What distinguishes modern humankind is our collective ability to build more complex tools and communities. In previous eras, these amplified muscle power. In the last half century, a new kind of tool has emerged-information technology. Its impact on society is now only dimly understood. We will explore the future of information technology by creating it and living in it, within the EECS Department, the Berkeley campus, the City of Berkeley, and beyond. Not only will be develop innovative, new technology, we will ap-ply expertise in human-centered systems to evaluate how well that technology lerages and enhances human interaction and intellectual activity.
Our expedition is called Endeavour , named for the ship Captain Cook sailed on his explorations of the Pacific. Cook's lasting contribution was a comprehensive knowledge of the great unknown of his day--the lands, resources, peoples, customs, and ideas across the sea. The sea is a particularly poignant metaphor, as it interconnects much of the world. We envision "fluid" information systems that are everywhere and always there, with components that "flow" through the infrastructure, "shape" themselves to adapt to their usage, and cooperate on the task at hand. We seek to develop a pervasive Informa-tion Utility, based on a new technology of fluid systems, enabling new approaches for problem solving and learning.
In the future, interconnected devices will be so commonplace that "the internet" becomes invisible. Devices span from the so tiny that the computer disappears, to servers so large that storage limits vanish. It will be possible to track and relate everything. The vastness of captured information shifts the management focus from simple queries to relation-ships, relevance, and flow. Such a system will enable pervasive, human-centered interaction, not just with information, but with expertise. Independent of place or time, anyone could access the combined work and collective intelligence of potentially unknown collaborators, dramatically reducing the effort to solve a problem or learn something new. This is achieved by an active system, working on behalf of the individual, gathering, accepting, filtering, aggregating, and organizing information, while protecting the owner's information assets.
Let's consider lifelong learning. A computer scientist, let's call her Cook, wishes to learn a new field like molecular biology. Imagine that she asks the question: "what have others like me done to learn about microbiology?" The system matches others with a compara-ble profile and shares with her their approach to learn this subject. Initially, it provides her the survey articles they read and the books they used. As she progresses, the system suggests more advanced texts. Her success in understanding, monitored by the system, leads her to the next level. Things she understands are presented quickly; difficult concepts are enforced with more readings, background materials, and discussions with a live tutor over the network. By presenting just the right materials, at just the right time, to just the right level of detail for her increasing expertise, Cook develops her desired level of knowledge in molecular biology. Her own endeavor further opens the route to others.
Our approach centers on enhancing the endeavor of understanding, while evaluating our technical developments by how well they amplify and leverage human intellect.
Extrapolating existing trends, desktop and server systems will have greater capacity, devices will become more diverse, and information management will be better integrated. But a larger fraction of our time will be spent pushing information-structuring, organ-izing, and storing the increasing volume within a changing environment. This will be far more expensive to administer and maintain than to build. We must wholly rethink system design: human time and attention, not processing or storage, are the limiting factors.
Our starting assumptions are: (i) a vast diversity of computing devices (PDAs, cameras, displays, sensors, actuators, mobile robots, vehicles), (ii) "unlimited" storage (servers supporting personal storage beyond a terabyte; anything that can be captured and stored in digital form will be), (iii) every computing device is connected roughly in proportion to its "capacity" (small devices with little bandwidth, communicating locally to larger devices with access to increased capacity), (iv) devices are predominantly "compatible," rather than predominantly incompatible (plug and play interoperation enabled by perva-sive technology for on-the-fly translation, emulation, and virtual machines). The only common element is communication. Devices will be so specialized that it will be unusual to have one with an "average" amount of processor, memory, disk, display, input, and connectivity.
Our view of the future demands a quantum change in information technology research: dynamic adaptation, self-organization, and personalization on a truly massive scale. Its scale and grandeur, its rapid evolution and the radical modes of its use, the heterogeneous nature of its component subsystems and their contained information, and the broad activities it supports, make our envisioned Information Utility enormously complex. This can only be managed by the system implicitly organizing its contents based on tacit in-formation extracted from the environment. Enabled by ubiquitous communication and comprehensive interoperation, implicit organization makes the utility more human-centered, by raising the level at which users interact with information.
It is expected that a full expedition will last for six to nine years, organized as two to three exploratory "voyages" of three years each. This first voyage will develop the initial conceptual architecture and proof of concepts in four critical areas: Information Devices, Information Utilities, Applications, and Design Methodology.
Information Devices investigates hardware architectures that integrate MEMS-based sensors, actuators, position locators, and communicators with imagers (cameras), displays, and hand-portable and other mobile computing devices including vehicles and mobile robots. A critical contribution will be to insure these devices are well integrated into the system architecture of the next century.
Information Utility is the core of the proposed research. We will develop the needed technologies to support fluid software: mechanisms for processing, storage, and informa-tion management that "flow" within the system, dividing into component "drops" that choose where to execute and how to access storage anywhere within the interconnected environment. Component software will discover and negotiate interoperation agreements with others in the local and wide-areas, providing cooperative processing. The Utility will also provide mechanisms to accommodate a broad range of input, output, and information access devices. It will implement "always available" processing and administrative scal-ability by exploiting the fluid software paradigm itself.
To ground the Utility architecture, it is crucial to incorporate stressing Applications. Ours include collaborative environments to support high-speed decision making and educa-tion/learning spaces.
Integral to the design effort will be a vigorous activity in Design Methodology:
formal methods for specifying the system, synthesizing it onto its underlying
hardware and soft-ware components, and rigorously evaluating it to insure
its usability by the intended user communities.
The first voyage forms the basis of an evolved architecture, scaled up experiments,
and a larger evaluation effort in the second three year voyage. The third
voyage would consist of a final refined architecture and major testbed deployments
in collaboration with indus-trial partners. This first expedition is organized
into a base program and eight optional tasks. The former focuses on the
broad research challenges described above. The options extend it in a number
of directions. They are strongly interdependent, building on specific aspects
of Information Devices and Utility functionality to enhance our Applications,
thus better demonstrating the latter's ability to amplify human intellect.
The base program and the eight optional tasks can be conceptualized by the
Expedition Cube shown in Figure 1 (Option 8, Scaled Up Field Trials, is
not shown).
The expedition is organized around the main activities of Applications, Information Util-ity, Information Devices, and Design Methodologies. The Base Program and Options are described in more detail in the following sections.
Professor Randy H. Katz will assume the role of Principal Investigator and Expedition Leader. The team brings to bear broad expertise in computing systems, operating sys-tems, CAD, MEMS, and user interface design and evaluation. Faculty investigators will all be involved in the Base Program, and will participate in the optional tasks as follows. A bold "X" indicates the expedition leader for that option.
| Options/Investigator | ||||||||
| Aiken | ||||||||
| Brewer | ||||||||
| Canny | ||||||||
| Culler | ||||||||
| Hellerstein | ||||||||
| Joseph | ||||||||
| Katz | ||||||||
| Kubiatowicz | ||||||||
| Landay | ||||||||
| Newton | ||||||||
| Patterson | ||||||||
| Pister | ||||||||
| Sangiovanni | ||||||||
| Wilensky |
Randy H. Katz, 17 July 1999, randy@cs.Berkeley.edu