The Convenience Of
Small Devices: How Pervasive Computing Will Personalize E-Business
When Mark Bregman, general manager, pervasive computing, talks about IBM's new pervasive-
computing unit, he is careful to stress that the organization is not actually developing any
technology. Bregman's unit, which formally came into being in the first quarter of 1998, is
charged with supplying the vision that will underlie such development. Its job is to research
the market, determine the most appropriate strategy for IBM and help all the other business
units make the transition into pervasive computing. With a staff of roughly a dozen --
expected to grow to no more than 50 -- Bregman has his hands full. But he sees his role
as temporary, destined to become redundant "once
pervasive computing becomes truly pervasive." In an
interview with IBM Research editor Rowan Dordick, Bregman airs
his plans and predictions for what he sees as the next step in computer evolution.
Do you have a simple definition of pervasive computing?
BREGMAN: Pervasive computing is about enabling people to gain immediate access to information and services anywhere, anytime, without having to scrounge for a phone jack. However, while mobility and wireless technology are a big part of it, it's really about making e-business personal. Thanks to the explosive growth of the Internet, people will soon expect to be able to engage in electronic business effortlessly.
Are you saying that pervasive computing will spur the spread of e-business?
BREGMAN: Yes. Most of our focus on e-business has been in the corporate context -- an employee sitting in front of a PC with a browser and performing transactions. To some degree, we've extended that thinking to the consumer, to personal e-business. To make e-business truly pervasive, however, we need to lower the barrier to services. Surprisingly, the PC is part of that barrier.
The PC is designed as a general-purpose platform that can run many different kinds of applications. Yet we have found that individuals prefer separate appliances.
Many people have a home PC as well as a separate fax machine and a stand-alone
answering machine. Of course, in principle, you could use your PC for faxing and answering the phone;
it's probably even cheaper. So why don't most people do that? It's not convenient. Convenience is
driving our thinking in how to personalize e-business.
What other sorts of specialized appliances do you envisage?
BREGMAN: Say you're at home and you want to order a book from Amazon.com. You've got to turn on your PC, wait for Windows to start up, click on the icon for the Internet dialer, dial your Internet service through a global network service provider, launch your browser, type in a URL and wait for Amazon.com to load. It's doable but unnecessarily complicated.
Now, what if you had a telephone-like appliance, a "screen phone," sitting on your kitchen counter? It's connected to a back-end infrastructure so you can push one button and bring up the Web. It even has buttons for your favorite sites. So now you just push a button, and you're at Amazon.com. There's a little keyboard with which to enter your order, and you can touch the screen to select from a menu. It also has a smartcard slot so that the transaction is secure.
A product that's tailored to a specific task will always be more convenient than a general-purpose device. So although people could prepare every meal with some fancy Swiss Army knife -- which would be more economical than buying many separate implements -- people don't do that. It's a behavior issue, not a technology issue.
But how far do you specialize? Might you end up with one of these pads for ordering books, another for the weather, another for stock quotations?
BREGMAN: We don't know the answer yet. But we do know that certain activities tend to cluster around certain devices. You don't have different TV sets to watch news, movies and sitcoms. But you do have a different product to listen to music. You might argue that the TV already has speakers, an amplifier, power supplies -- why not make that your AM/FM radio as well? Again, because it's not convenient.
Then why does the computer industry sell general-purpose platforms instead of specialized products?
BREGMAN: Because fundamentally the computer industry is still an industrial products business, not a consumer products business. Most consumer industries don't sell the customer a platform. Chrysler has a platform called a Jeep, but what they sell is a product with many different models.
The computer industry has only recently begun to change into an individual consumer products business, and I think specialization is a natural result of moving to consumer sales. That is what happened with electric motors. At one time, motors were sold as platforms. If you wanted to create something that required motor power, you bought a motor and integrated it yourself. But eventually, motors found their way into specialized products, where they became so ubiquitous and invisible that a lot of people don't even realize there's one in their refrigerator.
The same process of moving technology under the covers is happening in our industry. So far, IBM's business units have been selling platforms, whether they be hardware platforms like the System/390® or RS/6000® or software platforms like DB2 or the MQSeries®, or even our e-commerce solutions. But we've concluded there's a huge opportunity in moving to the next stage in the evolution of computing.
Is that why IBM has formed a pervasive-computing unit?
BREGMAN: If we want to enable personal e-business through pervasive computing, we need to provide our units with guidance. So we've created this separate organization, which has the task of working with all the existing units to execute what is needed for IBM to be the leader in pervasive computing.
Take the screen phone idea. We are working with a number of vendors who are going to be the branded manufacturers for these devices. There is an opportunity for our Microelectronics business to sell them the integrated chip sets. We also need to work with the telephone companies, because connecting these devices to their infrastructures will require changes in their billing, customer support, customer systems management and so on. And to manage secure transactions between servers and billions or trillions of devices, platforms such as those used by telecommunications companies will need to be enhanced. Finally, we need to understand how to bring the content owners -- banks, travel companies, retail companies -- into this new world.
No one piece of IBM is in a position to coordinate that activity, and frankly, if you don't do all of those things just right, you won't create the market. The goal of my team is to articulate these requirements and then work with the individual businesses within IBM so that they can take the actions that will enable the market to grow quickly.
What are some of the technologies that will be needed to implement the vision of pervasive computing?
BREGMAN: There are two separate questions we have to ask ourselves: What are the places where IBM can make money? And what are the areas in which IBM has to participate in order to ensure success whether or not we make money?
We can make money building chip sets for screen phones. And we have an opportunity with silicon germanium for wireless technologies, as well as with our CMOS technology because of the levels of integration and power we can achieve. There are also opportunities for our storage products, both traditional disk drives to handle the additional data guaranteed by these new devices, and innovative technologies such as our prototype one-inch disk drive, which is an ideal product for some of these new devices -- cameras, copiers and screen phones.
We'll also have opportunities to enhance our existing product line, as in the area of MQSeries [message-oriented middleware products], systems management tools, application development tools and so on, to deal with pervasive computing.
What about the other category you mentioned -- technologies that are essential for pervasive computing to flourish?
BREGMAN: On the software side, Java could play a critical role as an interface standard. And in infrastructure, there's an emerging area of "connectivity services" that require a new class of server. Different classes of devices will have different functions and capabilities, so they will not be able to access the same kind of content. Rather than write separate applications on the server side, the idea is to create an intermediate server that will sit between the network computing content and the devices. It will take content written in HTML or XML, say, and transcode it into a format for a particular device.
These new proxy, or content connectivity, servers also provide other functions, such as persistence in the network. For example, if I drive into a tunnel and lose my connection as I'm making an airline reservation, a proxy server will be able to complete the transaction and then notify me later.
Security is also going to be key, and we have a lot of activity there
(see Research, Number 4, 1997). Brokerage firms, for example, not only want to be able to provide their customers with timely, accurate information, they also want to give them the ability to immediately act on the information. That interaction needs to be secure. Unless we offer a solution that is both effective and easy to use, security could be a barrier to the emergence of this market. Fortunately, there are some key technologies that we can both develop and deliver, and promote as standards.
Standards will be critical. My organization is seeking to ensure that IBM is a part of those key standards organizations, such as the OpenCard Consortium for smartcards (see "Smartcards Get Smarter"), and the Wireless Application Protocol Forum. It's important that those standards be defined in ways that allow us to participate in this market and that don't lead to proprietary protocols.
Should IBM be a player in the device arena?
BREGMAN:We need to ensure that devices are available with the right characteristics. But we've made an explicit decision not to be in the device business -- with IBM-developed, -manufactured and -branded products -- because we're not well suited to dealing in such a huge diversity of products. Instead, we want to have IBM technology in everybody else's products.
Is bandwidth going to limit the growth of pervasive computing?
BREGMAN: It depends on your expectations. Dick Tracy watches that provide live person-to-person video are a long way off, if they ever materialize. Even if you could provide that kind of bandwidth, everybody would want it, and then you'd run back into a shortage of aggregate bandwidth -- the bandwidth multiplied by the number of users. The cost of building the infrastructure is going to be the limiting factor.
How large a market are you looking at?
BREGMAN: You can measure this market in lots of different ways -- by customers, devices, applications and so on. We estimate that in 2001 there will be about a $14.5 billion opportunity for IBM just selling technologies to device manufacturers. There's a similar market for services that IBM could provide, including systems integration, consulting, and managed operations and outsourcing. Then there's another opportunity that is harder to quantify: the servers, storage and systems management needed to support the devices themselves.
How will pervasive computing affect the market for general-purpose platforms like the PC?
BREGMAN: The PC will still be an important part of the business environment. There, it's a capital investment that's going to be amortized. And, although you're procuring it for a particular task or worker today, a year from now you'll redeploy it for another task or another worker. So there is a reason to invest in a general-purpose platform. If the PC were too specialized, it might not meet that need.
For home use, it's a different story. While some people need a platform for writing programs, a lot of PCs are in the home because people want access to games on the Internet, and if they had a game machine in an Internet box, they wouldn't need a PC. So the PC could become less important as an individual consumer product.
What is the biggest conceptual difficulty in selling your vision of pervasive computing?
BREGMAN: There are two. Some people can't get beyond the platform model. The reasoning is: "I can do everything with my PC, and the PC has 40 percent market penetration in the home today, heading for 100 percent. And once everybody has a PC, why would anyone want any of these other gadgets?" And I would answer: "You have an oven at home, right? Do you have a toaster oven as well? Why? You could make toast in the oven!" The point is, in individual buying there are motives, like convenience, that defy the rational economic modeling used in the corporate world. And the market is shifting from corporate buying behavior to individual buying behavior.
The other stumbling block is that we have yet to figure out in detail the business models for entering the market. The biggest frustration for customers, and for us right now, is we've got a strong vision, and everybody says, "That's great! When can you deliver?" But execution takes time.
IBM's senior executive team shares a strong conviction that we'll get those models figured out, and that we shouldn't let that prevent us from getting started. In some ways, it's like the PC business in the beginning: we didn't know how it was going to evolve, but we felt we had to get into it.
Can the PC business serve as a model, then?
BREGMAN: Not really. Pervasive computing is quite a bit different, because it assumes a distributed business model. When we launched the PC, it was an independent business, separate from everything else in the company, and therefore able to succeed or fail on its own.
Pervasive computing is in many ways a return to IBM's earlier model as an integrated business. There are competitors who can play in each separate area, but you really need the whole solution to make it work. We're trying to leverage all the parts of IBM because the value here is more in the whole than in the parts.
