All Commentary
Sunday, November 1, 1998

Does the Internet Prove the Need for Government Investment?

The Real Internet Grew out of a Spontaneous Ordering Process

Fans of tax-funded investment often cite the Internet as an example of the good that government can do. Sure, they say, the Net now has uncountable millions of components, from Web sites to computer networks large and small. But if it hadn’t been for those first critical investments by the government, we wouldn’t have the Internet today. Politicians from Vice President Al Gore to Speaker Newt Gingrich now call for more such investments by the government—and the taxes to fund them. We must find and identify promising new technologies and invest in them to build tomorrow’s information infrastructure, they tell us, a task too important to be left to private enterprise.

The actual history of the Internet suggests that this is far from accurate. The Internet today bears little resemblance either to what the government wanted to build or to what it actually built. The innovations in networking that produced today’s Net occurred as much despite government funding as because of it. If anything, therefore, the Internet represents the success of spontaneous ordering over central planning, not the successful design of a new technology by the state.

What Is the Internet?

Examining the strength of the statist claim that government investment created the Internet requires thinking carefully about exactly what the Internet is. In a 1968 paper titled “The Computer as a Communication Device,” two pioneering computer scientists, J.S.R. Licklider and Robert Taylor, set out the principles that continue to define the Internet today. According to Michael and Ronda Hauben, the principles consist of these ideas: “1. Communication is defined as an interactive creative process. 2. Response times need to be short to make the ‘conversation’ free and easy. 3. Larger networks form out of smaller regional networks. 4. Communities form out of affinity and common interests.”1 Perhaps most striking about this definition is its similarity to the description of an open society built around free markets.

Implementing this vision requires connections between computers. The hardware part of the Internet consists of millions of interconnections between computer systems across the world. My office computer, for example, is connected by a fiber optic cable to a server on my university campus. The server in turn is connected to the telecommunications lines owned by many companies. Those lines eventually lead, by an unknowable number of paths, to Freeman editor Sheldon Richman’s computer.

The software part of the Internet distinguishes its networking capabilities from that of the telephone system. Those capabilities enabled me to make use of these physical connections to submit this article by e-mail. My computer’s mail program chopped the article into many small pieces, or packets, addressed them to Richman care of his Internet service provider (ISP), and sent them out along many different paths using a technique called “packet-switching.” When the pieces reached the ISP, they were reassembled for downloading, making the document legible on his computer screen. The transmission of information in these small packets is a key feature of the Internet; it is what enables the Net to transmit much more information than the conventional telephone network, which requires dedication of a circuit between the telephones on each end for the duration of the call. Transmitting information over the Internet allows multiple uses of the same connection, mixing packets of information from my e-mail with yours, for example, and then sorting them out at the appropriate points. The “pipeline” between two points can therefore be kept continuously full, with no wasted time for pauses in conversations or while one caller answers the doorbell in the middle of a call.

The Internet is thus a complex system—a mixture of hardware and software and millions of computers of every conceivable make and design running different operating systems. And thanks to voluntary and mutual coordination by millions of people, as in a free market, they all talk to each other to provide us with an unimaginable variety of information and services, commercial and free. It is a network of networks.

Today’s Internet is the embodiment of a spontaneous order in many ways. No agency or board controls it. No central planner decides how it will operate. It ignores national borders. It has changed the world.

Yet a few years ago, little of what we know today as the Internet existed: no bookstores, no Web pages, little public access beyond academic institutions. Before the Internet, there was ARPANET—the Advanced Research Projects Agency Network—a U.S. Department of Defense (DOD) network that is often described as the forerunner of today’s Internet. The ARPANET connection is thus the source of the “we wouldn’t have it without the government” story.

The Birth of ARPANET

Concern over secure military communications in the early 1960s, sparked in part by an act of sabotage against three Defense Department microwave towers in Utah in 1961, led the U.S. military to commission a series of studies by the RAND Corporation. DOD wanted a decentralized communications network that could survive a nuclear attack and permit the United States to launch missiles in retaliation. The RAND work included a description of the theoretical basis for a packet-switching network. Packet-switching would permit messages to be routed around damaged parts of the network. RAND was not interested in implementing the reports, however, and nothing was done with them for several years. DOD continued to fund research; as a result, much of the pioneering work on networking was either classified or in unpublished DOD reports and so unavailable outside the government.2

Also during the 1960s, new operating systems allowed mainframe computers to handle more than one user at once through “time-sharing” methods. As computers increased in speed and flexibility, these operating systems allowed the huge costs of hardware to be spread over more users. Differences in capabilities among the various types of computers at different facilities produced interest in communicating between machines as well, in order to allow researchers to avoid duplicating hardware. Networking thus naturally followed from time-sharing.3 The Defense Department was particularly interested in these developments because of its extensive investment in computers at universities across the United States, and it continued to fund networking research, largely through the Information Processing Techniques Office (IPTO).

Far from setting out to design the Internet at the command of bureaucrats, researchers did much of the early work as “extras” on existing contracts, diverting resources to unofficial projects.4 Indeed, the first director of IPTO has been described as a “‘Johnny Appleseed’ with a mission” to promote networking, rather than as someone merely responding to his superiors’ goals.5 More than funding was diverted to these projects. IPTO also managed to hire whatever private researchers the director wanted.6

Even when the official decision was made to set up a network connecting research sites, no one knew what uses the network would be put to. For example, Henry Edward Hardy writes that “The popularity of electronic mail on the early ARPANET was unanticipated by its designers.”7 The informal “Network Working Group” shaped the network. But decisions were made not by the Defense Department or by university bureaucracies; rather they were made by free-wheeling technical working groups that formed spontaneously. “What began at the early meeting was the creation of a community which cooperated more-or-less harmoniously for over 15 years,” Peter Salus writes.8

While groundwork for the ARPANET was being laid, few outside the technical groups had much enthusiasm for the project. Many were skeptical about packet-switching. The project also struggled with “a feeling that with the government involved it would be five years late and nothing would work.”9 Only after a successful demonstration at a 1972 computer conference did much enthusiasm appear. (ARPANET was not the only government-funded networking experiment during the 1960s and 1970s. The French, German, British, and Japanese governments also funded network development. None was particularly successful and none produced anything like the Internet. American technical superiority doesn’t explain the success of ARPANET; other Defense Department networking efforts failed miserably.10)

Thus the story of ARPANET is fascinating but it is not the commonly accepted myth that government investment produced the Internet. Far from being centrally planned, ARPANET grew up as the result of the successful capture of agency (and private) resources by individuals pursuing their own academic interests. The availability of no-strings-attached federal defense dollars undoubtedly made it easier for the early networking pioneers to concentrate on the technical details of their work. Given the intensity of their interest in the subject, however, subsidies were hardly required. Other money would surely have been found.

Why Was There No “Private ARPANET”?

Government activity generally “crowds out” private activity by absorbing resources that could be used elsewhere. Computer networking is no exception. Not only did the government directly seize resources through taxation and lock up knowledge in classified documents, it also lured many of the best computer scientists to work on its projects, slowing private-sector activity.

Private networks were attempted, but they failed. Setting up a network required permission from the Federal Communications Commission, and existing communications companies like Western Union fought the creation of new networks. Even when a private packet-switching network, Telenet, began operation, “many millions in legal expenses” were required to fight the regulatory battle prompted by RCA, ITT, AT&T, and Western Union. This kept Telenet from making profits for years.11 Regulatory barriers to entry, not a lack of entrepreneurial activity, slowed the efforts to build private networks.

Despite these obstacles, a private network among universities, USENET, sprang up. It resembled today’s Internet much more than ARPANET did. USENET developed because the Defense Department limited ARPANET to a relatively small number of sites. People at other sites wanted a network too, and USENET quickly surpassed ARPANET in usefulness because it lacked the restrictions DOD money imposed on ARPANET. (USENET still exists; it is a collection of newsgroups devoted to every subject imaginable.)

ARPANET Was Not the Internet

The Internet had its genesis within ARPANET. The protocols, or common language that all computers can use to communicate with each other, were developed by ARPANET people. In 1979 the government’s Advanced Research Projects Agency established the Internet Configuration Control Board. Both ARPANET and the Internet are packet-switched networks. But there are many profound differences. Indeed, the Internet today is so vast and complex as to be virtually impossible to describe accurately. It is a highly decentralized open-ended process that is not governed in any literal sense. (The Internet Society, a nongovernment, nonprofit organization, was formed in 1992 to “maintain and extend the development and availability of the Internet and its associated technologies and applications.”) There are several defining characteristics that make the Internet what it is and sharply differentiate it from ARPANET.

  • Its constant and rapid growth: There are hundreds of thousands, if not millions, of connections on the Internet depending on how one defines “connection.” ARPANET, in contrast, grew slowly. In 1969 there were just four host computers connected. By January 1976 there were just 63. ARPANET’s limited scope was due to its restriction to sites with DOD funding.12
  • Its freedom with respect to content: It is nearly impossible to control how individuals use the Internet. Because it was built with government funds, ARPANET users faced a number of restrictions on their use of the network. Commercial use, for example, was banned. Even one of the most popular uses, newsgroups, quickly ran into censorship problems. Although the first newsgroups concerned primarily technical issues about the network, users quickly began to establish groups concerning other subjects of interest. When several users proposed a newsgroup dealing with recreational drugs, those in control of the network rejected it as too controversial. (Drugs weren’t the only subject that was rejected; a proposal for a newsgroup called “gourmand” was also turned down because the creator refused to change the name to “recipes.”)
  • Its quick and spontaneous evolution of standards: For example, the HTML language for producing web pages is quite recent but is now used throughout the world. The early network protocols developed by the hackers and graduate students who built the initial ARPANET software evolved into a system whose popularity prevented an international standards body from centrally planning a new system (OSI), although a number of government-sponsored networks attempted to impose it.13
  • Its obliviousness to geography and borders. Wired magazine recently sent Internet pioneer (and Grateful Dead lyricist) John Perry Barlow across Africa to test network accessibility in remote corners of the continent; he found an astounding degree of connectivity already in existence. Connecting to the world outside DOD-funded sites was a problem for ARPANET users. Even sending e-mail across a university campus from a non-ARPANET site to an ARPANET computer often required routing messages across continents.

The international connections that are such an important part of the Internet today were also a struggle for the early ARPANET pioneers. Once the network expanded beyond a few experimental connections to universities abroad, international standards bodies dominated by representatives of government postal and telephone monopolies began to interfere with the protocol standards.

The Internet after ARPA

ARPANET and its successors were eventually shut down beginning in 1990 and network traffic shifted to a mix of private and public transmission lines by 1995, a period of unparalleled growth in both network usage and technology. The ARPANET experience did contribute significantly to the development of the Internet in a number of important ways. The experience with networking gained from running ARPANET, for example, helped thousands of computer scientists make advances that benefit us all today. Even more important, the drawbacks to a government network prompted ARPANET users to develop techniques for outwitting government controllers. That experience helped create the Internet. As Jim Gilmore, an important figure in evading the early attempts at censorship told Peter Salus, “when faced with obnoxiously centralized control over something that ought to be free and/or distributed, I look for a low overhead way around that increases freedom in general.”14

There have been other sources of technical innovation as well, however, and they deserve as much credit as the Defense Department, if not more, for the Internet’s capabilities today. A recent article in Wired magazine attributed advances in credit-card processing and video-streaming technology to a company selling pornography on line. Those techniques are now used by mainstream vendors as well.15

Unlike the mythical Internet that sprang forth from the ARPANET, the real Internet grew out of a spontaneous ordering process of the interactions of millions of individual users. The uses we make of the Internet were unimaginable to the researchers and scientists who created the networking protocols and hardware advances we rely on today. Far from being the result of the government’s “strategic” investment in the original Defense Department networks, today’s Internet developed at most accidentally from and often in spite of those investments. The explosive growth in commerce, for example, became possible only when the government’s ban on commercial use of the networks it financed was lifted.

Moreover, the “strategic” nature of the early investment in networking is a myth. No one consciously created the Internet. While an international network of networks undoubtedly would look different today had ARPANET never existed, there is also little doubt that packet-switching and e-mail would have evolved anyway. Dedicated, motivated people with a need to communicate—for commercial and noncommercial purposes—would have surely seen to it.


  1. Michael Hauben and Ronda Hauben, Netizens: On the History and Impact of Usenet and the Internet (Los Alamitos, Calif.: IEEE Computer Society, 1997). The original paper is available at
  2. Ibid., p. 117.
  3. Ibid., p. 90.
  4. Peter Salus, Casting the Net: From ARPANET to INTERNET and beyond . . . (Reading, Mass.: Addison Wesley Publishing Company, 1995), p. 25.
  5. Hauben and Hauben, p. 86.
  6. Salus, p. 20.
  7. Henry Edward Hardy, “The History of the Net,” master’s thesis, Grand Valley State University, September 28, 1993; at
  8. Salus, p. 53.
  9. Ibid., p. 67.
  10. Ibid., p. 124.
  11. Ibid., p. 109.
  12. Hauben and Hauben, p. 42.
  13. Salus, p. 123.
  14. Ibid., p. 143.
  15. Frank Rose, “Sex Sells,” Wired, December 1997.

  • Andrew P. Morriss is the D. Paul Jones, Jr. & Charlene A. Jones Chairholder in Law and Professor of Business at the University of Alabama. He is coeditor (with Roger E. Meiners and Pierre Desrochers) of Silent Spring at 50: The False Crises of Rachel Carson, forthcoming from the Cato Institute.