5. Designing for Artifact Ecologies
As more people have access to an increasing number of
heterogenous devices, it becomes an important
consideration for interaction design.
A designer needs to take into account:
● The available hardware and software.
● The way interactions can be distributed across devices,
users, and environments.
● People’s perception.
8. Distributed User Interfaces
A DUI is any application interface that can be
distributed across different displays, devices,
and users engaged in co-located or remote
collaboration (Melchior, 2011).
11. How DUIs are Designed?
The predominant approach is genius design, where a small team
produces an artifact based on goals and requirements derived from
literature and occasional user studies. The user experience or usability
is usually not evaluated.
Inspiration often comes from observing how physical tools are used
and how they can be combined to support specific activities.
The main theoretical frameworks used are Proxemics Theory and
Activity-Based Computing.
12. Examples of proxemic interactions from
Marquardt et al. (2012) Activity-Based Computing system proposed by Bardram et al. (2012)
15. Challenges with Designing DUIs
The concept of distributed interactions is fuzzy, hard to grasp without
reading a significant amount of domain-specific literature, and is difficult
to apply in design.
There is a need for a well-established vocabulary, which is
supplemented by a set of examples and possible usage scenarios.
There is a need to find ways for introducing people to technical issues,
especially in situations, when they are not properly equipped to discuss
or relate to them.
16. Questions
How to actually design DUIs?
How to ensure a higher probability that the
designed artifact will resonate with users’
assumptions and will be appropriated by them?
17.
18. Research Goal
The aim is to provide interaction designers with:
● A set of options to be used in the design of distributed user
interfaces.
● A clear rationale to be able to choose among those options
and understand the implications of the choices made.
19. Possible Solution
The result could take the shape of a decision support
system for designing DUIs.
In addition, concrete examples need to be provided to
illustrate how this support system could be used to inform
the design of a real-life service.
Specifically, this could be realized through a collection of
design patterns.
20. DUI Design Patterns
Design patterns are a means for laymen to acquire a vocabulary that would
help them express and communicate their ideas (Borchers, 2000a).
A design pattern aims to present “a proven solution to a recurring design
problem” in a format that is easy to understand and can help generate new
ideas (Borchers, 2000b).
Individual patterns can be assembled into pattern languages, which have been
successfully used in architecture and software engineering as a means of
communicating design knowledge (Borchers, 2001).
27. DIxD Design Pattern Structure Human-Artifact Model Borchers’ Design Pattern Structure
Design motivation Why would the artifact be used? Problem statement + forces
Design goal Why would the artifact be used? Problem statement + forces
Setting Why would the artifact be used? Solution
Summary What can be done with the artifact? Solution
Examples What can be done with the artifact? Examples + Illustrations
Description How should the artifact be used? Solution
Enabling technology How can the artifact be operated? Solution
Diagram How can the artifact be operated? Diagram
Theory - Solution
References - References (implicit)
- - Ranking - not used (only later if we develop
a way to establish the relevance of each
design patterns OR if we are able to
compare competing solutions)
- - Context - not used (only later if our work
evolves into a pattern language)
28.
29.
30.
31.
32.
33.
34. Future Work
● [ST] Clarify the concepts and relationships in the DIxD
pattern language.
● [ST] Assess the expressiveness of the patterns language by
using it to analyze existing DUIs.
● [ST] Design a mobile DUI based on the DIxD pattern
language.
● [MT] Assess the relevance of the DIxD pattern language for
interaction designers.
● [MT] Explore how people interact with DUIs designed with
DIxD patterns and improve the language based on that.
35. References
● Melchior, J. (2011). Distributed user interfaces in space and time (p. 311). Presented at the Proceedings of the 3rd ACM SIGCHI
symposium on Engineering interactive computing systems - EICS '11, New York, New York, USA: ACM Press.
http://doi.org/10.1145/1996461.1996544
● Elmqvist, N. (2011). Distributed User Interfaces: State of the Art. In J. A. Gallud, R. Tesoriero, & V. M. R. Penichet, Distributed User
Interfaces (pp. 1–12). London: Springer London. http://doi.org/10.1007/978-1-4471-2271-5_1
● Rädle, R., Jetter, H.-C., Marquardt, N., Reiterer, H., & Rogers, Y. (2014). Demonstrating HuddleLamp: Spatially-Aware Mobile Displays
for Ad-hoc Around-the-Table Collaboration. the Ninth ACM International Conference (pp. 435–438). New York, New York, USA: ACM.
http://doi.org/10.1145/2669485.2676584
● Wroblewski, L. (2010). Touch Gesture Reference Guide. Retrieved from http://www.lukew.com/ff/entry.asp?1071
● Marquardt, N., Hinckley, K., & Greenberg, S. (2012). Cross-device interaction via micro-mobility and f-formations (p. 13). Presented at
the Proceedings of the 25th annual ACM symposium on User interface software and technology - UIST '12, New York, New York, USA:
ACM Press. http://doi.org/10.1145/2380116.2380121
● Bardram, J. E., Houben, S., Nielsen, S., & Gueddana, S. (2012). The Design and Architecture of ReticularSpaces – an Activity-Based
Computing Framework for Distributed and Collaborative SmartSpaces (p. 269). Presented at the Proceedings of the 4th ACM SIGCHI
symposium on Engineering interactive computing systems - EICS '12, New York, New York, USA: ACM Press.
http://doi.org/10.1145/2305484.2305529
● Borchers, J. O. (2000a). A pattern approach to interaction design. Presented at the Proceedings of the conference on Designing
interactive systems processes, practices, methods, and techniques - DIS '00, New York, New York, USA.
http://doi.org/10.1145/347642.347795
● Borchers, J. O. (2000b). CHI meets PLoP: An Interaction Patterns Workshop. ACM SIGCHI Bulletin, 32(1), 9–12.
http://doi.org/10.1145/333329.333330
● Borchers, J. (2001). A Pattern Approach to Interaction Design. Wiley. Retrieved from http://www.amazon.com/dp/0471498289/
● Mendel, J. (2012). A taxonomy of models used in the design process. Interactions, 19(1), 81. http://doi.org/10.1145/2065327.2065343
Editor's Notes
This diagram illustrates the evolution and narrowing down of my research topic. While initially it was quite broad, focusing on interaction design for mobile devices, it was eventually narrowed down to supporting the design of distributed user interfaces, specifically providing scaffolding for the envisioning stage of the design process. This narrowing down occurred mainly due to my work on exploring how people interact with ecologies of artifacts, as well as conducting a review of literature on the design of distributed user interfaces.
Having access to a wide range of devices is becoming more commonplace. A regular person might already be carrying a smartphone and a tablet or laptop, and a smartwatch on their wrist. These devices are always connected and travel everywhere their owner does. This puts pressure on interaction designers to change the ways digital artifacts are conceived.
For this reason it is no longer sufficient to design applications that work in isolation. A classical example of the shortcomings of this approach is using a messaging service, which rings all devices simultaneously when a new message is received. In an intense conversion this can become a great source of irritation for the user. Other issues lie in the area of data transfer. This is already being addressed by existing cloud storage services and interface migration, exemplified by Apple’s Continuity features for iOS and OS X and Google’s Chromecast, but clearly much more can be done in this area.
We propose to cluster existing research in the field of HCI in the domain of cross-device service design into 3 focus areas. Pervasive computing is seen as a domain focusing on the provision of hardware and software infrastructure to enable anytime anywhere access to information. Distributed computing focuses on designing interactions that span multiple devices, users, and environments. Ubiquitous computing focuses on the factors contributing to the perception of anytime anywhere information access and builds on contributions from both pervasive and distributed computing.
We acknowledge that the term “distributed computing” has an established definition in the computer science field and that our way of appropriating the concept does not necessarily match the original meaning. However, we use it here for the lack of a better concept.
This research focuses specifically on the distributed computing layer, working with the existing pervasive infrastructure, while only slightly touching the perception part.
One way of making the distributed computing concept tangible is through the design of distributed user interfaces (DUIs). The domain of DUI design is fairly young compared to other domains of HCI. A recently conducted literature review suggests that active research has begun around 2008-2009, which might be explained by the relevant technology becoming cheaper and more readily available.
Following is a definition of a DUI by Melchior.
Another definition is proposed by Elmqvist in his survey of the state of the art of DUIs. Unlike Melchior, Elmqvist focuses on the technological aspects of UI distribution, deliberately ignoring users.
It is curious to note that both definitions were proposed in 2011, suggesting that researchers have started making attempts to clarify the concept only fairly recently.
Following is an example of a functional DUI from the HuddleLamp project, published in 2014. The prototype allows users to employ multiple heterogeneous devices to create an integrated working space, where individual devices can serve as pieces of a larger screen or as hosts for specific pieces of content that the user assigns to them.
A recently conducted literature review aimed to provide insights into how DUIs are designed. The analysis was done based on a sample of 92 papers describing fully implemented DUI prototypes.
Here examples of prototypes based on Proxemics Theory, specifically F-formations and micro-mobility, and Activity-Based Computing are demonstrated. The examples of proxemic interactions illustrate how DUIs can be built to support collaboration scenarios in small groups, considering the ways people position themselves relative to each other and how objects are oriented and repositioned in the physical space to let them be fully viewed, partially viewed, or hidden from others.
The ABC example demonstrates what a Multi-Display Environment could look like and how an activity-centric UI could be designed for it. This interface represents the various actions corresponding to an ongoing activity, as well as relevant resources and participants.
The literature review suggested that research outcomes in the domain of DUI design can be categorized as mainly abstract or tangible. Between these a gap can be seen, because it is not a trivial task to go from a conceptual understanding of what a designer wants to do to a tangible artifact. Although there are examples of how DUIs are created is small teams and how challenges are solved on a case by case basis, there is a lack of support for designers for making this transition.
This notion is further supported by retrospective interviews conducted with members of the LearnMix team. The design of distributed interactions plays a core role in the project, yet it turned out to be quite challenging to appropriate.
The insights from the literature review and the retrospective interviews can be summarized as follows.
To guide further research 2 questions can be formulated.
Or in a nutshell how to help interaction designers bridge the gap between abstract ideas and tangible artifacts. The challenge then is to bridge the gap between what is already known about distributed user interfaces and what still needs to be known to actually design them.
This diagram demonstrates a pattern language developed to support the design of interactive tabletops as described by Remy et al. A pattern language should have a hierarchical structure. Basic or micro-patterns are positioned on the top and combinations and derivations further down. The items on different levels of the language should be semantically grouped together.
Here is another example of a pattern language for touch gestures. In this language gestures are aggregated into various groups according to user actions they enable or platforms that support them.
This is an example of how the touch patterns could be used to describe a sequence of actions a user would need to take while interacting with a tablet application designed for the LearnMix project. The purpose of the diagram is to demonstrate how individual patterns can be used as verbs that are combined into sentences describing how a certain interaction would occur.
The foundation of the pattern language we are currently working on came from the literature review on DUI design, where a set of 35 techniques used for interacting with DUIs was compiled. This Venn diagram illustrates how the interaction techniques can be distributed across different usage settings. A private setting would foresee using a device individually. A semi-private setting would foresee sharing a device with another person or a small group. A public setting would foresee using a device to display information to a large group. The overlapping areas include techniques that allow people to combine various types of usage. For example, one display could be used to show public information, but a private display could be used by an individual to show additional pieces of information from the public display, which are meant for private use.
The theoretical foundation of this work is based on the Human-Artifact Model, which has its roots in Activity Theoretical HCI and is the result of research by Susanne Bødker’s team. The Model considers the dialectic relationship between the assumptions of designers and users on 3 levels of Activity Theory: activities, actions, and operations. Analysis based on the model allows a researcher to better understand people’s motivations, goals, and the specific ways in which activities are operationalized through artifact usage. The Model is specifically meant to be used in the context of artifact ecologies, which refer to all digital artifacts a person has access to and uses to support her activities.
A structure of a design pattern proposed by Borchers was also used to inform the process of converting individual DUI interaction techniques to DIxD patterns. Please see the corresponding paper for the meaning of each of the components.
The following table demonstrates an attempt to map the structure of a DIxD pattern to the Human-Artifact Model and the components proposed by Borchers. The values for the design motivation, design goals, and setting come from surveyed literature. The enabling technologies originate from a recently proposed ubiquitous computing design space.
The sub-components of these elements are further elaborated on this mind map.
However, we reached a conclusion that the components of our DIxD pattern are too detailed and it would be beneficial to use more abstract concepts. For this reason we built on the distribution dimensions proposed by Elmqvist (2011) for the How? level. For the Why? level we tried to cluster the various design motivations and goals into higher-level categories. These categories still need to be clarified before they can be reflected in the DIxD pattern structure.
An example of what an individual DIxD pattern looks like is shown here. The design patterns are aggregated in a semantic MediaWiki to make it possible to create meaningful connections between them.
The next step is to design semantic queries that will allow designers to search for patterns that are relevant for their design problems, for example designing interactions for private and public settings, augmenting existing practices, or improving information management between devices.
Further, this diagram demonstrates our attempt to compile the collected design patterns into a pattern language. Here patterns are assembled across several levels. The first level includes micro-patterns, which are the smallest units of interaction that can no longer be broken down into smaller components. The second level includes slight variations of the original micro-patterns. The third level includes patterns that are combinations of several micro-patterns. These are aggregated under the label “1st Generation”, because these are the patterns that have published in literature for the first time. All remaining generations build on the initial micro-patterns, their derivatives, as well as subsequent generations.
In addition, the patterns in the diagram are color-coded to highlight that they belong to the same family. Patterns with the same color relate to each other and can be considered variations and improvements to the same pattern.
So returning to the issue of bridging the gap between abstract design concepts and tangible DUIs a more complete picture can emerge.
Here the double diamond concept is used to illustrate how we see a designer moving from an initial challenge statement to deliverable. This would entail going through the stages of discovery, reframing, envisioning, and creation, as described by Mendel (2012). Our contribution focuses on the envisionment stage, helping the designer to transition from scenarios and user stories to prototypes by exploring the various options for DUI design based on our design space and pattern language. This diagram also illustrates how this research is different from existing work surveyed in the literature review and what is discussed by Elmqvist (2011) as these works focus mainly on creation.