That vision and the challenges underlying it were discussed at the annual New Paradigms for Using Computers workshop at IBM's Almaden Research Center. Designers gathered here agreed that crafting a new way to interact with new kinds of smart systems will be the most important work in the next decade in computer design.

"Every revolution in computing has been driven by a fundamental change in the end-user experience," said Robert Morris, director of the R&D complex tucked into rolling brown hills south of San Jose. "The user experience is an integral part of our research activities here, and I think we are on the verge of something new in the user experience with the emergence of pervasive computing."

Dan Russell, recently named to head research here, set the stage of the discussion that focused as much on ethics and aesthetics as technology.

"The problem for us is not computation," said Russell, who acted as host of the event late last month. "Sure, we want to build petaflop computers. But the question is, how do we want to live our lives? When you enter a room, what do you want to happen?

"I want to live in a world where the technology disappears, and the right stuff happens. But the right stuff for me may not be the right stuff for you," he said.

Smart offices
Norbert Streitz, a division manager at the German National Center for Information Technology (GMD), described his group's work in creating two generations of what he termed "roomware." Streitz has helped develop collaborative computing spaces that span 15- x 4-foot wall displays, conference tables with 50-inch plasma displays and tablet-arm chairs with embedded pen-based computers.

Streitz's Darmstadt-based institute has created a software platform called Beach that links these systems with their various-sized displays via a 10-Mbit/second wireless 802.11 LAN. A version of the software dubbed Palm Beach will let data objects be transmitted from Palm Pilot handhelds to displays in walls, tables and chairs via infrared links. That software, still in a beta phase, is expected to be spun out as a commercial product to a German-based startup now being formed.

Wilkhahn, a German furniture maker that has worked with GMD on the projects, could announce plans to commercialize some of this work as early as October at the Orgatec office furniture trade show in Cologne, he said. But business issues lie ahead.

"The office-furniture industry is a medium-sized business, so it is not prone to taking risks," said Streitz. "They are also concerned about how they will handle new things like IT support for such products as well as the need for a retrained sales force."

Some at the conference question whether workers really want to use such products as part of collaborative teams or whether work will continue to be a more solitary process. They also raised issues about security for such collaborative systems, an area Streitz acknowledged he had not yet tried to address.

IBM's Almaden group also has been working on systems for the workplace and demonstrated an interactive bulletin board that would be based on a large-screen plasma display. Users would be able to pull data from their remote workstations and share it on the big screens just by standing near the displays with an RF ID tag, said Russell. The research group will present the concept as a product idea to IBM's commercially oriented pervasive computing group this fall, he said.

The world of pervasive computing will also be populated by digital consumer devices, ranging from digital jewelry to cellular phones for teens.

Design house GVO Inc. (Palo Alto, Calif.) demonstrated a cellular device that lets teenagers pass notes, play MP3 audio files, check movie listings and, almost incidentally, make phone calls. The handset is being test-marketed by Ericsson in the United States, Japan, Singapore and the United Kingdom.

Such a device could drive sales of third-generation (3G) cellular handsets and services if it could hit a price tag below $159, said Gary Waymire, a principal with GVO. "The challenge for Ericsson is to build a 3G device without all the bells and whistles, but well-targeted to a class of users," Waymire said. "This is a new class of device all about entertainment and communications. It's more of a Game Boy than a cell phone or a Palm Pilot."

Asko Komsi, a principal architect with Nokia Wireless Software Solutions (Mountain View, Calif.) and a conference attendee, agreed that young people will be among the early adopters of 3G services and that these sorts of applications could drive the next generation of merged voice and data networks.

Aiming at a more adult market, another attendee, Robert Mayo, a researcher at Compaq Computer Corp.'s Western Research lab in Palo Alto, said he is trying to design a keychain computer. The wireless device would automatically pick up information, such as electronic business cards, from other keychain computers or wireless kiosks. The data would later be downloaded to a PC.

A group of graduate students from U.C. Berkeley demonstrated a network-of-sensors design as one way to create a ubiquitous computing environment. The students linked about 15 sensor modules via a low-data-rate 900-MHz wireless network running a compact TinyOS software platform along with a low-overhead routing protocol.

The TinyOS needs only 3.4 kbytes of storage and delivered a data rate of about 800 bytes/s based on a raw transmission of about 10 kbytes/s, polling sensors 10,000 times a second.

"You are not going to be able to get Windows CE to handle this," said Robert Szewczyk. "You need quite a bit of concurrency for polling sensors and processing data. And there will be a wide range of sensor types, so you will also want quite a bit of code reuse."

Such a network has already been used to control an air conditioner. One researcher proposed linking accelerometers to the sensors and installing them in a building. If an earthquake struck, the network could automatically measure how much sections of a building moved and report whether they were safe to enter.

"Our routing protocol is like the active messages used in high-performance computing," said Szewczyk. "It's perfectly suited to our needs because it takes very little processing power. It's interesting that what worked for high-performance computing because it was very fast also works for us because it is very cheap and requires little power."

Indeed, among technology hurdles, "the biggest problem ultimately will be the power source," he said. "If these devices are to be installed in a structure and used for a very long time they will have to use power-harvesting techniques or batteries with an unusually high power density."

The Berkeley demo system now gets one year of life in its lowest power state but only one day when the sensors are running at full capacity.

Communications is the biggest power drain on the network, which has lead Berkeley professor Kris Pister to experiment with so-called corner-cube reflectors. In his design, a sensor basestation supplies a laser light source and the sensors employ a MEMS mirror to reflect the light around the network. Mirror planes on the MEMS are wiggled to reflect light as a means of transmitting data in an essentially energy-passive network.

MEMS and sensors are the key building blocks of the pervasive-computing era of smart objects, said Paul Saffo, director of the Institute of the Future and a speaker at the conference. "This decade is defined by cheap sensors," he said. "MEMS is the leading technology."

But much of the debate at the conference centered on whether ubiquitous-computing designers are creating a Frankenstein monster or a more-welcoming world.

"The blandness of suburbia is giving way to the greater blandness of cyburbia," Saffo said. "Everything distracts us, nothing quite connects, it's all like [Microsoft] Windows, and we don't seem to have much of a choice." Invoking Tennessee Williams, he warned, "We are moving toward the world of Blanche DuBois, where we depend on the kindness of computers. Communications are moving from people-to-people to machines-to-machines. The challenge will be, how do we break into that conversation when things are not working."

IBM's Russell took a more upbeat view, maintaining the new era will usher in an age of what he called calm computing.

"People face many interruptions that they can absorb, but there's a real cognitive limit to that," Russell said. "If you get more interruptions than you can handle, your performance falls down and you feel terrible, so we want to create systems that create fewer interruptions and make people less anxious."

Specifically, Russell's group is helping IBM's pervasive computing unit develop a version of IBM's WebSphere software tools that could be used across multiple devices and wireless networks. His team is also working to characterize a new very short-range wireless personal area network that could create a new form of I/O, possibly geared for gesture recognition, that Russell said will be demonstrated at next year's conference.

Indeed, said James Landay, an assistant professor at U.C. Berkeley, the larger goal for developers here is to develop so-called multimodal interfaces that employ a combination of speech and gestures to control a widening range of computing devices. But that's no small task. "This is a field where there are few cookbook solutions," said Landay.