new solutions.
Electronic forums and electronic town halls and their future role in the democratic process. What is the viability of such tools in the democratic process and how it is implemented?
Evaluation of music videos and video games to understand their applicability to the education process. The intrinsic lure of entertainment to youth offers an opportunity for students to author their own intellectual work.
Evaluation of the growing virtual office and telecommuting trends. These early adopters of NII-type technologies represent the cutting edge of a cultural change in society.
Finally, pre-competitive research on sociocultural trends is required given that this is not a typically profitable area for the private sector to invest in and given the need to ensure that these findings are publicly available and not held as proprietary information.

3.1.2 Understanding User Tasks

Rigorous investigation of existing and anticipated tasks associated with the NII is required to fully identify the scope of user requirements. Task models are the basis for all product design and the NII should be no different.

3.1.3 Understanding Environments

Users and technology cannot be considered in isolation. Rather, both exist in the context of environments. Traditional environments such as offices, hospitals and schools have received some investigation, but more is required with the onslaught of enabling technologies. Additional understanding of less traditional environments such as inner-city public access terminals, transportation and recreational users is also required.

3.1.4 Understanding Multiuser Collaboration

Much work and play is accomplished in the context of social activity; rarely are tasks done in isolation. The NII offers access not only to extensive sources of information but to varied and far-flung interactions. Users will have the ability to connect to friends, colleagues and experts who previously would have been accessible at best by phone or mail. The NII capabilities further such interactions by erasing the boundaries of space and time. We list four research topics here:

1) Much of the work and entertainment is social. Current single-user models of computer and networked use cannot meet such a need. Research is needed to understand the workplace's social context in order to bring this context into network/computer-mediated work. For example, medical providers need to have ad hoc "hallway" conversations with their remote colleagues just as easily as they now do with their colleagues in the same building. Electronic mail and telephony are not rich enough to provide sufficient social context for these ad hoc interactions.

2) As networked (NII-mediated) collaboration becomes more common, community access, privacy and community control become areas needing significant research. For example, medical patients depend on privacy, yet need to be able to find similar case studies and share information with support groups. Research is needed on access models for collaboration that can meet these diverse and often conflicting needs.

3) Through the NII, most users will participate in many different "communities," both geographically local ones and geographically dispersed communities. Research is needed on how users will do that, how they will socialize in the new networked communities, how different access rights will exist across communities and how users will "code switch" between them in the same sense that multilinguals code switch between communities. This has very strong user interface implications that are not well understood at present.

4) There are a variety of technical research issues such as different levels of interaction needed and desired for different tasks and different access equipment.

3.2 Findings and Recommendations: Technologies

3.2.1 Hardware

Many proposed uses for the NII cannot be successfully implemented today. Barriers include limitations of device performance, a high cost threshold and a lack of interface software. Development of enabling technologies would expand interaction via the NII at all levels (from lowest cost to the highest performance).

Development examples where stimulus is required include:

Remote expert image analysis applications (e.g., medical) require currently unavailable display performance--higher resolution, higher fidelity, and color and monochrome displays of high visual quality, free of the quite-visible annoying artifacts.
Mobil access to the NII will require thin, high-resolution, low-power, lightweight portable displays. The ability to view three-dimensional images requires that the NII include three-dimensional graphical user interface standards, hardware and software.
The general population has little computer literacy and will adopt the NII faster and in greater numbers if natural and more familiar I/O devices and interfaces are available (i.e., pen, speech, joystick and touchpad, either singly or concurrently).
Mass adoption of the NII demands low-cost NII interfaces that use the TV as the display device.

Other needed hardware items will be developed naturally in response to the economic stimulus created by the NII (bright projection displays, distributed storage, low data capacity backchannel, microphones).

3.2.2 Adaptive Intelligent Interfaces

Given the scope and magnitude of the NII, new approaches to end-user interfaces and their construction will be necessary. For example, in the medical field, new user interfaces that continually learn and remember information regarding specific areas of medicine will be essential to keep up with the information dynamics of recently published research and studies. Additionally, these types of interfaces will "learn" a person's viewpoint or interest to build self-describing search criteria.

These interfaces will be characterized by:

Their ability to adapt to users' unique needs and skills. This includes such things as recognizing capabilities and limitations of the user's current context, including hardware, and adjusting the interface accordingly.
Their ability to know what they do not know, engage in a dialogue with the user about that knowledge gap and learn from that dialogue for future use.
Their ability to learn about users, their tasks and the NII through diverse means appropriate for different situations. For example, learning from deliberate instruction or demonstration by the user, through exploration or through passive observation of user behavior. Systems with several learning strategies must also be able to reason about the appropriateness of those strategies to particular situations.
Their ability to intelligently filter information so individual users can choose to limit access to specific types of information (e.g., filters that recognize pornographic images, enabling parents to prohibit them from entering their homes).
Being as easy to construct a suite of agents personalized to an NII user as it is to build a playhouse with a child's interconnecting blocks.

Toward this end, a significant effort must be undertaken to create intelligent, adaptive computer-based systems that integrate seamlessly into the human use of the NII, including the technology base that enables systems developers to rapidly create and deploy systems that exhibit intelligent behavior.

3.2.3 Human-Centered Input Methods

Human-centered interfaces refer to natural human-machine interactions using advanced input methods. The closer NII interfaces match real-world concepts, the faster we will achieve acceptance of those services. Pen, speech and natural language are viewed as the first of these input methods.

Pen (and touch) enables the paper-like interface so common in education, authorization, forms and writing tasks. Speech has the potential to have a completely hands-free modality that can be envisioned as the ultimate interaction technique in the home and hands-free environments, as well as enabling the NII for the disabled and citizens with varied education levels. Together with natural-language comprehension and generation, these methods will enable users to interact with services using person-to-person metaphors.

In the future, NII interfaces will need to automatically sense additional physical information about the user, tracking the location and orientation of a person's sense organs (eyes, ears, mouth, head), as well as their hands (for gesture recognition) and potentially their whole body. This extends to sensing the location of a person within a room, building or city.

In order to be successful, we must evolve the research in navigation, recognition and semantic understanding technologies to the next stages of handling unconstrained input. Human-like feedback technologies, including natural voice response, anthropomorphic agents and intelligent dialogues complete the package and need to be a key focus area. Natural-language processing (NLP) is an important enabling technology for many of these features, and research should be continued in this area for the NII.

The use of the pen, speech and other modalities simultaneously has the potential of having a synergistic effect when applied to the communication of ideas (e.g., voice-over-ink annotations) and the recognition of written or spoken commands and data.

3.2.4 Visualization

NII participants will be bombarded with services and data from a multitude of sources. Our methods of presenting this information in digestible forms must be enhanced. Advanced visual (two-dimensional and three-dimensional), auditory, tactile and other sensory "views" can increase the user's ability to filter this information and identify, extract and enhance its content. Also, advanced techniques of cinematic direction enhance user interaction with anthropomorphic agents.

Examples include:

Navigating and sifting through abstract views of NII services that vary on a variety of dimensions.
Remotely analyzing and annotating (via voice, digital ink, etc.) medical images.
Remotely interpreting multidimensional data residing in centrally located, corporate/public-domain sources.
Interacting with anthropomorphic agents that provide varied, useful and natural views of themselves.
Acting within virtual environments for entertainment purposes.

Research is needed in:

Studies of how sensory dimensions (e.g., color, pitch, texture) can be used to code information and how users relate such sensory dimensions to form integrated perceptions.
Studies of multidimensional coding strategies that allow users to effectively navigate within a large body of elements and to narrow their search by successive filtering of these views.
Studies of how modern cinematic techniques are used to broaden acceptance of cinematic characters by an audience and how similar techniques could be enabled for use with anthropomorphic entities.

3.2.5 Authoring

By "authoring" we mean the creation and organization of the multimedia, hypermedia, video and virtual environment information presentations that will be made available on the NII.

Creating high-quality, effective multimedia/hypermedia for the NII, be it for education, entertainment or information presentation, takes too much time (sometimes as much as 1,000 hours for one hour of presentation) and requires considerable expertise. This represents a "production bottleneck" for one of the key commodities that will be distributed (and indeed created) with the NII.

The speed issue can be addressed with research to develop metaphors and approaches for authoring systems, with a goal of an order of magnitude decrease in authoring time. The expertise issue can be addressed by developing authoring tools that encourage and facilitate good design practice, by means such as intelligent agent "design critics," including "codes of practice" in the tools or in templates, style sheets or examples. One objective of this work should be to empower every individual user of the NII to be an information creator.

Similarly, shared virtual environments (including the simulation of physical processes that so often are an important part of these environments) need to be much more easily created than they now are. Tool kits that minimize and in some cases eliminate traditional programming are needed.

3.3 Findings and Recommendations: Metrics and Evaluation Techniques

The developers and consumers of NII software and hardware need to know how effective the products and services offered will be in fulfilling their needs. Consumers need to be able to make cost-benefit trade-offs for any particular access mechanism (e.g., I/O device, software options for accessing the NII) or application on the NII before they buy. The costs include usability issues like learning time, organizational-fit and satisfaction, as well as direct monetary costs. Developers need to understand the usability implications of their design decisions to direct design effort.

To fulfill this need, research is necessary to define metrics and techniques (observational, analytic and heuristic) for assessing individual usability, group usability and how NII systems will fit into a social or organizational context. An important step is to validate these metrics and evaluation techniques by comparing their results to rigorous empirical tests of real systems in real-world use. The definition and validation of metrics and evaluation techniques are each themselves an iterative, continuous process during the life of the NII.

3.3.1 Metrics

Currently, the most well-defined metrics are for individual performance: time to perform a task and time to learn to perform a task. Although these metrics are important to a large segment of U.S. workers, they do not cover many other aspects of NII uses such as:

1) Individual metrics:

Errors: Even the definition of "error" is ambiguous. Does an error occur whenever someone deviates from the optimal procedure for performing a task or only when someone behaves in a way that prevents the task from being successfully accomplished at all? Are behaviors then the cause of frustration errors? How do you consider quality in the error metric?
User satisfaction: Some work has been done (e.g., University of Maryland's Questionnaire for User Interaction Satisfaction), but additional validation and extension to groups is needed.
Quality of product: What constitutes quality? for individuals? for teams? for communities?
Creativity.
Fatigue, boredom, etc.

2) Group metrics:

In addition to the group analogs of the individual metrics, metrics specific to groups and communities need to be developed and validated, e.g., level of cooperation or homogeneity of a group.

3.3.2 Evaluation Techniques

Evaluation should occur at many levels, for example:

Individual components: e.g., Are high-resolution, cost-effective displays readable for some benchmark texts and graphics?
Integrated applications: e.g., How effective is combining pen with speech for a data acquisition task?
Use in context: e.g., Does use of the NII make a public school teacher more effective in teaching a particular curriculum, and if so, how?

Evaluation can reflect national priorities that industry will not do, e.g., does an entire school system meet its educational goals in a cost-effective way using the NII, and if so, how?

Three types of evaluation techniques have arisen from human factors and human-computer interaction (HCI) research: observational, theory-based and heuristic. Currently, observational techniques--i.e., watching users as they use an application--are dominant in the computer applications industry and will undoubtedly continue to be useful in the development of the NII. In fact, the NII poses the opportunity for forums where users of applications directly share product experience and potential users can judge whether they wish to invest both time and money in a particular application. However, the NII also poses unique problems for evaluation. In particular, when applications move from being a single-user in front of a workstation to groups of users geographically and temporally distributed, observational methods are not easily used. Due to this and other drawbacks of the observational approach, the NII requires research into other methods of evaluation.

Theory-based and heuristic evaluation techniques are promising approaches in that they 1) provide evaluation at many stages in the design process, 2) remove the dependence on the availability of representative users and 3) remove the dependence on highly skilled usability testers. Each of these techniques needs to be validated by comparing its predictions to rigorous, empirical trials of real NII applications in real-world use. Also, each of these techniques needs to be usable by members of typical design teams in real development organizations.

Theory-Based Simulation

A modern engineering discipline requires modeling techniques and simulations to rapidly manipulate and evaluate design ideas. Design for the NII requires theories particularly associated with modeling users, their tasks and activities, work flow processes, environments and human communication. To fill this need, research is necessary to develop and validate simulations of users, tasks and environments (i.e., simulated users in a simulated community), of NII hardware and software technology and of the discourse through which these entities communicate. The methods for producing and using such simulations must be themselves usable, cost-effective and usefully fit into the design cycle. Additional research is required in:

Cognitive models to simulate different types of individual users who have different capabilities, backgrounds or training. This would reduce the need to find people to represent a user group, relaxing the dependence on specialized user groups like experts or physically challenged users. A start has been made with the GOMS modeling approach (Card, Moran and Newell) that predicts expert performance time and learning time, but needs further expansion to other metrics.
Social theories to simulate the sociocultural implications of the NII. Researchers in computational organization theory have demonstrated initial progress in predicting behavior of real organizations in real situations, but continued research is necessary to serve the purposes of the NII.

Heuristic Techniques

As design experience accrues through the continuous evaluation of the NII, this experience could be captured and applied to new designs through heuristic techniques. Such techniques have started to develop, but need to be expanded to group applications and validated with observed performance in the field. These nascent techniques include: "Heuristic Evaluation" (Nielsen), Claims Analysis (Carroll and Rosson), guidelines, Cognitive Walkthroughs (Lewis and Polson) and prototyping tools with design critics (Fisher). Research in refining these methods and creating new ones should be supported to fill the needs of the NII.

3.4 Findings and Recommendations: Standards

The NII needs to establish a core set of user interface standards to provide a framework for developing products and services that are as ubiquitous and as easy to use as basic telephony.

User interface standard specifications must ultimately provide uniform hardware and software benchmarks to measure both performance and compliance. Visible and explicit evaluation of competing solutions drives technical refinements in the appropriate direction and rewards cost-effectiveness.

Standards are needed to enable easy-to-use services across different platforms and different media.

Recommended research should focus on two areas:

Defining key elements of the user/service functions, capabilities and commands. For example, the automobile has a small number of standardized functions (steering, acceleration and braking) that make basic functionality accessible to all users, but the rest of the automobile is left to the creativity of the designers. Likewise, research should be done to identify the core functionality that needs to be standardized for basic use (e.g., the ability to undo actions, the ability to interrupt processes), leaving the vast majority of decisions to the ingenuity of NII designers.
Usability-enabling software, including user-centered application programming interfaces and protocols to support essential user-level functions across different devices and platforms.

3.5 Findings and Recommendations: Pilot Projects and Testbeds

Evaluation of the NII itself and its applications should be early, ongoing and its results should continually feed back into design. One approach would be to create a national "virtual" user interaction laboratory for defining, evaluating and conducting research that will foster collaboration and motivate NII product developers to make ease of use a high priority. There are also several testbeds already in place that should be studied immediately to begin the evaluation/design symbiotic relationship. For instance, research should be conducted into the individual use, social use and cultural implications of testbeds such as Mosaic, the World Wide Web and NSF's Scientific Collaboratories so that successes, failures and unexpected results can be understood and fed back into the design of NII products that come after. New testbeds and/or pilot projects such as those outlined below should also be developed to foster further development, introduction and evaluation of new ease-of-use technologies.

3.5.1 Health Care

Health care offers many interesting challenges and opportunities.

The realization of "virtual consultation" will:

Improve quality of care by spreading the availability of scarce consulting expertise.
Reduce costs by eliminating travel and duplication of physical resources.
Improve the timeliness of health care by overcoming geographic barriers to consulting expertise.
Improve the care giver's access to information and colleagues.

Patients will be empowered by:

Providing improved access to medical literature, case studies and personal records.
Improving care environments with access to support groups.
Providing home care with remote supervision and monitoring.
Improving care environments for both care givers and patients by providing more of the rich social context for interaction that is missing in unidimensional interactions such as mail, electronic mail and telephony.

3.5.2 Education

Students need interactive access to educators with expertise not locally available, in either a virtual classroom or monitoring environment. Students need to collaborate among themselves, forming geographically and culturally disparate cooperative learning projects.

"Virtual" instruction, tutoring and consultation will:

Improve the quality of education by allowing greater specialization of educators, yet greater sharing of that expertise.
Reduce education costs by spreading the availability of scarce educational expertise, reducing travel, and sharing information and resources.
Enable cooperative learning environments and improve the quality of education, particularly where the social aspects of learning that are so lacking in "distance education" today are included.
Improve the quality of education by promoting community and industry involvement.
Improve the scope of successful education by facilitating the education of many who are awkward socially (and prefer anonymous networked interaction) or mobility disadvantaged.

3.5.3 Business

Conference or meeting attendees need to participate without the time and expense of travel. Spontaneous, informal and low-overhead access (consistent with the goals of the NII) allows more-productive collaboration between meetings as well. The potential effects on business are:

Reducing costs by reducing travel.
Increasing job opportunities by creating access to jobs--without collocation/relocation--for those unable to come to the workplace. Job information can be made available without expensive exploratory travel.
Potentially changing the face of American business in unforeseen ways, only understood through social research. Just as the automobile caused the development of suburbs, use of the NII for work might have equally far-reaching changes.

3.5.4 Entertainment

This area offers the possibility of social games to geographically disparate (often anonymous) participants. The NII could increase the entertainment market by adding social dimensions not found in current games, i.e., social dimensions that appeal to many of the population who do not like current computer-based entertainment (games of movies).