The focus of this thesis was on the use of small peripheral desktop displays that utilize motion-based and in-place animation techniques to convey the state of and changes to dynamic information resources. Experiments and user studies explored relative advantages of the different animation techniques and the impact of user characteristics on preferences and performance. A widget set supports programmers in building awareness applications by providing animation capabilities and related support.
The pilot study examined the use of animated displays as a way to maintain information awareness. The participants in the study seemed willing to sacrifice desktop space for heightened awareness. Most of the participants were content with the amount of information provided (though some desired more), and no one listed distraction caused by the animation as a reason for not using the program. The results described in this study were a stepping stone to better understanding the potential use of animation in maintaining awareness.
The empirical evaluations explored the balance between distraction, reaction, and comprehension and memorability when using peripheral animated displays. The first experiment showed that fading, tickering, and blasting peripheral displays can be non-distracting yet can effectively communicate information, and that the type of animation impacts performance on awareness tasks. The second experiment showed that changes in size and speed can impact performance on awareness tasks.
The development and analysis of Agentk provides programmers to include animated effects in their applications using a familiar programming interface. Use of the widgets makes it easier for end-users to have a sense of control over the information flow while maintaining a desired level of information awareness. The advantage gained from using Agentk is that constant, cyclic animations can be developed easily and rapidly, a definite advantage given the immediate needs that arise in the awareness problem domain.
The observational study provided a way to predict animation preferences using readily available characteristics about the users. Four user categories were identified that show promise in being able to associate user characteristics with animation preferences. This association would allow designers who have an understanding of their target population to construct displays that better communicate information.
Overall, this research has developed an understanding of the nature of the awareness problem and of how users' wants and needs differ in maintaining awareness. It is reasonable to conclude that the use of animation can assist in maintaining awareness without causing undue distraction in particular situations. While the experiments and studies described in this thesis looked only at short-term situations, numerous situations exist when a animated peripheral interface that is used even for short periods of time could potentially be beneficial: tracking breaking news stories on hot topics, monitoring the weather before a meeting in another building, or keeping an eye on volatile stocks. As with most tools, peripheral animations can be and have been misused, but when used properly, this thesis has shown they have the potential to be beneficial.
To increase the chances that peripheral animated displays will be used wisely and to improve the user perception of animated displays, programmers should not necessarily write applications that employ animation with the expectation that they will be used continually, but rather for short, well-defined periods of time. Perhaps they should be designed only to monitor the traffic between 5 pm and 6 pm every weekday, or to watch the scores of selected baseball games during the pennant drive and playoffs. If programmers do anticipate that it will be necessary to run the application at all times, alternate (non-animated) information delivery mechanisms should be made available as well.
Will people use peripheral animated displays to maintain awareness? The results of the experiments and studies suggest that they will, as do the popularity of certain peripheral displays that have been in use for many years. Consider instances when users are willing to tolerate constant animation. Glancing at a clock on the computer desktop to obtain the time is far easier than running a command or even looking at a wristwatch, and looking at a load monitor while running a compute-intensive program is far easier than running commands to determine the system load. Besides being immediately available, another distinction is that the changes to the display are small, subtle, and predictable, allowing the user to adapt to the changing display to the point where it is not in the least distracting. By providing smooth animations found in Agentk, the applications will be smoother and less likely to distract the user unnecessarily. The programmer and user can assist in this by choosing appropriate sizes and speeds for the application based on the anticipated informational goals of the user. It is necessary to understand the trade-off between the importance of the information being communicated and the resources necessary to display and process it. People will use applications and designers should create applications if they feel the trade-off is justified.
The choice to study fading and tickering effects was based on the expectation that other in-place animations would show similar results to fade and other motion-based animations would show similar results to ticker. Initial evidence from these studies and elsewhere seem to support this claim; for example, the first experiment showed that the in-place blast effect resulted in similar performance to the in-place fade, and the Maglio work described in Chapter 2 showed that motion-based vertical tickering resulted in similar performance to horizontal [45]. However, further evaluations should be performed to explore the similarities and uncover possible differences.
Another area for further exploration is the effect of peripheral animation on various different central tasks. In my empirical evaluations, I selected browsing tasks to be performed because several pilot study participants indicated that for them the most likely time for using animated displays was while browsing the Web. However, many potential users may want and need to use animated displays at other times. While the evaluations in this thesis showed that browsing tasks are not affected by the presence of peripheral animations, the Maglio study showed that performance on another task, document editing, does seem to be affected [45]. Perhaps the most affected tasks are highly interactive tasks, or perhaps cognitively demanding tasks, or reading-intensive tasks. Further studies using different central tasks would help in understanding the degree to which different tasks are affected by the presence of animation.
Finally, while the experimental results in this thesis have indicated that performance differences exist based on characteristics of the animation being used, additional research is needed to explore why differences exist. Additional experiments that used eye tracking could determine how people budget their time between primary and peripheral tasks. Are certain animated effects or characteristics of animation more likely than others to attract an individual's attention? Do users read the information in peripheral displays or merely glance at them in search of certain key words or names? Are users less likely to glance at a peripheral display when performing a more cognitively demanding task? Or do they glance at the peripheral display at about the same rate but perform worse on the central task? Empirical evaluations similar to the ones described in this thesis, but augmented with eye tracking capabilities, could shed light on these and other similar questions.
The development of any user interface toolkit can be a never-ending process. There are always more features that can be added. This section discusses extensions and improvements to Agentk suggested by end-users and programmers.
The Agentk toolkit provides support for using animated effects in peripheral awareness applications. However, only a small subset of all possible behaviors are represented. Extending the widget set to include other behaviors such as shrinking, growing, start-and-stop, and swiping will provide a wider variety of techniques for programmers to choose from and should attract a larger population of programmers and end-users.
In addition, the collection of options should be extended to include Disney-style techniques such as motion trails and slow-in/slow-out [41]. These effects have the potential to lessen the distraction caused by the animations and improve the ability of the widgets to communicate by generating smoother changes.
Ultimately, I hope to create a framework within which any programmer can add new widgets. In creating Agentk, it has become clear that much of the code can be shared among widgets. Already I have isolated and encapsulated routines that have been and can be reused in other widgets. By only requiring programmers to specify a brief description of the behavior of the widget and the options, they are more likely to expand the behavior of the toolkit, benefiting not only themselves but future users.
While the use of peripheral animation has proven to be useful in helping to maintain awareness, future research should move beyond looking at animation alone and consider how it can be combined with other communication media to further enhance performance on awareness-related tasks and activities. The Agentk toolkit described in Chapter 6 included two graphical augmentations to the animated displays, automatic highlighting and history-based shadowing, that showed promise in informing the user of changes. Agentk also allowed text and images to be combined in animated widgets (as seen in the CWIC passive image browser described in Chapter 6), though the effectiveness of fading images has not been analyzed.
Auditory cues have proven useful in assisting with the monitoring of information [16]. As was seen in this thesis, there is a fine line between monitoring and awareness -- often the tasks intertwine. As such, combining auditory cues with animated peripheral displays may help draw the user's attention in important situations. This was examined briefly in [45].
Ultimately, the goal is to automate the awareness process in the spirit of prior work in the automation of graphical information presentation by Mackinlay, Casner, and others [44, 12, 15, 71]. Much of the research to date has focused only on characteristics of the information and characteristics of the task at hand. This research extends the work by considering that users must accomplish not just a sole task but multiple tasks at any given time. I want to support easy identification of awareness mechanisms for arbitrary dynamic resources and arbitrary situations. As a first step, perhaps the user would specify only the size and position of an available display area and the information resource to be displayed, and an automated system would then identify the best communication mechanism for that situation. In the end, the process should be automated to the point that communication mechanisms are categorized based on various features of situations in which they would be used. The nature of the mechanisms that are selected for a given situation would depend not only on the information in the resource, but on the level of importance the user gives the information at the time and on general characteristics of the user. When information changes, it should be delivered to users in an appropriate manner befitting their current and future needs.
There are many possibilities for the application of this research to building systems to address awareness needs and related problems.
The user studies that were conducted took place in an academic setting, but the potential for use extends to other settings as well. Consider the implications of this work on a classroom lecture situation. Students could submit questions silently and electronically, and they would appear on the lecturer's display. When the question appeared, the lecturer could make the decision whether to address it immediately if it were a timely and relevant addition, or to wait until later if it were a point that was to be discussed in the near future.
In another example, consider the incorporation of peripheral animations in a shared break room. A wall-sized flat-panel display could present peripheral displays for the entire room, showing interesting headlines, images from the Web, local news of interest, and other topics that could spurn conversation and interactions. Using cameras and microphones, the wall-sized display could be connected with other displays in other common areas, helping to facilitate interaction and establish community.
As a third example, consider the application of animated peripheral displays in a virtual reality (VR) environment. One common use for VR environments is in virtual walkthroughs of places unfamiliar to the users to acclimate them. The results from this work suggest that certain types of animated peripheral displays would be better suited than others for informing and alerting the user of buildings, roads, and other elements in the environment.
Another area of interest for the application of the thesis results is in addressing the needs of different user populations. While the testing focused on a fairly narrow population, this work has potential benefits for different populations, such as older adults or users with special needs. The population that was examined in this thesis may be the ideal one for using animated displays as many of them grew up using highly interactive interfaces. Examining a group that does not have this background would provide an interesting point of comparison. Users with special needs, in particular users who have limited use of their hands or arms, may benefit from having peripheral animated displays. Information could be delivered to them without having to manually scroll. Many of the RSVP and automatic scrolling experiments described in Chapter 2 are targeted for this group [36, 28, 37].
In summary, the work described in this thesis made the following contributions:
These contributions extend the understanding of the utility of animated peripheral displays in assisting in information awareness situations.