Climbing the Digital Mountain

By Vinton G. Cerf

The funny thing about climbing mountains is that sometimes you cannot tell how far you have climbed until you turn around and look back. In January 1988, there were approximately 28,000 host computers on the Internet. By January 1999 the number had grown to over 43,000,000. The growth factor is slightly shy of doubling every year. Assuming continued growth at this rate for the next several years, Internet will have nearly a billion devices on the network by the beginning of 2007. For such phenomenal growth to occur, over 250 million devices would have to be added to the Internet in the year 2006. Such growth does not seem likely if it were to depend solely on personal computers, workstations and servers of various kinds. What is more likely, should these future statistics prove to be valid, is that an enormous number of low-cost Internet-enabled appliances will enter into the market. How might this happen? And what effect will it have on the Internet's operation, need for capacity, and performance?

We can find a glimpse of tomorrow's Internet in the trends of today. There are several thousand radio stations that put their audio on the Internet, to be heard through PC-based client software that interprets streams of packets as sound. One could as easily imagine a single-purpose device whose principal use is to "tune" in to Web-based sources of digital sound and to render them audible. This is already happening with MPEG3-encoded audio which can be downloaded and played at leisure or played in real-time after an interval of buffering to overcome variations in packet inter-arrival time across the Internet. In addition to turning the music industry upside down, replacing physical distribution with online, digital distribution of music, appliances such as this may also point the way towards other special-purpose devices for rendering Internet-delivered digital content in a variety of ways: as video, as formatted text, as images, as sound or a combination of all of these. Indeed, the richness of the delivered content is only limited by our ability to digitize, encode, compress, deliver and interpret either in real-time or after recording at the receiving site(s). Already, much progress has been made in the multicasting of digital content, but there is still a great deal of room for improvement in the efficiency with which such applications can be implemented.

Such applications will fuel unprecedented demand for digital capacity. In addition to the enormous demand for transmission capacity, performance requirements will be similarly aggressive. Delays in the Internet must be minimized and throughput maximized. Of course, all of these applications can also be contemplated in private or virtual private networks that depend on packet technologies similar to or even identical to those used in the Internet for their efficiency and their flexibility. Such applications will fuel unprecedented demand for digital capacity. In addition to the enormous demand for transmission capacity, performance requirements will be similarly aggressive. Delays in the Internet must be minimized and throughput maximized. Of course, all of these applications can also be contemplated in private or virtual private networks that depend on packet technologies similar to or even identical to those used in the Internet for their efficiency and their flexibility.

When we turn around to see how far we have come ten years from now, what will we see? First, we will find that a vast array of single purpose or at least simple-purpose devices will have joined the masses of personal computers, servers, mainframes and supercomputers on the Internet (or intranets) of the day. Second, these devices will be capable of absorbing and interpreting new software so that new "models" of products may merely be downloads of upgrading software as opposed to new physical devices. Third, because the devices will be programmable and capable of responding to external controls, new services will likely arise to manage, control or at least interact with these new creatures in the Internet zoo. A side-effect of this versatility is that products that might otherwise have been simply products may become part of a service rendered through the network in conjunction with cooperating devices. For example, a box of laundry soap could become part of a service if the washing machine is Internet-enabled and can receive configuration instructions from the Internet. A user might browse a web-site, inquiring how to set up the washing machine to deal with a particularly badly stained bit of clothing, and have the appropriate instructions conveyed through the net to the washing machine. Plainly access controls will be important, or the fifteen year-old next door may reprogram your household appliances while you are away at work or on vacation!

Video cassette recorders may well be on the Internet as well and, under control of software at distant web sites, might be made to record selected television programs coming over old-style analog transmission channels or through modern digital channels. Of course, by that time, the conventional video cassette may well have been replaced by a solid-state, high density, holographic memory, and the conventional television set by a holographic device capable of rendering scenes in three dimensions without the need for special glasses or goggles. Indeed, as we struggle to imagine what may be commonplace in a decade, we are confronted with the challenge of imagining new ways of doing old things as well as trying to think of the new things that weill be qenabled by the technologies of the future.V

It seems entirely likely that untethered communication will play a key role in the evolution of Internet and its applications. We already see strong evidence that continuous, wireless communication is highly valued in today's society. In some developing countries, where wireline telephony often has been very slow to develop, cellular telephone service is growing at rates exceeding 65% per year. A part of this is a consequence of the low cost of implementation of the service and a part of it is a direct result of the competitive marketplaces created by deregulation of telecommunications services. While the economics of asynchronous, low altitude satellite communication remains to be seen, one can readily posit wireless multicasting of Internet packets over digital broadcast satellites. Indeed, such services could also be found on digital cable systems as well. Combining geo-positioning services, such as the Global Positioning Satellite service, with wireless communication and connectivity to Internet may open up a variety of new applications. Devices that "know where they are" can use this information to access geographically indexed databases anywhere in the Internet to provide position-dependent information including directions for getting from where you are to where you want to go.

The desire to "stay in touch" at all times may lead to wearable devices that are integrated with articles of clothing or, at the least, can be conveniently worn for easy access. Perhaps in 2010 it will be natural to strap on a "Batman Belt" containing a variety of battery-powered devices all capable of interacting through the wireless Internet with similar devices and with servers of all kinds. One hopes that a commensurate convenience will have been achieved for the software for these gadgets. It is not appealing to imagine that one might have to "boot up" the telephone by waiting five minutes for the operating system to configure itself each time the device is turned on. Users will not tolerate less than instant availability, nearly 100% reliability, and the minimum possible delay in access to the services supported by these devices.

Whether such scenarios can and will become reality is a matter for speculation, but there is no doubt that elements of these ideas will be prevalent. Nor is there any doubt that significant engineering barriers must be overcome for such services to be widely available. Not the least of these will include robustness of the devices themselves, battery lifetimes, ease of use and cost. It will be easy to see how far we have come and how we have gotten there, in ten years' time - but we are going to have to live through it to see it in hindsight. There are a thousand paths into the future, but which one we take is no more predictable than the discovery of the transistor in 1948 or the integrated circuit in 1958. What we can know, however, is that the path will be filled with interesting, unexpected, surprising and sometimes baffling turns, taking us farther and farther up the digital mountain that lies ahead.