What do we learn from the Arpanet, the direct precursor of the Internet? Many things. But the surprising lesson is refinement. We too easily believe that a technological wonder jumped fully formed out of the mind of its inventor. Yet the Arpanet required many revisions as developers worked to understand it’s basic concepts and the implications of those concepts. It was a process in common to the way that IBM refined its original Personal Compu
ter, Microsoft refined the Widows operating system and Apple refined the iPhone.
In the case of the Arpanet, the protagonist is a computer scientist named J. C. R. Licklider. In 1960, he conceived an idea for how computers could support human activity. He called that idea “Human-Machine Symbiosis.” Over the next 5 years, he refined that idea. He explored the elements that it might need, and the way it might work. He recognized that it would require interactive computing, graphics, artificial intelligence, central archives of knowledge, and a network that would connect them all.
During that period, he also had the good fortune to be a leader of the one government agency that could develop his ideas about “Human-Machine Symbiosis”, the Defense Department’s Advanced Research Projects Agency or ARPA. ARAP began by developing interactive computing at MIT, computer graphics at Utah, and artificial intelligence at Stanford. In 1966, it began the project that would connect these sites together, the project known as Arpanet.
The following article, by Steve Crocker and David Alan Grier, shows how Licklider refined his ideas and how the Arpanet was merely one element of his Human-Computer Symbiosis.
The Context for Arpanet
J. C. R. Licklider
David Alan Grier
George Washington University
ICANN / Shinkuro LLC
In the fall of 1962, four years before he started the project that would come to be called the Arpanet. Robert Taylor was asked to join an informal government coordinating committee. The committee would meet at the Pentagon’s Advanced Research Projects Agency (ARPA). It was organized by J. C. R. Licklider, the newly-appointed director of ARPA’s newly-formed Information Processing and Techniques Office (IPTO) and it would transform computing research.
The committee had representatives from eight government agencies that funded computing research. The Atomic Energy Commission sent a representative, as did the research offices of each of the three services: the Office of Naval Research(ONR), the Air Force Office of Scientific Research(AFOSR), and the Army Research Office (ARO). The National Science Foundation (NSF) sent a representative even though it did not have a formal computing research program.  To the NSF, computing was a tool that supported other forms of research, most commonly physics. It had about $2.9 Million in its budget to fund university computer centers. On top of that, it provided about a hundred thousand dollars of grants for mathematical research relating to computing. 
In addition, the committee included a representative from the National Institutes of Health (NIH) and Robert Taylor, who represented the National Aeronautics and Space Administration (NASA). At NASA, Taylor oversaw a program that studied specialized computing systems for spacecraft control, simulation and general data display. As he recalled, the group would “exchange information about who they were funding, and look for opportunities to collaborate, and remove undesirable overlap.” 
Among the research agencies, IPTO was the new upstart but it had the potential to change computing radically. It was one of the smaller offices within ARPA yet its “modest budget was greater than the sum of all the other budgets represented on that committee,” Taylor recalled.  Furthermore, Licklider had a very expansive view of computation. While he appreciated the value of numerical calculation, he felt that computers could do more to support research, strategic planning, decision-making and organizational operation. He was also willing to spend his budget in ways that would make his vision a reality. “To begin is everything,” he wrote late in the fall of 1962, “even if it is necessary at first to build research systems along lines that would be uneconomic for widespread application.” 
Beyond this informal government committee, Licklider had already made connections with the existing institutions of computing research. Before coming to ARPA, he had worked at the Cambridge consulting firm, Bolt, Baranek, and Newman (BBN). In that position, he had gotten “acquainted with people in the computer world.” He had connections at Lincoln Laboratories, and also followed what was being done by “people at MIT, and Harvard, and UCLA.” He also knew the directors of the two most prominent university computing centers, Phil Morse of MIT Computer Center and George Brown at UCLA’s Western Data Processing Center.  These two centers were partially funded by IBM, which was the dominant computing company at the time. They provided computing services not only to their host institutions but also to other organizations their regions. Most importantly, they served as centers of instruction and research, as there were no university computer science departments in 1962. Most academic research, instruction and even software development was done by computer center personnel. 
Such connections shaped Licklider’s view of computing and what computing could accomplish. During his time at ARPA, he would establish an agenda for the agency that would guide not only his decisions but those of his successors, Ivan Sutherland, who served as IPTO director from 1965 to 1966, and Taylor, who became director in the summer of 1966. It was Taylor who started the Arpanet project with a meeting that has since become central to ARPA lore. 
In a few short years after Licklider arrived at ARPA and formed the Information Processing Techniques Office (IPTO), there were active, major research projects across the U.S. Their emphases were on interactive computing, primarily via time-sharing, artificial intelligence research laboratories, graphics research, and advanced computer architecture projects were transforming the computer science research landscape. CMU, Harvard, MIT, Stanford, Stanford Research Institute, UC Berkeley, UCLA,, Utah and others were the sites of newly created centers of excellence. Young professors who become the stars of the emerging computer science research community aggressively pushed the limits of how computers could interact with and support human reasoning, and hundreds of graduate students learned both the trade and the vision.
Part of the vision that Licklider brought to ARPA and that was shared by his successors was the connection of machines and people across these laboratories. The critical moment was reached in February 1966 when Taylor asked his boss, Charles Herzfeld, the Director of ARPA, for funds to begin work on the network. “I gave him a number or reasons why I thought the Arpanet project ought to be launched,” he recalled, and received a positive reply. It “was literally a 15 minute conversation.”  The speed is a little surprising because IPTO had recently terminated a networking project at UCLA.  In that short meeting, Taylor must have made the case (or perhaps reminded Herzfeld that others had made the case) that such a network made sense, that it had benefits for all concerned, that it was feasible with current technology (or technology that could be easily invented), and that ARPA had sufficient resources to build it.
In his brief meeting, Taylor may or may not have raised three organizational issues: Who would build the network? What organization would hold and operate it? Finally, who would develop and expand it? These questions were not entirely answered when the IPTO office produced its program plan for the network, eighteen months later. That document, written by Taylor’s two colleagues at IPTO, Larry Roberts and Barry Wessler, set the basic goals for the project and outlined how the network would be built. It did not, however, identify the kind of contractor who would manage the development of the network. In the field that should have held that information was the phrase “To be Determined.” By that point, it was all but certain that no external organization would serve as principal investigator. Instead, IPTO personnel would lead the project. 
Taylor may have had a vision for how the network would have been developed, operated and expanded, or he may have let the logic of events suggest solutions. In either case, his steps seemed to follow Licklider’s advice about the importance of starting the project and building a good research environment, even of that environment would “be uneconomic for widespread application.”
To understand the organizational issues that Taylor faced in 1966 and to appreciate the nature of his accomplishments, it is useful to return to the computing environment that existed in 1962, when J. C. R. Licklider created IPTO and began to push his vision for interactive, networked computation. Rather than being wholly apart from that environment, Licklider’s ideas drew from the strengths of early 1960s computing and the aspirations of its leaders. To those strengths and aspirations, Licklider brought a broad research perspective that grasped that computing could do more than simply solve systems of partial differential equations, the common physics application of the time. He also understood that “to begin is everything,” that it was just as important to demonstrate what computers to do today as it was to plan the ideal system for the future. 
Licklider began combining his broader vision for computing and his practical sense for how to implement that vision in his 1960 paper “Man-Machine Symbiosis.” This paper came out of his early experiences with interactive computing at Lincoln Labs, a systematic observation of his own activities, and some careful thought about what a computer could do for his work. He started with a time-and-motion study of his research, as if he were an efficiency expert or an industrial engineer. “I served as my own subject,” he reported. “I tried to keep track of what one moderately technical person actually did during the hours he regarded as devoted to work.” 
Such an approach was not uncommon in the late 1950s. At that time, computers were industrial objects. They operated in factory settings and were managed by people who were just as likely to have a background in industrial management than in programming or software development. Even the popular culture of the time, such as the 1955 movie “The Desk Set”, identified computer managers as efficiency experts, though the character in that film thought the term “a bit obsolete” and preferred the term “methods engineer”.
In his time-and-motion study, Licklider concluded that perhaps as much as 85% of his time was “devoted mainly to activities that were essentially clerical or mechanical.” Most consisted of task that could be easily handled by a computer of the time: “searching, calculating, plotting, transforming.” Two sets of tasks would require some more research. They required the machine to determine “the logical or dynamic consequences of a set of assumptions or hypothesis” and thereby help prepare “the way for a decision or an insight.” 
Licklider was not alone in a new role for computers in decision-making. Others, including the mathematician Richard Hamming, the future Nobel Laureate Herbert Simon, and the staff of the NSF were all thinking about how to use computers in support of human decision-making.. As early as 1956, the NSF had focused its attention on the problem of organizing scientific knowledge, arguing that better systems for storing and retrieving documents would save time for the scientists, “time they can actively spend on research.” While they accepted the limitations of the current computers, they felt that a “promising approach is the mechanization of systems to store and recover science information.” During the late 1950s, they had held conferences on information systems for scientific research and funded feasibility study at Case Institute of Technology. 
Much of the work developed or sponsored by the NSF dealt with networks only in passing. It was primarily focused on an idea called “the information system,” a computer that could store large amounts of text and produce that information in response to queries. The concept grew out of the hypothetical Memex in Vannevar Bush’s 1945 article “As We May Think” and was commonly discussed in popular culture. The play “Desk Set” revolved around a fictional information machine named “EMERAC”.
Licklider was certainly familiar with the thinking at the NSF and some of the other projects. He referenced many of them in his book, Libraries of the Future. However, he moved beyond a broad discussion of these systems to create a concrete research program, a program that had specific goals. He identified four technologies that would be needed for human-computer symbiosis: interactive computing, natural language recognition, graphical computer interfaces, and new programming models.  He then demonstrated how these technologies could be used to human-computer symbiosis.
In the first paper that presented his vision the 1960 “Man-Machine Symbiosis”, Licklider briefly noted that information systems could provide some of the “symbiotic functions suggested earlier in this paper,” by incorporating functions of libraries and bringing information to individuals as they did their work. He then moved beyond the individual computer system and argued that machines might be interconnected. “The picture readily enlarges itself into a network of such centers, connected to one another by wide-band communications lines to individual users by leased wired services.” In suggesting these connections, he was motivated by practical concerns. The connections would make these services widely available and hence, “the cost of the gigantic memories and sophisticated programs would be divided by the number of users. “
Licklider further revealed the pragmatic nature of his vision, when he conceded that he might not fully foresee how people and computers might interact. “Man-computer symbiosis is probably not the ultimate paradigm for complex technological systems,” he acknowledged. He speculated that artificial intelligence or even chemically simulated brains might be more effective than the technology he was describing. Still he argued that such technologies were at least 20 years in the future and that man-computer symbiosis should be operational long before that point. The intervening years, he argued “should be intellectually the most creative and exciting in the history of mankind.” 
Licklider returned to the themes of “Man-Machine Symbiosis” shortly after he became the director of IPTO. This second paper, “On-Line Man Computer Communication,” argued that “to begin is everything” and then connected his ideas to existing computer centers, institutions that provided computing services to broad populations. In universities such centers relatively new. Only six years prior, the NSF had complained that only “a few large computing centers around the country are available for basic research problems and these generally on a part-time basis only.” MIT had been able to create a general computing center only in in the following year, 1957, with help from IBM.  Contemporaneously, UCLA built a similar center on its campus, also with help from IBM. 
The director of the MIT center was the physicist Philip Morse. Morse had long been an advocate of computing as a service for research. During the second world war, he had argued for funding centralized computer centers and making those offices available to war time researchers. These offices did calculations not with computing machinery but with human computers, clerical workers who did computation by hand or with the aid of adding machines. “I have not been a developer of either hardware or software in the computer field,” he would later explain. “I’ve been a developer of the use of the machine”. 
As the director of the MIT Computing, Morse was interested in developing tools that made computing easier to use. In his first year as director, he had received the largest amount of funds that the National Science Foundation yet had granted for the study of computing methodology. It was neither for hardware nor software but for a handbook on how to compute higher mathematical functions. The handbook, commonly known as “AMS 55,” has become one of the most widely circulated books on computation. That grant was followed, five years later, by funds to create the Compatible Time-Sharing System or CTSS.  Morse was never the technical leader of the CTSS project. The computer scientist John McCarthy had conceived the system as a way to test if the timesharing concept “could be demonstrated by some judicious programming.” Fernando Corbató led the actual development team. Morse only raised the funds and gave it the blessing of the institution. 
Time-sharing was of interest to Licklider because it was an economic way of implementing interactive computing. He had no interest in the conventional operational model for computing centers, a model he once mocked as being “patterned after that of the neighborhood dry cleaner (‘in by ten, out by five’).” Computer centers ran their machines in batch mode, processing jobs more or less in the order in which they were received. The interval between submitting the program and receiving the output made it difficult to bring the computer into a person’s thinking process. In his second paper on human-computer interaction, the 1962 “On-Line Man-Computer Communications,” Licklider argued that batch mode was wholly “inadequate for creative man-computer thinking,” which required a “tight, on-line coupling between human brains and electronic computers.” To get that tight coupling could assign a single computer to each user or employ timesharing. The “economic obstacle fades,” he noted, “as the cost of a computer is divided among several or many users.”
As director of IPTO, Licklider would devote much of his initial budget to projects that would produce software tools that would support his research agenda. Many of the investigators on early IPTO projects had already demonstrated their ability to develop useful tools. This list included Marvin Minsky, John McCarthy, Herbert Simon, Allen Newell and Edward Feigenbaum. They had already helped create tools such as the CTSS Timesharing System, the programming language IPL-V and the language LISP. None of these projects were done as ends in themselves but were steps in larger research agendas devoted to artificial intelligence.
For Licklider, the most important tool was timesharing. “Among the first nine organizations that the IPTO funded in 1963,” noted historian Chigusa Kita, “eight were related to time-sharing.” The four major contractors were MIT in Boston, Carnegie Mellon University in Pittsburgh, the System Development Corporation in Santa Monica and the University of California in Berkeley. [38 MIT was developing the Multics system, using many of the people who worked on the CTSS system. SDC was creating a time-sharing system for the AN/FSQ32 computer, a machine that had been built for the Air Defense system SAGE. The University of California was modifying a commercial computer, the Scientific Data Systems 930 to run time-sharing software. Kita has noted that none of these programs could support some of the things that Licklider desired, notably interactive graphics. However, Licklider desired to expand the use of interactive computing. To do that, he needed a stable, flexible timesharing system. 
Time-sharing would have the side effect of encouraging researchers to collaborate, both unconsciously and consciously. Unconsciously, all users would be sharing the same resource. The computer cycles that went to one researcher could not be used by another. Consciously, the users were making tools for a common a system that could not only be shared by others but could be used simultaneously. Towards the end of the paper, Licklider hinted at this kind of collaboration by stating that he was trying to “couple man to computer as closely as man is now coupled to man in good multidisciplinary scientific or engineering teams.”
Licklider had great faith in multi-disciplinary teams. “I fell in love with the summer study process that MIT had,” he later explained. “It was getting physicists, mathematicians — everybody who could contribute — to work very intensively for a period of two or three months.” He had participated in two early in his career. “They had one on undersea warfare and overseas transport — a thing called Project Hartwell.” It was followed by a second effort, called Project Charles, on air defense. 
By the spring of 1963, Licklider was trying to bring some of that multi-disciplinary cooperation to the projects that were working under ITPO. That effort produced his third description of human-computer symbiosis, embedded in a memo Licklider sent to investigators apologizing for deferring a scheduled meeting. The memo, usually identified by the address line “Members and Affiliates of the Intergalactic Computer Network,” is usually cited because it contains a clear example of how Licklider would like to use the network. However, the main theme of the document is collaboration. “At this extreme,” he wrote, “the problem is essentially the one discussed by science fiction writers: ‘how do you get communications started among totally uncorrelated ‘sapient’ beings?’” 
Licklider seems to have grasped that it would not be enough merely to connect his research groups with a complex piece of technology. He also had to establish a framework for collaboration. “We are in great haste to construct a magnetic telegraph from Maine to Texas,” the naturalist Henry David Thoreau had famously written in the 19th century, “but Maine and Texas, it may be, have nothing important to communicate.” Licklider argued that perhaps the main reason that he desired to bring the researchers together was “to explore the possibilities for mutual advantage in these activities – to determine who is dependent upon whom for what.” Finally, he hoped to identify the groups that would receive a “bonus benefit” from collaboration, the groups that would be able to do more because they could borrow the work of others. Ultimately, he hoped to have an electronic platform that would support such discussions. For the immediate future, he would be grateful to see it “all on one blackboard.” 
Licklider would sketch the technology for a network, or at least his understanding of that technology, in his fourth work on human-computer symbiosis, Libraries of the Future. (1964) The book was a report on a study of libraries by the Council on Library Resources. The study committee included not only Licklider but also a surprising number of people who involved IPTO. Both Marvin Minsky and John McCarthy, IPTO funded researchers, worked with the group. Tom Marrill, who had done network experiments a Lincoln labs, also contributed. Working directly with Licklider was Phil Morse, the director of the MIT computer laboratory and Emanuel Piore, the head of IBM research.
In presenting his vision for a network, Licklider presents ideas that are consistent with his prior discussions of network but were far more complicated than the ideas that guided the initial Arpanet. This “plan is not a final plan or even a mature plan,” he confessed. “Perhaps it should be regarded only as a set of suggestions, made by a small group without expertness in all the potentially contributory disciplines.” 
It is far from clear that his network sketch influenced the design for Arpanet. Two key ideas were incorporated into the design much later: packet-switching, and routers – the Interface Message Processors.  However, in writing the report, he saw why it was important to start work on these ideas immediately and not wait for a perfect or an elegant solution. The report gave me “an early appreciation of the coefficient in the exponential increase in the power of computers.” Computers were not only getting faster, they were getting faster in a systematic way and at an exponential rate. “We figured that every two years the cost effectiveness of computer hardware was doubling.” 
Eventually, Licklider, as well as everyone else, would call this observation “Moore’s Law” after Gordon Moore, who was then the head of Fairchild Semiconductors. In fact, Licklider was looking at larger phenomena that predated Moore. Moore would articulate his observations almost a year after the publication of Libraries of the Future, and would restrict them to digital integrated circuits. Licklider recognized that not only computer circuits, but memories and full systems were following the same pattern. Our planning “should not be, limited by literal interpretation of the existing technology, “he wrote in the book. The current rate of growth suggested “that the basic “mechanical” constraints will disappear.” 
Expanding the Vision: Libraries and Information Systems
By the time he finished Libraries of the Future, Licklider had left IPTO and moved to IBM. His successor, Ivan Sutherland, largely continued Licklider’s and its focus on human-computer symbiosis. “Licklider had established a baseline and a direction that the office was going,” Sutherland observed. “So there were essentially no great battles to fight.” His major contributions to the work of IPTO were to establish a graphics research laboratory at the University with Dave Evans and to fund work on the ILLIAC IV high speed computer.
Sutherland made one effort to build a network, which he assessed as being a “major failure.” In fact, it was IPTO’s second attempt at networking. Taken together the two represented steps towards the Arpanet project, though neither produced a complete or sophisticated network project. The first attempt, under Licklider had connected the computer at SDC with the timesharing system at the University of California at Berkeley. “The experience,” explained the project report, has “contributed greatly to the problems involved in using time-sharing systems from afar.” 
The second effort, overseen by Sutherland contracted with UCLA’s Western Data Processing Center to connect three IBM computers on its campus. It “was not something that they wanted,” he claimed. They weren’t “interested in it from a technical point of view,” nor did they think that they would “reap enormous economic benefits from” the work. One of the present authors (Crocker) worked on this project as a student and remembers the conclusion differently. He viewed the Western Data Processing Center staff as being very interested in the technology of the network. However, they were unable to overcome the managerial and political differences among the three sites.
The proposal for the UCLA network shows that it would have been very different from Arpanet had it actually worked. It connected three identical computers, rather than the heterogenous machines that would ultimately be connected to the Arpanet. It will “increase computation efficiency by distributing the computing load to idle computers,” stated the proposal. 
The author of the UCLA proposal did not address the three key organizational questions of the network: the role that the individual computer centers would have in the construction, operation, or development of the network. Licklider returned that very issue in his last article on human-computer symbiosis, the 1965 “Man-Computer Partnership.” Conceptually, this article adds little to the ideas that described in his prior four works. He focuses on interactive computing, graphic interfaces, natural language recognition and new programming models. As with prior articles, gives a detailed description of the kind machine he would like to use and how that machine would interact with him.
Unlike prior articles, Licklider spends more time on the software that his desired machine would need than the hardware. This system will require “a great and ever-growing collection of public programs and data,” he wrote, as well as “a marked advance in the design and master of languages for communication between men and machines.” At that point, he then turned to one problem Robert Taylor would soon face in building the Arpanet. He argued that a network of such machines would need the “development of groups or communities of sophisticated users, capable not only of using the computer’s services but also of contributing to advancement of its services.”
Conclusion: Starting the Network
Licklider’s 1965 paper would be his last contribution, both visionary and practical, to the environment that would spawn the internet. In his five descriptions of Human-Computer Symbiosis, clearly presented the goals of the project and the value of interactive computing, information systems, and computer communications. These documents allowed Robert Taylor to focus on the concrete steps of building the network. His first task was to find a leader for the project and start working to identify the technology for the system. He had already identified a computer scientist who might be capable of building the network, Larry Roberts of Lincoln Laboratories. He would need several months to convince Roberts to join IPTO and start to build the network.
Roberts would also have to deal with two other key elements of the network. He would have to organize the separate groups at geographically dispersed sites to accept and use the proposed network. It was a problem that had been considered and discussed by groups such as the NSF and the American Association for the Advancement of Science(AAAS) but no one had made much progress towards solving it. About the time Taylor had begun his work, a AAAS writer had described geographically dispersed researchers as “invisible colleges” in the magazine Science. He had argued that such groups, which consisted of “hundred or so really active and knowledgeable people in any particular part of the research front of science” were the “natural units of which science is composed”. He had argued, as other had before, that technological tools would be able to connect these researchers by organizing publications and information. However, like most prior commentators, he had little sense of the effort this work would entail.
In the spring of 1966, Taylor was thinking about connecting the natural units of science or at least the natural units of IPTO science to link themselves into a common community. The “objective in the Arpanet,” Robert Taylor would later recall, went beyond developing the technology and designing a functioning communication devices. It “was to get these time-sharing communities, which had been built up locally” and to “build them into metacommunities by connecting them.”  Not all of the IPTO researchers would embrace the network, fearing that the project would take resources from their work. Taylor argued that it was important “to get the researchers interacting,” and organized meetings among the principal investigators. He also organized meetings among the senior graduate students, the people who actually do much of the work to build and operate the network. 
Looking back at that point in 1966, the second IPTO director, Ivan Sutherland argued that Taylor had assembled all of the elements that were needed for the Arpanet. He had a solid technological base in timesharing. He had computers that were doubling in power every 18 months. He had a colleague in Larry Roberts, who was capable of leading the world. He also had the managerial skills that would make the make the project successful. Bob had “a much higher degree of people skills than either his predecessor,” Sutherland noted with a reference to himself, “or his successor.”  with those elements, and the vision of J. C. R. Licklider, Roberts would start the Apranet.
 _______, “Fifth Annual Report of the National Science Foundation,” Government Printing Office, Washington DC, 1955.
 _______, “Resource Sharing Computer Networks,” Program Description, Information Processing Technology Office, Advanced Research Projects Agency, June 3, 1968. (Personal Papers of Steve Crocker)
 ________, “University of California at Los Angeles”, Proposal to the Advanced Research Projects Agency, February 11, 1965. (Personal Papers of Steve Crocker)
 _______, “Seventh Annual report of the National Science Foundation,” Fiscal Year 1957, Government Printing Office, Washington DC, 1957.
 _______, “Sixth Annual Report of the National Science Foundation,” Government Printing Office, Washington DC, 1956.
 _______, “ Twelfth Annual Report of the National Science Foundation,” Government Printing Office, Washington DC, 1962.
 Abramowitz, Milton and Stegun, Irene, A Handbook of Mathematical Functions, AMS 55, Natural Bureau of Standards, Government Printing Office, 1964.
 Abbate, Janet, Inventing the Internet, MIT Press, 1999.
 Aspray, William, “An Interview with Robert Taylor,”28 February 1989, Charles Babbage Foundation, OH 154.
 Aspray, William, “Oral History of Ivan Sutherland,” 1 May 1989 Pittsburgh, PA, OH 171, Charles Babbage Institute.
 Aspray, William and Norberg, William, “An Interview with J. C. R. Licklider,” 28 October 1988, Charles Babbage Institute, OH 150.
 Aspray, William; Williams, B. O., “Arming American scientists: NSF and the provision of scientific computing facilities for universities, 1950-1973,” IEEE Annals
of the History of Computing, Fall 1994, 16:4, 60 – 74
 Baran, Paul, “On Distributed Communications Networks,” Rand Paper p-2626, Santa Monica, CA. 1962.
 Bush, Vannevar, “As We Now Think,” Atlantic, July, 1945.
 Campbell-Kelly, Martin, “Data Communications at the National Physical Laboratory(1965-1975),” IEEE Annals of the History of Computing, 9:4, 1988.
 De Solla Price, Derek J., “ Ethics of Scientific Publication” Science, New Series, Vol. 144, No. 3619 (May 8, 1964), pp. 655-657
 de Solla Price, Derek, “Networks of Scientific Papers,” Science,
Vol 149 no 3683 (July 30, 1965 p 510-515.
 Evans, David and Feigenbaum, “Report on Contract SD-185,” January 15.1964, https://stacks.stanford.edu/file/druid:gm914kt1551/gm914kt1551.pdf, Last access April 10, 2018.
 Frand, Jason, “History of Western Data Processing Center,” http://www.frandweb.net/jason/anderson_years/clay.sprowls/history/WDPC/wdpc.htm. Last accessed 2018.04.09.
 Grier, David Alan, When Computers Were Human, Princeton University Press, 2005.
 Hafner, Katie, and Lyon, Matthew, Where Wizards Stay Up Late, Simon and Schuster, 1996.
Hamming, R., “The Mechanization of Science,” Proceedings of the 1961 16th ACM National Meeting, 1961, 31.101-31.104
 Kita, Chigusa, “J. C. R. Licklider’s Vision for the IPTO,” IEEE Annals of the History of Computing, 25:3, July-Sept 2003, p 62-77
 Licklider, J. C. R., Libraries of the Future, MIT Press 1965.
 Licklider, J. C. R., “Man – Computer Partnership,” Ekistics, Vol. 20, No. 118 (SEPTEMBER 1965), pp. 165-169.
 Licklider, J. C. R., “Man-Machine Symbiosis”, IEEE Transactions on Human Factors in Electronics, HFE-1, March 1960, p 4-11.
 Licklider, J. C. R, “Memo for members and Affiliates of the Intergalactic Computer Network,” Information Processing and Technology Office, Advanced Research and Projects Agency, April 23, 1963, Personal Papers of Steve Crocker.
 Licklider, J. C. R and Clark, Weldon, “On-Line Man-Computer Communications,” AIEE-IRE ’62: Proceedings of the May 1-3, 1962, Spring Joint Computer Conference.
 McJones, Paul, “Oral History of Robert (Bob) W. Taylor”, recorded: October 10 – October 11, 2008 Woodside, California, CHM Reference number: X5059.2009
 Marchant, William, The Desk Set, Samuel French, New York, 1955.
 Marill, Thomas and Roberts, Lawrence, “Computing Network Development to Achieve Resource Sharing,” AFIPS Spring Computing Conference 1966.
 Mertz, Richard, “Oral History of Phil Morse,” Computer Oral History Collection, December 16, 1970, National Museum of American History.
 Norberg, Arthur and O’Neil, Transforming Computer Technology, Johns Hopkins University Press,, 1996.
 Pelkey, James, “Interview of Joseph Carl Robnett (J.C.R.) Licklider” June 28, 1988, Computer History Museum X5671.2010
 Simon, Herbert, “The Corporation: Will it Be Managed By Machines?” in Management and Corporations, 1985, Anshen, Melvin and Bach, George Leland, editors, McGraw Hill, 1960.
 Walden, Henry David Thoreau, Walden, Ticknor and Fields, Boston, 1854.
 Walden, Dave and van Vleck, Tom, “The Compatible Time Sharing System,” IEEE Computer Society, Los Alamitos, CA, 2011.