🥇The History of the Internet: ARPANET — The Birth

Other articles in the series:

By the mid-1960s, the first time-sharing computing systems largely followed the early history of the first telephone switchboards. Entrepreneurs created these switchboards to allow subscribers to access services like taxis, doctors, or fire brigades. However, subscribers soon realized that local switchboards were equally effective for communication and socializing with one another. Similarly, time-sharing systems, initially designed for users to 'call' computing power for themselves, soon transformed into utility switchboards with integrated messaging systems. In the next decade, computers would undergo another phase in the history of telephony – the advent of interconnected switchboards forming regional and long-distance networks.

Protosnet

The first attempt to integrate several computers into a larger entity was the project of an interactive computer network SAGE, the American air defense system. Since each of the 23 SAGE control centers covered a specific geographical area, a mechanism was needed to transmit radar tracks from one center to another in cases where an aircraft crossed the boundary between these areas. SAGE developers referred to this task as 'cross-telling' [cross-telling], and they solved it by creating data transmission lines based on dedicated AT&T phone lines extended between all neighboring control centers. Ronald Enticnap, part of a small delegation from the Royal Armed Forces sent to SAGE, led the development and implementation of this subsystem. Unfortunately, I did not find a detailed description of the 'cross-telling' system, but evidently, the computer at each control center determined when a radar track crossed into another sector and sent its records over the phone line to the computer of that sector, where they could be received by an operator monitoring the local terminal.

The SAGE system needed to convert digital data into an analog signal for telephone lines (and then back at the receiving station), which provided AT&T the opportunity to develop the 'Bell 101' modem (or dataset, as it was initially called), capable of transmitting a modest 110 bits per second. Later, this device was named the modem, due to its ability to modulate an analog phone signal using a set of outgoing digital data, and demodulate bits from the incoming wave.

Bell 101 dataset

Thus, SAGE laid an important technical foundation for later computer networks. However, the first computer network with a legacy that was sufficiently long and influential was a network with a name still known today: ARPANET. Unlike SAGE, it integrated a diverse set of computers, both with time-sharing and packet-switched processing, each with its unique set of programs. The network was designed to be universal in scale and operation, and it was meant to meet any user needs. The project was funded by the Information Processing Techniques Office (IPTO), led by director Robert Taylor, who was the head of computer research at ARPA. But the very idea of such a network was conceived by the first director of that office, Joseph Carl Robnett Licklider.

Idea

As we learned earlier, Licklider, or 'Lick' to his colleagues, was trained as a psychologist. However, while working with radar systems at the Lincoln Laboratory in the late 1950s, he became fascinated by interactive computers. This passion led him to fund some of the first experiments in time-sharing computers when he became director of the newly formed IPTO in 1962.

By that time, he was already dreaming of a way to connect isolated interactive computers into a larger superstructure. In his 1960 paper on 'man-computer symbiosis,' he wrote:

It seems reasonable to envision a "thinking center" that can incorporate the functions of modern libraries and the anticipated breakthroughs in information storage and retrieval, as well as the symbiotic functions described above in this work. This picture easily scales into a network of such centers, interconnected by broadband communication lines, and accessible to individual users through leased telephone lines.

Just as the TX-2 ignited Lick's passionate interest in interactive computers, SAGE may have inspired him to imagine how to connect various interactive computing centers and create something akin to a telephone network for intellectual services. Wherever this idea originated, Lick began to disseminate it within the community of researchers he had established at IPTO, and the most notable of such communications was a memo dated April 23, 1963, addressed to "Members and Sections of the Intergalactic Computer Network," which included various researchers funded by IPTO for time-sharing computer access and other computing projects.

The memo appears disorganized and chaotic, clearly dictated on the fly and not edited. Therefore, to understand what Lick intended to convey regarding computer networks, one must ponder a bit. However, certain points immediately stand out. First, Lick stated that the "various projects" funded by IPTO actually belong to "one domain." He then discusses the need to allocate funds and projects to maximize the benefits of this endeavor, since among the network of researchers, "for progress, each active researcher requires a software base and equipment that is more sophisticated and comprehensive than he can create in a reasonable time." Lick concludes that achieving this global efficiency requires some personal concessions and sacrifices.

He then begins to discuss computer networks in detail (as opposed to social networks). He writes about the necessity of a certain language for managing networks (which would later be called a protocol) and his desire to one day see the IPTO computer network composed of "at least four major computers, possibly six to eight smaller computers, and a large assortment of disk and magnetic tape storage – not to mention remote consoles and teletype machines." Finally, he elaborates over several pages on a concrete example of how interaction with such a computer network might develop in the future. Lick imagines a situation in which he conducts an analysis of some experimental data. "The problem is," he writes, "that I don’t have a decent program for plotting graphs. Is there a suitable program somewhere in the system? Using the doctrine of network prevalence, I first poll the local computer and then other centers. Let’s say I’m working at SDC, and I find what seems to be a suitable program on a disk in Berkeley." He requests the network to execute this program, assuming that "with a complex network management system, I won’t have to decide whether to transmit the data for programs to process it somewhere else or download programs to run on my own data."

All together, these fragments of ideas reveal a larger scheme envisioned by Licklider: first, to divide certain specialties and areas of knowledge among researchers funded by IPTO, and then to build upon this social community a physical network of IPTO computers. This physical manifestation of IPTO's "common cause" would allow researchers to share knowledge and gain access to specialized equipment and software at each workstation. In this way, IPTO could avoid wasteful duplication while enhancing the capabilities of every dollar in funding, giving each researcher across all IPTO projects access to the full spectrum of computing resources.

This idea of resource sharing among members of the research community through a communication network planted the seeds within IPTO that would later grow into the creation of ARPANET.

Despite its military origins, ARPANET, which emerged from the Pentagon, had no military justification. It is sometimes said that this network was developed as a military communication network capable of surviving a nuclear attack. As we will see later, there is an indirect connection between ARPANET and an earlier project with such a purpose, and ARPA officials periodically talked about “hardened systems” to justify the existence of their network before Congress or the Secretary of Defense. In reality, IPTO created ARPANET purely for its internal needs, to support the research community — most of whom could not justify their activities through defense work.

Meanwhile, at the time of the release of his famous memo, Licklider had already begun to plan the embryo of his intergalactic network, of which Leonard Kleinrock from the University of California, Los Angeles (UCLA) would be the director.

The SAGE model OA-1008 console, complete with a light gun (at the end of a cable under a transparent plastic cover), a lighter, and an ashtray.

Prerequisites

Kleinrock was the son of Eastern European working-class immigrants and grew up on Manhattan in the shadow of the George Washington Bridge [connecting the northern part of Manhattan Island in New York City and Fort Lee in Bergen County, New Jersey.While in school, he took evening electrical engineering classes at the City College of New York. Hearing about the opportunity to train at MIT with a subsequent semester of full-time work in the Lincoln Laboratory, he eagerly seized it.

The laboratory was created to meet the needs of SAGE, but since then has expanded to many other research projects, often only indirectly related to air defense, if at all related to defense. Among them was the "Barnstable Research" project, a proposed Air Force concept to create an orbital belt of metal strips (like dipole reflectors), which could be used as a global communication system. Kleinrock was captivated by the authority of Claude Shannon. from MIT, he decided to focus on communication network theory. Barnstable's research gave Kleinrock the first opportunity to apply information theory and queue theory to data transmission networks, and he expanded this analysis into a full dissertation on message-switching networks, combining mathematical analysis with empirical data gathered from simulations running on TX-2 computers in Lincoln's laboratories. Among Kleinrock's close colleagues in the lab, who, like him, used computers on a time-sharing system, were Lawrence Roberts and Ivan Sutherland, whom we will meet a little later.

By 1963, Kleinrock accepted a job offer at UCLA, and Licklider saw him as an opportunity. He was an expert in data transmission networks, working alongside three local computer centers: a main computing center, a health computing center, and the Western Data Processing Center (a cooperative of thirty institutions that shared access to an IBM computer). Moreover, six institutes from the Western Data Processing Center had remote connections to the computer via modems, and the System Development Corporation (SDC), sponsored by IPTO, was located just a few kilometers from Santa Monica. IPTO commissioned UCLA to integrate these four centers as the first experiment in creating a computer network. Subsequently, according to the plan, communication with Berkeley could explore the problems inherent in long-distance data transmission.

Despite the promising situation, the project failed, and the network was never built. The directors of the various UCLA centers did not trust one another and did not believe in the project, refusing to relinquish control over computing resources to each other's users. IPTO had little leverage in this situation, as none of the computing centers received funding from ARPA. This political problem highlights one of the key issues in the history of the internet. If it's very difficult to convince diverse participants that organizing communication among them and cooperation benefits all parties, how did the internet even come to be? We will repeatedly return to these questions in the following articles.

The second attempt by IPTO to build a network was more successful, possibly because it was much less ambitious — it was just a simple experimental check. In 1965, a psychologist and a student of Licklider named Tom Merrill left the Lincoln Laboratory to capitalize on the growing excitement around interactive computers by starting his own shared access business. However, after failing to gain enough paying clients, he began seeking other sources of income and ultimately offered IPTO to hire him for a study on computer networks. The new director of IPTO, Ivan Sutherland, decided to bring in a large and respected company as a partner for support, and he subcontracted the work to Merrill's company through the Lincoln Laboratory. Another old colleague of Kleinrock, Lawrence (Larry) Roberts, was appointed to lead the project on the lab's side.

Roberts, as a student at MIT, became skilled in working with the TX-0 computer built by the Lincoln Laboratory. He would sit for hours mesmerized in front of the glowing console screen and eventually wrote a program that poorly recognized handwritten symbols using neural networks. Like Kleinrock, he eventually began working at the laboratory for graduate credits, tackling problems related to computer graphics and computer vision, such as edge recognition and generating three-dimensional images, on the larger and more powerful TX-2.

For most of 1964, Roberts primarily focused on image processing. Then he met Lick. In November of that year, he attended a conference on the future of computers, sponsored by the BBC, held at a hot springs resort in Homestead, West Virginia. There, he engaged in deep discussions with fellow attendees until late at night, hearing Lick present his idea for an intergalactic network for the first time. Something clicked in Roberts' mind – he was excellent at handling computer graphics, but was essentially limited to a single unique TX-2 computer. Even if he could share his software, no one else would be able to use it since no one had equivalent hardware to run it. The only way for him to spread the impact of his work was to publish his findings in scientific papers, hoping someone could replicate them elsewhere. He concluded that Lick was right – a network was indeed the next step needed to accelerate research in computing.

Roberts eventually began working with Merrill, attempting to connect the TX-2 from the Lincoln Laboratory to the SDC computer in Santa Monica, California, using a telephone line that spanned the country. In an experimental project that seemed to be copied from Lick's memorandum on the 'intergalactic network,' they planned to have the TX-2 halt mid-computation, use an automatic dialer to call the SDC Q-32, run a matrix multiplication program on that computer, and then continue the original computations using its response.

In addition to the meaningful use of expensive and advanced technology to transmit the results of simple mathematical operations across the continent, it is worth noting the dreadfully low speed of this process due to the use of the telephone network. To make a call, a dedicated line had to be set up between the caller and the recipient, which usually went through several different telephone exchanges. By 1965, almost all of them were electromechanical (this was the year AT&T launched the first fully electronic exchange in Sayreville, New Jersey). Magnets moved metal bars from one place to another to ensure contact at each node. The entire process took several seconds, during which the TX-2 had to simply sit and wait. Moreover, the lines, which were perfectly suited for conversations, were too noisy for transmitting individual bits and provided very low bandwidth (a couple of hundred bits per second). A different approach was needed for a truly effective intergalactic interactive network.

The Merrill-Roberts experiment did not demonstrate the practicality or usefulness of a long-distance network, showing only its theoretical viability. But even that was enough.

Solution

In mid-1966, Robert Taylor became the new, third director of IPTO, following Ivan Sutherland. He was a student of Licklider, also a psychologist, and joined IPTO after previously managing computing research at NASA. Apparently, almost immediately upon his arrival, Taylor decided that it was time to realize the dream of an intergalactic network; he launched the project that created ARPANET.

Money from ARPA was still flowing in, so Taylor had no trouble obtaining additional funding from his boss, Charles Herzfeld. However, this solution carried a significant risk of failure. In 1965, there were quite a few lines connecting the opposite ends of the country, and no one had ever attempted anything like ARPANET. Other early experiments in creating computer networks can be recalled. For instance, Princeton and Carnegie Mellon built a grid of shared-access computers in the late 1960s in collaboration with IBM. The main distinguishing feature of this project was its uniformity — it used exactly the same hardware and software computers.

On the other hand, ARPANET would have to deal with diversity. By the mid-1960s, IPTO was funding more than ten organizations, each with a computer, and all of them operated on different hardware and software. The ability to share software was rarely available even among different models from the same manufacturer — this was only attempted with the latest line of IBM System/360.

The diversity of systems posed a risk that added significant technical complexity to the network's development and the possibility of resource sharing in the style of Licklider. For example, at the University of Illinois, a massive supercomputer was being built at that time with ARPA funds. ILLIAC IV. Taylor felt it unlikely that local users from Urbana-Champaign could fully utilize the resources of this enormous machine. Even systems on a much more modest scale — the TX-2 at the Lincoln Laboratory and the Sigma-7 at UCLA — typically could not share software with each other due to fundamental incompatibilities. The prospect of overcoming these limitations by gaining direct access to the software of one node while being at another was attractive.

In a paper describing this network experiment, Merrill and Roberts suggested that such resource sharing would lead to something akin to a Ricardian comparative advantage for computing nodes:

The establishment of the network can lead to a certain specialization of the collaborating nodes. If a node X, due to the presence of specific software or hardware, excels particularly at matrix inversion, it can be expected that users from other nodes in the network will take advantage of this opportunity, inverting their matrices on node X instead of doing it on their home computers.

Taylor had another motivation for implementing a resource-sharing network. Purchasing a new computer for each new node in IPTO, which had all the capabilities that researchers at that node might ever need, was an expensive endeavor. As more nodes were added to IPTO's portfolio, the budget was dangerously expanding. By linking all funded IPTO systems into one network, it would be possible to provide new grant recipients with more modest computers, or even to avoid purchasing them altogether. They could utilize the necessary computing power on remote nodes with surplus resources, and the entire network would function as a public reservoir of software and hardware.

After launching the project and securing its funding, Taylor's last significant contribution to ARPANET was selecting a person to directly oversee the development of the system and ensure its implementation. The obvious choice was Roberts. His engineering skills were unquestionable, he was already a respected member of the IPTO research community, and he was one of the few people with real experience in designing and building long-distance computer networks. Therefore, in the fall of 1966, Taylor called Roberts and asked him to come from Massachusetts to work on ARPA in Washington.

But it turned out to be difficult to entice him. Many academic supervisors at IPTO were skeptical of Robert Taylor's leadership, considering him lightweight. Yes, Licklider was also a psychologist, without an engineering background, but at least he held a doctorate and had certain merits as one of the founding fathers of interactive computing. Taylor was an unknown individual with a master's degree. How could he lead complex technical work in the IPTO community? Roberts was also among these skeptics.

But the combination of carrot and stick did its job (most sources indicate a predominance of the stick with a practical absence of carrots). On one hand, Taylor put some pressure on Roberts's boss at the Lincoln Laboratory, reminding him that much of the lab's funding now came from ARPA, and therefore he should persuade Roberts of the advantages of this proposal. On the other hand, Taylor offered Roberts the newly established title of 'Senior Scientist,' who would report through Taylor to the ARPA deputy director, and also become Taylor's successor as director. Under these conditions, Roberts agreed to take on the ARPANET project. It was time to turn the idea of resource sharing into reality.

History of the internet: ARPANET — the beginning

Protosnet

Idea

Prerequisites

Solution

Further reading