InformationWeek

The Internet: Where's It All Going?--Three key players in the development of the Internet see almost limitless possibilities for their creation

here's the internet headed? Some experts predict the Net, with the World Wide Web as its centerpiece, will have 1 billion users by decade's end. Hype or hope? To find out, InformationWeek went directly to the Net's sources: Vinton Cerf, Bill Joy, and Tim Berners-Lee. Cerf and a small band of computer sci entists spent the summer of 1969 changing computers from mere number-crunchers to powerful communications tools. Cerf was then a 23-year-old grad student at UCLA. Today he's MCI's senior VP for data architecture and president of the Internet Society, which oversees the Net's standards, guidelines, and practices. Joy developed U.C. Berkeley Unix in the early '70s. Its networking protocols and implementations would later become a foundation of the Net. Today, Joy is co-founder and VP of research at Sun Microsystems. Berners-Lee and colleagues at the European Particle Physics Laboratory (CERN) in Geneva created in 1989 a uniformed network on the Internet for in-house physicists: the World Wide Web. Today, Berners-Lee is director of the W3C Consortium, the official Web standards organization based at MIT. All three pioneers spoke recently with contributing writer Larry Lange.

So where's the Internet headed?

Vinton Cerf CERF: It's likely that homes will come equipped with LANs that link most appliances so power companies can adjust electrical peak demand and that enable remote diagnostics to run on computer-controlled devices. Internet services will be readily accessible and interworkable with video services. Computer software will help set up teleconference calls from home, support telecommuting--which will be mandated in more states where pollution from commuting has become a major problem--and enable multiparty games.

We're also seeing the emergence of financial-transaction support on the Internet, and perhaps bill-paying will at last become a part of that landscape. Teachers and parents will be able to confer by E-mail, and Johnny won't be able to claim that there is no homework because you'll see it on the Web page for his school and classes. Mobile access to the Internet by way of digital radio-equipped notebook computers will become commonplace.

Infrared communication will be used for local links so that lecturers can automatically download their notes, charts, and graphs into students' notebooks. Business cards will be replaced with "info-squirts" between notebooks. People sitting down at face-to-face meetings will be able to "internet" instantly this way. Of course, security is going to get very interesting in this environment.

JOY: I would segment the Internet first into commercial [markets], and then the home. In the commercial market, what companies want to do is disseminate information. So there's an awful lot of business to be had just disseminating stuff. If you look in the commercial marketplace, like entertainment or education or the information highway kind of stuff, you have to distinguish between information whose value is had by "having people have it." That would typically be the function of marketing or advertising. In fact, with advertising, so people can get the information, you pay for impression.

It seems to me right now that the best kind of information to disseminate is the stuff you w ant people to have, information you'd even be willing to pay for people to have. You don't have to worry about people copying it, so you get around a lot of these problems people have in trying to charge for things.

That's historically been true of things in an academic environment, where you have research results you want to share. For us at Sun, it's a natural, because we understand that philosophy of sharing information. When you sit down and say, "Well, I want to sell you a stock quote, and I want to charge, say, 20.03 cents for every something or other," that's much more difficult right now to implement. We think there are a lot of things you can do without getting to that point.

BERNERS-LEE: I held a session at the first World Wide Web conference in Geneva in July '94 to poke developers into the virtual reality direction. In particular, we aim to promote a route for incremental moves from existing static images toward full VR, so programs on slower machines--or with slower links-- can still negotiate some representation of the VR world that they can handle. This is the way.

To what degree will the Internet be commercialized?

Bill Joy JOY: Everybody wants to do that, to be the first bank of the Internet: "Wire me your credit card, and we'll take a percentage." We're looking at what kind of technology you'll need in servers. We've been working with the government to try to make sure that the protocols for electronic commerce are open so that we don't get the first monopoly "NetBank."

CERF: I think people who say that "X has to be free" don't really understand that everything must be paid for somehow. Choosing to use taxes to pay for it does not make something free. It does subsidize things, and one has to think twice about what should be--and what need not be--subsidized. Universal telephone service was subsidized not by the government directly, but by business ratepayers. Of course, the government indire ctly contributed by permitting a regulated monopoly in exchange for this subsidization mechanism.

Infrastructure always has to be self-supporting, even if this is by way of taxes, or it cannot survive. Unlike the road system, the Internet can and should be paid for as a commercially viable service. Special segments of the marketplace may well need assistance, and this can be provided by government or other support mechanisms.

On the whole, competitive provision of service still seems the most attractive formula.

BERNERS-LEE: The Internet, specifically the Web, is moving from appearing as a neat application to being the underlying information space in which we communicate, learn, compute, and do business. Corporations join W3C because they need that space to be stable and want to have a say in the way it evolves. They realize that the Web--as a highway and marketplace--has to be there as a precursor to all the fancy ways in which their products will be able to compete.

Many compani es would like to earn money on the Net but aren't sure how. What are MCI and Sun doing to make money online?

CERF: We're at the beginning stages of rolling out our internetMCI services. Together with marketplaceMCI, these are a very broad set of services, linked with the earlier networkMCI business services: ISDN [integrated services digital network] teleconferencing, E-mail, news services, and so on. I'm very happy with the backbone MCI has built and feel relieved to know that we can scale it up to 155 Mbps. We're providing transit services for many regional reseller nets and are linked to the rest of the worldwide Internet community through a number of packet exchanges, as well as supporting a number of direct international connections.

Dial-up access and dedicated service to our customers are all part of the plan, including access via other switching systems such as ISDN, hyperstream frame relay, SMDS [switched multimegabit data service] and ATM [asynchronous transfer mode].

Also, I think our Web home page, MCI's Grammercy Press is the most enjoyable and innovative work on the Web so far.

JOY: Right now, for a company that wants to be on the Internet, Netra [Sun's line of Internet servers] is really the machine of choice, because it's been proven and we've worked for so long on it. It's so robust, we rely on it heavily ourselves at Sun. It's for everybody who wants a good, reliable Internet gateway machine. We're looking at what other things we can add to it.

As far as our Web page goes, someone told me that it was the No. 1 hit site on the Net. What makes a Web site popular is that there's current and complete information.

A site tends to get stale, so there's been a big effort in Sun to link all the people who provide solutions on Sun systems by either letting them get some space on the server, or linking to their own existing machines in a very organized way, so you'll be able to find the informa tion you need.

How big a problem is security on the Net? What solutions are likely to emerge?

JOY: We're trying to get past the technical problems and ask, "How do we build robust, secure servers while preserving the anonymity of people in transactions and address some of the privacy issues?" We've been talking to a number of people, and I hope we'll be able to announce some deals where we're at least part of the experiments.

Tim Berners-Lee It's certainly an area where a lot of different people want to decide the answer, but eventually, it's going to have to be a cooperative thing.

I also think firewalls are important, as there's always going to be marauders out there on the Net. We wouldn't be connected to the Net without a firewall, because our internal network is so important to our business. We don't want every hacker in the world just wandering around our network.

CERF: The marketplaceMCI effort is a bold move t oward adding value to Internet access by making commerce on the Internet possible. We'll work with our partners and merchant customers to help them use the technology of the Web to provide market presence and financial transaction support in a secure and global environment.

Our objective is to bring the Internet within reach of the general public, and to make it fertile ground in which new products, services, and businesses
can thrive.

Are there any lessons others can learn from the way you're using the Internet today?

JOY: We're trying internally to cut down on the amount of paper that we use at Sun. With the hypertext kind of paradigm that you see on the Web, you can imagine putting all of your corporate documentation on there--even things like design files for machines. For example, I now have a lateral file cabinet full of those manuals that came with all the machines they set up, and it's indexed by part numbers. It sure would be nice if there was just one browser where all of that stuff could get hyperlinked together on the Web. We've made a corporate commitment to work on that, and it's already making a big difference.

Right now, we have several viewers. We have our employee manuals online in one viewer, the stock plan administration's in another, health care in another, and expense reporting in another. They're all in different databases, so you can't easily make a link.

Say you have an expense report, and you have some exceptional thing and you want to point to the place in the employee manual where it authorizes that. That's a problem today, in general, because there are no authoring tools in these Web browsers.

But you can imagine that making up an expense report can be like authoring a page. We think it's really going to make a big difference in corporations. People in companies spend a lot of time in engineering and management just looking for things, and that would really help solve that problem alone.

What needs to happen before the Internet beco mes an everyday household application?

BERNERS-LEE: Clearly, we need a good market in dial-in PPP [Point-to-Point Protocol] access, with a number of compatible PPP or SLIP [Serial Line Internet Protocol] providers in each area. We can assume that a Web browser and TCP [Transmission Control Protocol] software will be packaged with any operating system from now on: For example, see IBM's OS/2 WebExplorer and the coming Windows 95.

The terminal equipment is still expensive, though I'm looking forward to the day when my daughter finds a rolled-up 1,000 pixel by 1,000-pixel color screen in her cereal packet, with a magnetic back so it sticks to the fridge.

CERF: We plainly need to provide better tools for securing Internet access. This would allow companies and individual subscribers to protect their internal resources while allowing them to reach out and be visible on the global Internet. We're also discovering new applications--packet voice, packet video, multicasting and multipart y conferencing, games, and so on--that require extensions to basic "best efforts" packet service.

Privacy concerns are rising and need to be addressed. Creating a routing system for a billion networks--the current scaling target--is a major challenge.

Mobile access is yet another. I think technology costs are dropping fast enough that it isn't out of the question in 10 to 15 years, maybe sooner. For instance, watch how quickly cellular telephone service is spreading in developing countries where wireline services have been slow to emerge. PCs and LANs are rapidly absorbed in these countries since they're easy to install and apply locally.

What about the advent of information haves and have-nots? Is it likely? If so, will it be a problem?

CERF: While this is something we must be vigilant to avoid, I feel compelled to say that we have always had some gaps in access to information. What one hopes is that Internet access will be widely available, and that much of its information w ill be provided at no additional cost.

Library access is already becoming a reality. In Maryland, for example, "freenets" are emerging that are locally subsidized and linked to the Internet, and local bulletin board systems are connecting.

There will always be premium services, but I don't think it is a foregone conclusion that we will end up with the apocalyptic haves/have-nots gap.

JOY: One of the issues right now is that if you want to look at Web content, you really need 56-Kbps speed. We need to work on compression so people can use it at 14.4 Kbps, or something that you can get in an inexpensive modem or pocket PC.

Are Sun Microystems and MCI finding ways to prevent this schism? Is preventing it even possible?

JOY: There are a lot of things you can do around the variable between the corporate Net and the public Net. To speak to the corporate side, one of the things we're doing here at Sun on a project we have in Aspen, Colo., is to ask, "What kind of services do we need for the community--home, school, and city government?" That's a whole different problem because, frankly, people [at home] don't have the same networking and ability to get information as they do in a company. At work, everyone's got a workstation or a PC with a network right on their desk.

CERF: MCI has participated in the provision of Internet service to the academic and research community since it became involved in the NSFNet [National Science Foundation Network] backbone. It contributed much to this effort, including goods, services, and cash. It has contributed its high-bandwidth facilities to the gigabit research effort and, more recently, has joined with others to help bring schools and libraries up on the Internet. MCI understands the meaning of public service and the importance that its services have now--and will have for
future generations.

MCI is also a founding member of the Internet Society and has freely contributed in-kind and financial support to its work.

Do you recall the first words communicated over the Internet?

CERF: I wish I could remember, but the chances are that it was something like, "This is a test." Engineers, most of the time, aren't poets.

How did the name 'World Wide Web' come about?

BERNERS-LEE: It's obvious, of course. The "Web" is for its decentralized nonhierarchical topology, which is key--and "World Wide" goes better with "Web" than does "global."

In the early stages, people told me it was a great project, but that it would never take off with a name like that. They suggested instead names from Greek mythology. I also got a lot of grief for making an acronym with more syllables than the name itself.

What's your favorite Web site?

BERNERS-LEE: In a historic sense, the Vatican Library exhibit on the Web was--and still is--a milestone, because it had quality information that was artistically and beauti fully presented. Basically, we showed everyone what could be done with the Web.

Comments on this story?

InformationWeek http://techweb.cmp.com/iwk