We hope to identify and enumerate the considerations that will most help designers choose where along the spectrum they want their systems to fall. For example, when is it appropriate for a user to trust the devices in a multi-device environment? Seamless interaction presumes that users trust devices completely with their information. However, that trust may not be justified; devices may capture information and use it maliciously. In practice users may want to modify the components, content, and behavior of interfaces when interacting across multiple devices. Security is closely related to the issue of trust; the more seamless an interface, the greater the risk to the security of the user's device (and by extension the user's information).

The development of a good set of usage scenarios will also help to create a specification or requirement for the technological support needed to support these systems as well. For example, how can a system support seamless connection and association when different devices have different form factors of input, display output, network capabilities? What technologies are available today, and what technologies need to be developed? More generally we wish to address the question of whether we can design and build a gradient of levels of seamful/seamless systems and identify gaps in today's technologies that prevent, or at least make it difficult, to support the range of possibilities along the spectrum. The following two sample scenarios are meant as starting points for our discussions, to give prospective participants an idea of the domains and issues we hope to explore. Part of the workshop goals are to develop a broader set of examples, and to make their design choices explicit. Both should aid future development ubiquitous computing environments, systems and applications.

Sample Scenario 1: PDA Interactions with a Ubiquitous Environment

Imagine a ubiquitous computing environment in which all the surfaces are interactive: walls, tables, kiosks, windows and so forth. As a user walks around in such an interaction space with her PDA, she can use it as an input device to interact with other devices and content in the environment. Should the (ubiquitous) system keep track of which other interactive surfaces she is close to? As she walks up close to a table, should the associated connection between her PDA and a device in the environment, say a table be made automatically for her so that a swipe of her stylus on her PDA on an image will "move" the image onto the table? Or should there be an explicit action on her part to initiate a "linking" between her PDA and the table? Each approach has implications for privacy, awareness and interaction. After that, should her PDA stylus input be redirected onto the tabletop? When she "leaves" the table (walks away or powers off her device), should the table automatically "wipe" away here content, leaving no indication or memory of her being ever there? Or should she make a conscious decision as to whether to leave or delete her content, and take an explicit action to do so? The choices a designer makes in the above scenario will create and support different usage domains, user experience and applications.

In this example, we once again have a multi-device ubiquitous computing environment. This time, however, the user wishes to access her own content ­ she has her cell phone with her email on it. Although she can interact with her email on her own device, she wishes to exploit one of the devices available in the space. She might want to divide the interface to her email client across her cell phone (a trusted device) and a device in the local environment in different ways depending on how much she trusts the other device. At home or in her office, for example, she could transfer it completely, as she fully trusts the other device. In a more public setting that she (mostly) trusted, she might transfer the inbox, but use her cell phone to approve requests (e.g., read email, compose email). A number of other possibilities exist; she could initiate transfer of a single message from the cell phone, read and reply on the other device, allow composing messages on the other device, but not transfer headers and remove all contact information from the body if replying to an existing message, and so on. Another consideration is the visibility of the other device she wishes to use. Is it a desktop (traditionally a single-user device) or a larger (and more public) wall display? She may choose to display different content based on the size and accessibility of the other device and its display. The interactions here represent a more seamful interface, in which the user initiates the division of the interface based on the environment at hand. single-user device) or a larger (and more public) wall display? She may choose to display different content based on the size and accessibility of the other device and its display. The interactions here represent a more seamful interface, in which the user initiates the division of the interface based on the environment at hand.