Critical Information Design

Abstract

In virtual worlds such as Second Life, every user has the potential to become an interface designer; indeed, the success of Second Life relies upon the creativity of its user community. Unfortunately, too often inexperienced designers fail to critically evaluate their creations in virtual worlds, and combined with the inherent limitations of the system, this results in user frustration, a sense that a task is taking “too long” to accomplish, and ultimately, an unwillingness to participate in Second Life. Prior research explored the messages the tools of design send within Second Life–what metaphors are successful, and which are not; how those metaphors play upon affordances already created in the “real world;” and ultimately how the confirmation or violation of user expectations of object behavior impacts the affective state of the user. This paper will use a short case study and additional literature to introduce critical design practices in interaction design, and propose design solutions for Second Life artifact creators.

Literature Review

Metaphors are excellent tools, and lend themselves exceptionally well to 3D virtual worlds, by virtue of the system’s ability to mimic real world objects. In Second Life, these metaphors can evoke powerful positive affective responses in users, but have the potential to also cause confusion in users.My prior research investigates how metaphors confirm or violate user expectations; if an object behaves in an unexpected way, then this is more likely to cause a negative response. However, the greater point is that metaphors are tools to be used, not design frameworks in and of themselves. To that end, this paper looks at two key methodologies–pragmatic design and interaction criticism–to identify the means of approaching design in virtual worlds.

Pragmatic design is an established design concept; essentially, designers should work with an intended user response already in place. These responses are varied, but can be sensual, emotional, spatio-temporal, and compositional in nature, in addition to cognitive and analytical responses. Sensual responses are based on physical appeal; emotional responses, similarly, are affective reactions to design. Spatio-temporal responses are situational in nature–how a user responds to design within the context of a place and time. Finally, a compositional response pertains to how well a user makes sense of the how and why of a given design. Incorporating all these responses during the design planning and evaluation period is critical for more than simply virtual world software; indeed, designers of all stripes must consider the impact an aesthetic or functional design will have upon its audience. Much of Second Life anticipates the first three design responses already, as are evident in the lushly detailed zen gardens, fantasy sims, academic campuses, and government outreach islands like NASA and NOAA. However, the compositional aspects of design–the narrative construction of experience–are sometimes overlooked in Second Life. I posit that when a user fails to make sense of how or why something behaves as it does in Second Life, it is due to a design which does not effectively use metaphors, or take advantage of the system’s unique affordances or features.

Interaction criticism adds a practical and reflective component to activity theory and activity-based design–through prototyping, evaluation, and then expert judgement of a system’s aesthetic, emotional, and experiential impact. Rather than approach these highly interpretive aspects of design with academic analysis, interaction criticism advocates an analysis based on a critic’s expertise, which parallels the skills of art or literary critics. As Bardzell & Bardzell perceive interaction criticism, design is an inherently communicative act; thus, the evaluation of traditionally rhetorical concepts as part of the critical process is appropriate.

Methodology

This experiment required a single user to visit a library in Second Life and meet a simple objective–in this instance, to take a screen capture of the user’s avatar reading a specific book. Audio capture was taken using Apple’s Garageband; during analysis the researcher used Ambrosia’s WireTap Studio to mark instances of increasing frustration. The researcher lists the following characteristics of frustration:

1. Verbal cues: “I don’t get it” or expressed lack of understanding.
2. Auditory cues: sighs, nervous laughs, or other vocal stops.
3. Behavioral cues: lengthy (in excess of 30 seconds) breaks in communication or interaction

For the purposes of this study, frustration directly related to hardware or network problems–notably “rezzing” issues–are excluded from data analysis.

Results

The task instructions are to take a picture of the avatar sitting and reading a book at a library. The subject interprets this as a parallel set of tasks to completing it in the real world: find the book, then return to a seat and open it. In actuality, the task is much simpler: since the chair is scripted, then the subject need only sit in the chair to read the book automatically. As a result, the subject needed approximately 18 minutes to complete the task; the overwhelming majority of that time was spent browsing individual titles, though other objects provided distraction at various points as well.

Consequently, the subject demonstrates progressive frustration and uncertainty throughout the exercise. At several points, the subject expresses her confusion or uncertainty about a given task, and as her annoyance becomes more pronounced, she goes so far as to say “well, that was misleading” as a criticism of the task’s instructions. Additionally, the subject began to sigh impatiently, use more vocal stops such as “umm” or “well,” and pauses in interaction became increasingly common, and of longer duration.

Eventually, with some additional guidance from the instructor, the subject discovers the instructions for use of the scripted chair, and begins to leverage her prior experience with Second Life to properly equip her avatar and complete the task. Repeatedly, the subject expresses annoyance, though this seems interchangeably aimed at either the system itself, the design of the task, and herself.

These figures display markers the researcher added to the audio recording, which identify instances of different vocal cues or expressions by the subject throughout the task. Clearly, these expressions become increasingly frequent as the task progresses and the subject begins to experience difficulty completing the task.

Of note, two clear expectation violations are present within the recording–one positive violation and one negative one. The subject expressed pleasant surprise at discovering she could move her cursor over titles to get information; while this is common system behavior in Second Life, sometimes it is unreliable due to either network connectivity or the designer’s intent. The other expectation violation pertains to the discovery that an avatar is actually an automated service; since the subject initially identified the avatar as an individual, the discovery that she was, in fact, “a robot” resulted in modest confusion and mild disappointment.

Several times throughout the exercise, the subject expressed a wish for a catalog to locate books with. The subject expressed frustration at being unable to search for an object’s location efficiently, and wished to have an index that would facilitate locating materials. The lack of efficiency in discovering objects, coupled with Second Life’s slow response, led to increasing frustration on the part of the subject.

Discussion

Given the problems the subject experienced completing the exercise, and the clear miscommunication between user, experiment designer, and environment designer, some potential solutions and alternatives present themselves. These solutions may be undertaken by either the user-designer, by Second Life’s engineers, or by a standards-based collaboration between both entities. First, designers should leverage the affordances unique and already present within Second Life. Second Life allows objects to more or less advertise their function via scripts; indeed, floating text which encourages users to “take a copy” of an object, “donate money” or “Sit here” are common. The immediate problem might have been resolved with “sit and read!” floating over the chair in the task. In a greater sense, Second Life’s engineers at Linden Labs need to develop a standard means of knowing how to interact with an object without “touching” it, similar to the cursor/color change which indicates a clickable URL in a web page. While Second Life is partially there–text is often used as an indicator, and when mousing over specific objects such as chairs the cursor changes–perhaps a more immediately obvious scheme should be used.

The inherent difficulty in SL is that most objects can be manipulated in some way or another; the question is bridging the link between manipulation and completing specific tasks with the object. Second Life is far from perfect in this instance; while the author can reach down, pick up a scarf, and put it around her, her Second Life avatar must first pick up the object, put it in her inventory, take it from her inventory and drop it on her avatar. These steps are not intuitive to novice or even necessarily to experienced users, and the process must become more streamlined.

Another option is to improve access with search techniques. Second Life’s search functionality is extremely limited, and

Figure 4. In this mockup of a local search interface for Second Life, search results and terms are highlighted visually.

when seeking information from an inanimate artifact–such as reading text on a poster, or looking for an object–the user is restricted to time-consuming visual browsing. A catalog–such as the one posited by the subject–can prove useful, but without specific location coordinates included in the database, may have only limited appeal. An alternative solution would be an “environmental search” that leverages existing object metadata–with a simple text field, users could complete Google-like keyword searches to find objects located nearby. A mock-up of this display is included here.

This design lends itself well to Second Life, since not only does it introduce a new quick searching mechanism, but a bare bones keyword search interface is already familiar to users who have already used a web search engine. By using a visual indicator–like a glowing outline around relevant items–and maintaining transparency, users suddenly gain new abilities to quickly find what they are looking for, without systematic visual inspection of the screen before them.

However, is a totally predictable virtual environment necessarily desireable? Wright et. al. discusses the benefit of enchantment in design–while enchantment is an immeasurable quality, designers still aim to inspire it in their products. This is especially clear in the work Second Life creators produce, ranging from extraordinary apparel to astonishing landscape objects. If users are not forced by systemic design to explore the environment, and instead can take a shortcut to find what they think they need, is the opportunity for enchantment lost? These are questions outside the realm of this paper, and potentially beyond academic study.

Conclusion

Summary

Ultimately, this case study demonstrates some of the inherent problems of a system which sometimes hides the intent or functionality of an artifact, and highlights the potential for a poorly designed system to create frustration in a user. Some potential solutions, such as a local search display or development of uniform object standards might alleviate the situation and go far to retain users and improve the overall experience of navigating Second Life. Incorporating theoretical elements of design as part of the Second Life creation process, however impossible it is to enforce, would go far towards improving user comprehension as well.

Areas for Future Research

Clearly, an excellent area for future research is presented by the search prototype. Future research should focus on building a prototype or functional mockup, testing with a limited group of users, and evaluating the results, time a user spends searching versus exploring, and whether the search interface has any effect upon frustration levels of a given task. Moreover, search features should be explored and tested as well: should searching be limited to keywords? What metadata should be included in searching? What effect will search interfaces have upon designers–will designers begin to use targeted keywords in their content, in order to rise in search ranks? All these questions are worth pursuing by future researchers.

Additionally, future research should continue to develop a framework for interaction criticism for evaluation of interface aesthetics and appeal. To do so, communication and rhetorical theory might be a starting point for future researchers, especially if one approaches interface design as a communication medium between designer and user. Some rhetorical techniques and communication assessment methods are already present within HCI, therefore using one to enrich the critical language of the other should be possible.

Value of Study

On a personal note, using Second Life as a virtual world study arena gave me the opportunity to being exploring user interface design from a theoretical and pragmatic perspective, with relatively low barriers to entry. Since the interface itself has significant room for improvement, identifying why elements don’t work, and how other users interpret the interface is a useful critical experience–proposing solutions to these problems is even more valuable. On the other hand, my interaction with other Second Life citizens outside my small community has been very limited, and the communication theorist in me would love to conduct a more detailed study of communication patterns between users, as well as work on true experiment design.