EVE M. PROPER

This paper was written for a 300-level history class called "Science, Technology, and Government." I received an "A" on it.

The Birth of the Internet: From urban legend to the mundane

One of the stories we tell ourselves -- one of our paradigms of how technology works -- is that the military develops some esoteric technology for their own use. After a few years, or perhaps when the war is over, the technology is adapted for civilian use, much to everyone's benefit. This story even comes true, sometimes: Tang is a frequently cited example, although it was created for NASA and not the military.

We are told that the Internet, too, is another example of this. Developed by the Advanced Research Projects Administration (ARPA) at the height of the Cold War to protect our military's communications in the event of a nuclear war, it eventually spawned today's bigger, better, faster Internet. Of course, there are a few more details, but that's the basic idea.

Or is it? Considering the newfound popularity of the Internet and the vast media coverage it has received, there is remarkably little written on its history. Occasionally a journalist tosses off a remark about the Net being born for military defense and makes some remark about swords and plowshares, but very little research has been done into the matter. The public is more concerned about the specter of "Net porn" ruining their children.

Perhaps one of the reasons this story is questioned so infrequently is that a little knowledge doesn't poke any holes in the theory. Most people who pride themselves on being wired can tell you that today's Internet developed from ARPAnet, developed by ARPA (later rechristened the Defense Advanced Research Projects Administration). As the latter name would suggest, DARPA commissioned and funded projects related to the national defense. So far so good, and that's where most people's knowledge ends.

Unfortunately for any poetic metaphors about swords and plowshares, the story of ARPAnet's impetus is only an urban legend. Like any good urban legend, its source is cloudy, but magazines as respected as Time have reported it in good faith. And when respected publications repeat myths as truth, their veracity becomes engraved in the public consciousness. Not all ARPAnet creation stories repeat this; while American Heritage magazine in 1995 published a story claiming the usual nuclearholocaust birth story, Science magazine got the story right back in 1990. (Ironically, the former story criticizes two more respectable magazines for "omitting" the true story of the Internet's creation.)(1)

One possible explanation for the legend that has grown up is the involvement of Paul Baran. Back in 1960, Baran, then at RAND, was worried about exactly the problem the Internet was supposedly invented to solve. If the U.S. was attacked by nuclear missiles, its missiles might survive but its communications network would not. He developed an idea for digital transmission of data. Instead of sending out a whole message at once on a line temporarily dedicated to that purpose (a la the telephone), why not break the message up into shorter pieces and send them out over different routes? The receiving end could assemble the parts and request any it was missing. The system depended on redundancy -- multiple links between communicators. If one was taken out, others would exist.

Baran suggested his idea to RAND and AT&T, but no one listened. No one thought it would work. Five years later, RAND and the Air Force were ready to do it, but the idea died on the drawing board: Baran didn't think that the Defense Communications Agency would be able to complete it. Several years later, Donald Davies in England independently developed the same idea. He called it "packet-switching," and the name stuck -- unlike Baran's term, "distributive adaptive message block switching."

Baran's (and Davies') idea was later used in developing the ARPAnet, and it's easy to see how Baran's motivations could be attributed to the rest of the Internet's fathers by association. However, the ARPAnet project was commissioned without any knowledge of Baran's work (he had since given up and moved on to other projects). It was only after development began and ARPA's researchers began looking for ways to make their network run smoothly that they were referred to Baran's and Davies' ideas. Neither Baran nor Davies ever actually did any work on the ARPAnet project; their packet-switching theories were simply implemented by the ARPAnet crew.(2) This is not to downplay the importance of Paul Baran in the creation of the Internet but to say that his motivations were not those of the rest of the ARPAnet developers.

What was their motivation, then? If the Internet isn't an artifact of Cold War nuclear holocaust terror, then what is it the result of? As mundane as it may sound, ARPAnet was developed to further timesharing. (Perhaps this is another reason the nuclear holocaust-prevention scenario sprung up: it was much more glamorous than a simple question of resource allocation could ever hope to be.)

In the 1960s, of course, personal computers were a long ways off in the future. Minicomputers were a recent development; smaller than mainframes, their lower cost made them available to more companies and institutions. Like mainframes, they were too expensive and large to give to many people. People who wanted to run something on a computer would get it punched up, since this was the day of punch cards, and send it off to the lab. The lab workers would run each "batch," one at a time, and results would be back within a few days.

Things dramatically improved with the widespread use of terminals. Terminals are connections to the computer, generally from somewhere nearby. The terminal has no central processing unit (CPU) of its own; it does have a keyboard, and maybe a monitor, that allows the user to interact with the computer. The terminal essentially sends information back and forth between the user and the computer. This wouldn't be so exciting in and of itself if it weren't for the fact that users never use 100 percent of a computer's resources. This means that several people can interact with one computer via different terminals simultaneously without any appreciable loss of speed. This "time-sharing" idea made much more efficient use of a computer's resources, important in a time when computers were scarce.

But what if a bunch of computers were connected? Then researchers at one institution could use the resources of another institution. Groups working on similar projects could view each others' results, for example, or watch a program run on a computer hundreds of miles away. ARPA thought this was an idea with promise; in the mid-1960s it agreed to fund it.(3) But this also posed difficulties. After all, no one had built a wide-area network before. How was data to be transmitted? An even greater problem was the variety of different computer systems. Today's choice of Windows, Windows NT, OS/2, UNIX, and Macintosh at the PC level may seem overwhelming, but in the 1960s there was no standardization. Different product lines by the same company ran on entirely different systems, and programming was still done in machine language much of the time. Perhaps ARPA could get two Sigma-7's to talk to each other, but what about a Sigma-7 and a Honeywell?

Rather than have computers at different sites talk directly to each other, ARPA decided to use Interface Message Processors, or IMPs. These machines were similar to telephone operators, receiving and routing messages. Their operation, however, was to be invisible to the user. Data would be transmitted via packets in "envelopes," meaning that the delivery systems never read the data they were transmitting. So while ARPA was busy setting the standards for message transmission and learning about packet-switching, a small Massachusetts consulting company was designing the IMP on an ARPA contract. Bolt, Beranek and Newman sat down with a few Honeywell DDP516 computers (the first ones were "ruggedized" to withstand any conditions(4), although that feature was later dropped) and started writing the IMP software.

Four sites were initially selected to receive IMPs; these sites had to develop their own software to enable their machine to communicate with the IMPs. These initial sites were UCLA, SRI, UC Santa Barbara, and the University of Utah. (ARPA itself and Case Western Reserve University were two of the first dozen sites to be connected.) All of the initial sites received federal research dollars, a requirement that was not going to change for a long time. Eventually, other institutions began to chafe at their exclusion: Usenet was developed as an alternative to ARPAnet by Duke University and the University of North Carolina.

To make a long story short, ARPAnet did grow over time, connecting more research institutions, mostly universities. File transfer, one of the original hopes for the network, was slow to develop, but something the internet's fathers had not initially conceived of was using a large piece of internet bandwidth -- electronic mail. E-mail provided researchers with a quick, convenient way to communicate; it was as quick as a telephone call, but it did not require both parties to be available at the same time. While much e-mail was work-related, much of it was personal, something that led to contention within the early networked community. Many users felt that personal communication was not only insecure (it was easy to read other people's mail then, as it still is) but also inappropriate. However, e-mail continued to flourish.

Other networks, both public and private, were slowly added on in the 1970s. The internet took a leap in size when the ARPAnet was joined by NSFNET in the 1980s, created by the National Science Foundation. The NSFNET became much larger than ARPAnet; it was upgraded in 1987 and today remains the "backbone" of the internet. Technically, the term "Internet" with a capital "I" referred to ARPAnet, while "internet" with a lowercase "i" referred to any other networks. This distinction became less than academic in the 1990s, when ARPAnet ceased to exist.

The network had grown to such proportions that government support was no longer needed, and the original equipment was no longer state-of-the-art. Some IMPs were destroyed; one was given to the Computer Museum in Boston for display. The original sites were still maintained but on different hosts. The original network may have ceased to exist, but its offspring was thriving. However, "thriving" was still a relative term; internet users were quite limited in scope still. Ordinary citizens who were online were generally connected to national services such as CompuServe, and limited to viewing the proprietary content that their company provided. As late as 1990, Science ran an article touting the benefits of the internet; perhaps, it said, the internet could benefit even more scientists. For this to happen, though, it said that it would become necessary to increase its ease of use, until using the network was as simple as dialing a telephone.

The internet of 1990 was still entirely text-based. Most browsers were text-only. This changed in 1993, when Mosaic was developed at the University of Illinois, UrbanaChampaign, at the National Center for Supercomputing Applications. It was the first graphical browser for the World Wide Web (WWW), a part of the internet. Finding a web site was now as simple as typing in its name, and sites could be navigated using the click of a mouse and not the stroke of a key. One of Mosaic's principle developers, undergraduate Marc Andreessen, went on to cofound Netscape only a few years later. Mosaic, of course, was not the first or the only application to simplify use of the internet, but it had perhaps the greatest impact on "ordinary" computer users. One of its benefits was that other parts of the Internet were eventually tied into the WWW; e-mail, for example, can be sent via Netscape Navigator and Microsoft Internet Explorer, two descendants of Mosaic.

Notes

1--It's amazing how many sources get the story of the Internet's early years at least partially wrong. Hafner and Lyon mention that Time made the mistake of repeating the national defense story; Congressional Quarterly Weekly Review called the early internet part of "a secret Cold War experiment"; Computerworld says that the internet was designed to connect DOD computers; and American Heritage is discussed above. This list doesn't include passing references to the internet's origins.
2--I initially thought this idea was original, but the authors of "A Brief History of the Internet, version 3.1" and the homesteader at http://www.dsu.edu/~anderbea/history both suggest that it was probably Baran's reports for RAND that led to the rumors.
3-- Hafner and Lyon attribute the suggestion that led to ARPRnet to Bob Taylor at ARPA.
4--The original ruggedized Honeywells were designed to withstand battlefield conditions, although that's not why BBN selected them. Frank Heart at BBN was concerned about curious graduate students playing around with the machines and fouling something up.

Bibliography

Anthes, Gary H. "The History of the Future." Computerworld, October 3, 1994: 101-105. Cerf, Vint. "A Brief History of the Internet and Related Networks." http://www.simmons. edu/~pomerant/techcomp/cerf.html. No date given.
Diamond, Edwin and Stephen Bates. "The Ancient History of the Internet." American Heritage, October 1995: 3445.
Frentzen, Jeffrey. "WinMosaic eases Internet surfing." PC Week, December 20, 1993: 63-64.
Glider, George. "Inventing the Internet Again." Forbes ASAP , June 2, 1997: 107-120.
Hafner, Katie and Matthew Lyon. Where Wizards Stay Up Late: The Origins of the Internet. New York: Simon & Schuster, 1996.
Leiner, Barry M., Vinton G. Cerf, David D. Clark, Robert E. Kahn, Leonard Kleinrock, Daniel C. Lynch, Jon Postel, Lawrence G. Roberts, and Stephen Wolff. "A Brief History of the Internet, version 3.1." http://www.isoc.org/internet-history. Posted February 20, 1997.
Mayr, David. "The History of the Internet and the WWW." http://ourworld.compuserve. com/homepages/dmayr/history.htm. Last revised June 1997.
McHugh, Josh. "Packet Man: Robert Khan." Forbes, July 7, 1997: 328-329.
Mills, Mike. "Networks: From a Defense Lab ... To a Worldwide Web of Users." Congressional Quarterly Weekly Review, April 3, 1993: 830831.
Palca, Joseph. "Getting Together Bit by Bit."Science 248 (1990): 160-162.