Markus Latzina, Joerg Beringer
The term transformative has different connotations, but it usually refers to the concept of significant, irreversible change, often considered game changing due to its ability to redefine an entire value system. Recently we have observed new technologies that are deemed to be transformative, for example, by enabling us to maintain relationships remotely via social networks, access information anytime anywhere via mobile devices, or obtain search results instantly.
What has not yet become transformative is the user interface itself. The fundamental design rationale for designing applications goes back to the 1980s, when mainframes were replaced by personal computers and PC software was packaged into applications. Since then, the concept of a software application packaged at design time with proper functionality has dominated the design of end-user software. In this approach, each application comes with its own data model, and the user interface displays these data structures as application content. Data objects presented at the user-interface level are designed to match the user's mental model, but this also ties them to the semantics defined by the application scope.
The application scope is typically defined by a specific set of use cases identified prior to implementing the application. Generally, system design is understood as a solution to a well-defined set of requirements. Requirements derived from an agreed-upon set of use cases are mapped to features . This design rationale implicitly creates hard boundaries with respect to its coverage of use scenarios and ability to consume application content.
A software application design is considered to be usable if it enables people to perform the identified set of tasks and matches their mental model. However, in traditional software solutions, the task models inform only the design and are implicit at runtime. Transformative User Experience frames contexts in terms of spaces of interaction potentials where users can realize their current goals by moving across various task contexts along self-determined transformative vectors. This approach requires a much more explicit presence of task models in a product's runtimenot just at design time. To allow users to create a proper task setting, a system must provide mechanisms to detach application content and move it into different task contexts. Software applications often lack this concept of elasticity, primarily because applications are designed for specific tasks. But in reality, task contexts are often not mechanistic; rather, they grow organically as needed in a given situation . Task flows do not always follow the typical task life cycle, alternating between non-routine and routine situations and materializing as an idiosyncratic one-time practice. We therefore see a need to identify principles of designing for elasticity to support seamless transitions across contexts rather than forcing users to switch between isolated packaged applications or parts thereof.
Instead of designing for many discrete applications, the Transformative User Experience approach aims to natively support a larger variety of task flows by replacing application boundaries with elastic, situational environments that allow transitions between different task states. Imagine businesspeople who work collaboratively on a large display to discuss business issues and make decisions (see Figure 1). This display must be able to surface relevant content. During the discussion, content may be moved, clustered, annotated, or synthesized to analyze information and capture insights. Areas on the display might represent certain task contexts typical for knowledge-intensive work, such as prioritizing, querying, inspecting, and displaying analytical information.
|Figure 1. Knowledge work on a large display (Source: SAP).|
From this example, it should become obvious that content, application, and device need to be decoupled as much as possible to allow users to focus on information without being confined to a particular pre-packaged application context. With the continued proliferation of cloud computing and virtualization of data persistency, content will be increasingly decoupled from containers and accessible independent of a particular application. This requires the definition of new principles of how data objects (content) interact with their context (container) and how this flexibility can be leveraged to enable end users to compose and transform content as required in an actual task situation [3,4,5]. When content is decoupled from the container (application, device, software stack), there is an opportunity to design natural, task-motivated new environments in which multiple capabilities and data coexist and can be put in contextual relationship at use time rather than at design time.
This runtime flexibility challenges the traditional application design, where content not only resides in different repositories but also is surfaced within rather confined application stacks, making it difficult to be consumed within a situational, task-oriented context (see Figure 2). As content becomes more pervasive, user interfaces will increasingly emerge at runtime from the actions of users. This trend requires new standards for interchangeable content and the modeling of task contexts, since neither will come as fixed ingredients in traditionally designed applications.
|Figure 2. Decoupling of capability, content, and technical container.|
Our goal is to develop a repeatable design approach for this kind of transformative user experience and establish a framework for a novel system architecture that makes this quality a default for next-generation products. This change in design thinking is also reflected in the emergence of new disciplines within HCI that are postulating a more balanced relationship between users and systems, with respect to adaptation and distribution of control:
- Appropriation. With increasing mobility and the pervasive use of software in social contexts, the application is more likely to be exposed to unexpected contexts of use. Requirements engineering and design validation therefore need to look beyond the core use cases defined for a product and continuously adjust to new and unexpected usages of a product [1,6,7,8].
- Meta design. In meta design, a product is designed for continuous change, either explicitly, by engaging users (using approaches characteristic of open source communities or resembling end-user development), or implicitly, by admitting ways that facilitate adaptation in the course of continuous use and according to the dynamics of the socio-technical ecosystem [9,10].
- End-user development. Today's user is a generative consumer, a prosumer who is both consuming and creating content. To support this type of user, the use time (application at runtime) needs to provide generative capabilities that are commonly found in design times. In such cases, the boundary between design time and use time may blur altogether.
- Organic design rationale . Gesture-based direct manipulation of work artifacts and open-ended navigation within social or information networks are becoming a prominent user interface paradigm in which navigation and manipulation are based on natural spatial metaphors (e.g., deep zooming or ostensive navigation) .
- Distributed cognition focuses on the cooperation between systems and between systems and their environment. The system does not simply follow the user's input; rather, the user and the system are part of a cooperative network. The interaction between them is bidirectional, and contexts are shared.
Encouraged by these novel conceptions of system behavior and the relationship between system and user in general, we believe the design of static user interfaces for applications must be complemented by a more organic and flexible approach that enables users to access content and move it across the boundaries of an application. We call this approach Transformative User Experience, since it features a set of repeatable system behaviors and design principles for building adaptive user interfaces that emerge over time in symbiosis with the task flow of the user. We describe technical system behavior and introduce a new design rationale for creating an information architecture that develops bridges between operating-system level and purposed applications.
We are developing principles of how to construct transformative user interfaces that embody aspects of appropriation and meta design as intrinsic capabilities to support task life cycles in a much more fluid and situational way. The resulting architecture explicitly acknowledges the existence of control and expertise outside the system and scales for designing next-generation systems that contextualize content within a hybrid cooperative fabric of multiple applications, devices, and users.
To move information from one context to another and to make contexts elastic to match situational task conditions, Transformative User Experience defines two key concepts: a container representing task context and a content item representing a data entity. So that content items may be presented in any container, the container can query the content item for information on how to present it. Each content item provides data about the present item, but not the rendering itself, which makes the item representation agnostic to the technical aspects of the container. Containers can select between a variety of archetypical presentation styles, such as data point, line item, or business-card views. To further adapt content items to the local task semantics of the container, the container can treat the content item as a generic item and add enablements for contextual behaviors. This object casting allows the container to convert a generic system object or a local application object into an object of use that is contextually appropriate with respect to the semantics of the container.
Dourish illustrates the difference between a system object and an object of use with the example of a document that is primarily a generic system object owned by a generic text-processing application . But when used in a personal context, it becomes meeting minutes, a patent application, or a contract, depending on the use. In comparison, our focus is on the context as defined by the situational task rather than on the personal context.
In our approach we combine cognitive theories of action planning and task accomplishment with semantic approaches that introduce frames for expressing situational semantics. Our ideas are also grounded in the behavior setting theory of Roger Barker , who observed in many empirical studies that behavioral patterns depend on environmental settings and that the meaning of objects in those environments changes accordingly . With the Transformative User Experience approach, we aim at enabling the same type of interaction pattern with respect to the technical container, resembling the behavior setting, and the content item, resembling the behavioral pattern.
In the large-display scenario, the container is initially empty, and as such its task context is undefined. Let us assume that users in front of the display are discussing a new organizational structure and begin to pull up names of managers who could play a major role in a new business. At this point, the display becomes a collection of content items (people) and is still not tailored to a specific domain. When a user adds an org-chart view and begins to sketch org units, the context of the large display becomes focused on organizational structures. People's names within this context are potential candidates for manager positions, and users can annotate pros and cons for each nominee. The large display then turns into a planning tool and the container context becomes an organizational structure, with managers and organizational units as the content items. When several users create different variants of the org chart, the org chart itself becomes a decision option (content item) within a larger decision-making context. When the discussion is wrapping up, follow-up actions are defined and users agree to get in touch with some of the proposed managers. Now the manager object has assumed the role of a to-do item for the respective task owner, which can be accepted, rejected, and signed off as completed.
Today an explicit design rationale for systematically constructing such an organic, open-ended interaction framework has not been articulated. Transformative User Experience attempts to identify the principles of how to construct transformative interfaces in a native way to empower the user to incrementally establish new task contexts with intended meanings within the system itself. While our principles for content composition resemble the classic drag-and-drop principles of direct manipulation interfaces, Transformative User Experience aims at leveraging the interaction between content and container in ways that acknowledge the potentials for semantic plasticity of both.
We believe that Transformative User Experience is a promising conception for bridging the omnipresent chasm between the levels of content-agnostic operating systems on the one hand, and packaged applications on the other (see Figure 3).
|Figure 3. Transcending the chasm between operating system and packaged applications.|
An integrated work environment for knowledge workers requires more than generic file-storage systems and tool palettes that launch packaged applications. The seamless transition between searching, collecting, using, and transforming content is what characterizes knowledge work. Currently, an intermediate layer that supports goal-oriented, lightweight situational actions in a more fluid way than today's application paradigm is missing. This layer requires a new design rationale suitable to augmenting knowledge work and would create a symbiosis between system and users in which the system supports the emerging intentions of users.
Instead of launching packaged applications and navigating along static screens, users should be able to navigate in an ostensive manner between content, context, and applications, and to create content and contexts at runtime. When the user can actively transform content and create task contexts and the user interface is no longer restricted to a priori packaged applications, the difference between use time and design time begins to vanish: Design time becomes use time.
3. Sikka, V. Timeless software. 2010; http://vishal-sikka.blogspot.com/2008/10/timeless-sottware.html
4. Schneyer, M. Containers and devices, and how to decouple the right way. 2010; http://markschneyer.com/2010/10/containers-and-devices-and-how-to-decouple-the-right-way
5. Melichar, C. Content vs. container. 2006; http://www.intermediablog.com/2006/12/content_vs_cont.html
7. Salovaara, A. Studying appropriation of everyday technologies: A cognitive approach. Proc. of the 27th International Conference. Extended Abstracts on Human Factors in Computing Systems. ACM, New York, 2009, 31413144.
11. Werner, H., Latzina, M., and Brade, M. SymbikA new medium for collaborative knowledge-intensive work. Proc. of the International Conference on Education, Informatics, and Cybernetics. International Institute of Informatics and Systemics, Winter Garden, FL, to be published.
Markus Latzina is a senior UX architect at SAP. His aspirations are oriented toward developing a generative design theory by adopting a praxeological, humanistic stance that puts designful thinking at its heart. He holds a master's degree in psychology from the University of Tübingen and completed graduate studies at Sonoma State University with a Fulbright scholarship.
Joerg Beringer is VP of New Product Design at SAP Labs in Palo Alto, CA. As a design entrepreneur and thought leader he contributed to establish innovative products for knowledge-intensive work and collaboration at the workplace. His current focus is combining aspects of business intelligence and personal productivity to define new product experiences that form a symbiosis between business and personal needs.
©2012 ACM 1072-5220/12/0300 $10.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2012 ACM, Inc.