JCR Licklider's vision for the Information Processing Technology Office (IPTO)

The Information Processing Techniques Office (IPTO) of the Advanced Research Projects Agency (ARPA) of the U.S. Department of Defense was established in 1962 to implement flexible military command and control systems. When setting the IPTO's research funding agenda, its first director, JCR Licklider, emphasized the development of time-sharing systems . This article will explore how Licklider's early vision of a "network of thinking centers" laid the groundwork for the IPTO's most famous project , the Arpanet .

JCR Licklider was the first director of the Information Processing Technology Office (IPTO) of the U.S. Department of Defense's Advanced Research Projects Agency (ARPA). As Martin Campbell-Kelly and William Aspray point out in Computer: A History of the Information Machine , IPTO support will The time-sharing system brought computing into the mainstream because it had the largest budget for computer research of any federal agency. Licklider's strategy is to allocate the IPTO's budget to a handful of organisations. Of the first nine organisations that IPTO funded in 1963, eight were related to time-sharing systems. However, the fact that Licklider prioritized time-sharing systems is somewhat odd, since time-sharing systems were not even mentioned in the 1961 government or Department of Defense report that led to the creation of the IPTO. Furthermore, even mentioned, early time-sharing systems were unable to support human-machine interaction through graphics, which is critical to the main goal of IPTO—the development of command and control systems. Thus, Licklider's subsequent acclaim as the granddaddy of graphical, interactive computing seemed out of place in the early 1960s.

The beginning of IPTO

IPTO was created when the Kennedy administration demanded improvements to the command and control system for U.S. defense purposes. As a result, the Director of the Office of Defense Research and Engineering tasked ARPA in June 1961 with the development of such a system. ARPA asked the Institute for Defense Analyses (IDA) to conduct a study of digital computer applications based on three objectives:

1. Study command and control problems to determine the criteria for the effective application of computers for command and control.

2. Propose future DOD development goals for automated command and control capabilities and develop guidelines to help future planners characterize individual systems.

3. Delineate problem areas that require accelerated research.

The research culminated in a report called Computers in Command and Control , which was submitted to ARPA in November 1961, arguing that the ultimate goal of the Department of Defense was an automated command and control system . While the team found that the technical feasibility of using computers is currently generally overestimated and that "the commander must remain the primary information processing element of the system," the team listed "several potentially productive options associated with the chain of command." areas of research and development”. These situations are as follows:

a. Develop improved planning, analysis, and programming techniques.

b. Develop improved programs and languages for communication between machines and users.

c. Fundamental research aims to improve our understanding of the complex processes of pattern perception, concept formation and identification, problem solving, learning, and decision-making.

d. Research aimed at improving the reliability of computers and their associated hardware.

In the Institute for Defense Analysis (IDA) report, the time-sharing system was never mentioned. The likely reason is that, around 1960, the time-sharing system could not support the intensive use of monitors on the terminal.

In fact, Wesley Clark, the designer of the MIT TX-2 computer, opposed time-sharing in MIT's Long Range Computation Study Group in 1960, He believes that the time-sharing system has serious limitations. Ironically, as early as 1954 Wesley Clark had "specified in a memo titled ' The Multi-Sequence Program Concept ' how to write a subroutines so that they can be executed by multiple different callers at the same time" and designed the I/O interrupt system in the TX-2 computer. Wesley Clark himself implemented the basic technique of allowing I/O devices to share CPU time, which could serve as the basis for a time-sharing system. However, the research team led by Wesley Clark of MIT Lincoln Laboratory did not use this technique for time-sharing systems, but for the I of the TX-2. Use of /O to connect devices to make computers more "approachable". In 1963, Ivan Sutherland developed Sketchpad on the TX-2, a digital drawing system with a light pen that Licklider greatly admired.

Oddly, while Licklider fully understood the importance of advanced human-computer interaction, whenever he had a choice, he always leaned toward time-sharing systems. For example, in the early days of the IPTO, when Licklider decided to fund Douglas Engelbert's group at the Stanford Research Institute, Licklider asked Engelbert to use a time-sharing system for his people computer interaction research . Engelbart was puzzled:

Licklider moved quickly. By early 1963, we had a funded project. But while I suggested using a local computer and building an interactive workstation, Licklider asked us to connect the monitor to the System Development Corporation (SDC) AN/FSQ-32 computer in Santa Monica, Experiments were performed under the Q32's novel time-sharing system. (Converting the Q32 to a time-sharing machine was an IPTO project for SDC.) ...for various reasons, the first year was very inefficient relative to the purpose and plan of our project, which is not the case in a pioneering enterprise common.

Apparently, due to Licklider's instructions, human-computer interaction research in Engelbart's group was slow. Licklider's decision went against his image as a "visionary of interactive computing."

Licklider published a paper titled " Artificial Intelligence, Military Intelligence, and Command and Control " in Military Information Systems in 1964, starting from The economic perspective explains his support for the time-sharing system. In that book, based on the 1st Information Systems Science Conference held in November 1962, he argued that complex information systems required a large number of programming tasks and that large and fast computers based on the concept of "reasonably synchronous time-sharing" could be used economically. Used to facilitate the work of programmers. However, this reason is not sufficient to explain his tendency towards time-sharing. Because, as Licklider himself put it in another 1962 paper on online human-machine communication, "In critical military systems, such as SAGE [semi-automatic ground defense environments], economic constraints are more less, the need for human-computer interaction is greater or more pronounced.” And at that time, “computing costs were decreasing; it was no longer completely uneconomical for a person to think in real time with a mid-range computer.”

Therefore, the development of the time-sharing system was somewhat out of place for the Department of Defense's new command and control office. Licklider successfully used this economic rationale to formally propose the development of time-sharing systems, but he claimed that the main goal of the IPTO was to develop the interaction between humans and computers. Why did Licklider develop human-computer interaction through time-sharing systems? In analysing the reasons behind this seemingly contradictory goal, let's examine the resources that Licklider has used to fund the research agenda that the IPTO has formed. In analysing the reasons behind this seemingly contradictory goal, let's analyze the sources of Licklider's research agenda to fund the IPTO. Clearly, the time-sharing system was important to Licklider's main focus, which he called "a thinking center ." Before joining IPTO, he already had the concept.

"Thinking Center"

Back in 1957, Licklider joined BBN, an acoustic consulting firm based in Cambridge, Massachusetts, to lead a group of experimental psychologists working in the field of acoustics. While the group already had an analog computer, Licklider asked Leo Beranek, one of the founders of BBN, to buy a digital computer, which he did. Later, in 1959, Licklider had access to a prototype of the Digital Equipment Corporation (DEC) PDP-1, the most advanced machine ever designed after Clark's TX computer. Soon, Licklider and his colleagues, such as Edward Fredkin of BBN and John McCarthy of MIT, began developing a time-sharing system for BBN based on the PDP-1.

The Digital Equipment Corporation (DEC) PDP-1 was of interest to BBN because the computer "is characterized by high speed and relatively little memory" and "its low cost means that it is generally not shared by a large number of users". The BBN time-sharing system was experimental as it initially allowed only 2 users to use teletypewriters, which was later increased to 5. In all likelihood, Licklider and his colleagues were not motivated by practical needs, but wanted to confirm the possibility of extending this technique to larger systems.

Licklider participated in one of a series of lectures held in 1961 to mark MIT's 100th anniversary, in which he discussed how time-sharing systems made digital computers "at your fingertips" at the university. and". In another talk on "Time-Sharing Computer Systems" by John McCarthy, Licklider's 1960 paper on " Man-Computer Symbiosis " was presented as "Thoughts on Time-Sharing Systems" "The latest paper" was presented.

In this paper, Licklider not only proposes the concept of a new human-machine thinking system, but also develops a research agenda for this goal, detailing memory hardware, memory organization programming, computer languages, and I/O devices and other necessary elements. Although the time-sharing system is not the core of the paper, it is clearly presented as a prerequisite, "Speed Mismatch between Men and Computers":

Today's mainframe computers are too fast and too expensive to think collaboratively with a single human in real time. Obviously, in order to be efficient and economical, computers must distribute their time among many users. Timesharing systems are currently under active development. There are even arrangements to prevent users from "breaking" anything other than their own personal programs.

Some time in 10 or 15 years, it seems reasonable to envision a "thinking center" that would combine the functions of libraries today, and the expected advances in information storage and retrieval and the symbiosis suggested earlier in this paper Function. This vision can easily be scaled up to such a central network, interconnected by broadband communication lines, and connected to individual users by leased line services. In such a system, the speed of the computer would be balanced, and the cost of huge memory and complex programs would be divided by the number of users.

Licklider sees time-sharing as an economic problem and a key technology for sharing intellectual resources, the "thinking center." He even mentioned such a central network.

At the time of the article's publication, Licklider was not only involved in the development of the time-sharing system, but during the winter of 1960-1961 he also led the BBN group researching future libraries . Both of these efforts are steps towards the realization of a center of thought. In his research agenda for the BBN research group, he encourages members to study the state of the art in five areas: storage and retrieval of information; memory organization; human-computer communication; language processing; and artificial intelligence . As he wrote in his 1960 paper, these were "information storage and retrieval and the symbiotic functions" whose advances were thought to be prerequisites for the realization of a thinking center.

In terms of implementation, BBN, through its subsidiary Telcomp, introduced time-sharing in 1963, a "leased-wire service" that Licklider believed was the technology needed to connect these central networks. So, when Licklider was chosen as the first director of the IPTO, he had already taken steps to start realizing his vision of a knowledge resource-sharing network.

In this context, it is understandable that he envisioned a network of time-sharing systems from the very beginning of the IPTO. Licklider showed his enthusiasm by describing a network in an internal ARPA memo (sometimes referred to as the " Intergalactic Computer Network memo ") in 1963. He sent a memorandum to IPTO-funded researchers to "Members and Affiliates of the Intergalactic Computer Network" in preparation for the April 25, 1963 meeting. The memo is his careful computer research agenda for funding. In the memo, Licklider explained:

I realize that there may be only rare occasions when most or all of the computers in the entire system operate together in an integrated network. In my opinion, it is interesting and important to develop a comprehensive network operation capability.

He continued, "Let me imagine that I'm sitting at a console that includes a picture tube monitor, light pen, and typewriter." The eight-page memo, three of which detailed an experimental psychologist's hypothetical Computer network usage. Here he demonstrates his comprehensive understanding of the center of thought. At the end of the memo, he noted: "The truth is, in my opinion, if we try to make the most of the facilities we have, there will be many or most problems that the military needs to address urgently." Therefore, although no military There was an immediate need for time-sharing systems or their networking, but Licklider nonetheless tried to convince researchers that the development of time-sharing systems would then benefit the military.

His consistent focus on the network of thought centers is visible even after his stewardship of the IPTO ended. In Licklider's only book , Libraries of the Future , the core concept of a "procognitive system" extends his view of such networks. In the book, he explains that the network consists of four levels of subsystems, all based on time-sharing techniques, pointing to IPTO-funded MIT, Carnegie Institute of Technology, Systems Development Corporation, RAND Corporation, and BBN Corporation The time-sharing system is the best way to think about layers 3 and 4 of the network.

From these facts, we can see that Licklider added time-sharing to the IPTO's agenda to realize his vision of a network of thought centers, and he even prioritized it. But how can the picture of thinking centers easily expand into such a network of centers?

A true SAGE system [semi-automatic ground defense environment]...

Archived JCR Licklider papers include an interesting 1957 typed manuscript (with some handwritten corrections): "Real SAGE Systems or Thinking in the Direction of Human-Machine Systems." ( The Truly SAGE System or Toward a Man–Machine System for Thinking ) Licklider's Network of Thinking Centers is a transformation of SAGE as an advanced information network. To understand its historical significance, we need to examine Licklider's relationship to the SAGE project in the 1950s.

Licklider's relationship with SAGE began in 1951 when, as an associate professor at MIT, he was selected as "a token psychologist" for Project Charles, a A project established by the Provincial Institute of Technology to study the possibility of improving air defense systems. MIT Lincoln Laboratory was established in the early 1950s with the main task of developing the national air defense system. Although Licklider was a formal advisor to the Psychology Group at Lincoln Laboratory, he worked closely with the group's leader, George Miller, to set up a research effort to detect information from simulated graphical data obtained by radar. group.

The group worked with electrical engineers on data transmission and with engineers from Polaroid on analog graphics technology, but it had little direct connection to digital computer development at MIT Lincoln Laboratory. However, Licklider "had the opportunity to learn about computers, radar equipment and communications with Lincoln Laboratory." As Licklider has often mentioned, his exposure to advanced digital computers in his laboratory in the late 1950s was a crucial experience in forming his vision for the use of advanced computers in the future.

Wesley Clark, the principal designer of Lincoln Laboratory's TX computer, recalls introducing Licklider to one of these computers (in the citation, "LINC" was designed by Clark in the early 1960s the name of the minicomputer):

I just showed him its usefulness and powerful display and expected him to draw his own conclusions about the enormous value of interacting directly with a computer. I learned this "on-line" operation with Whirlwind and the Memory Test Computer (MTC), and I designed the TX-0, TX-2, and of course, the LINC, all in the same , used in a direct interactive way. All of these computers have display ranges, and "interactive" means using them extensively. Licklider was clearly interested to see how we did it all. Like many other visitors, he understood the enormous value of direct access to computers and wanted to follow our example at Lincoln Laboratory.

As Clark points out, it was around this time that Licklider recognized the potential for widespread use of displays.

Around the time he first saw the TX-2, Licklider also had some interaction at a meeting with Lincoln Laboratory's Communications Technology Group, which works on pattern recognition, led by Oliver Selfridge (Oliver Selfridge) led. Selfridge and GPDinneen presented a paper on pattern recognition at the Western Joint Computer Conference in 1955. At the same session, Clark also presented a paper on pattern recognition. Thus, through his contacts with colleagues at Lincoln Laboratory, we can assume that Licklider acquired the knowledge of pattern recognition by digital computers. Armed with this advanced computational knowledge, Licklider even began to think about an advanced information system through SAGE's holistic approach.

As Licklider wrote in "The Truly SAGE System," the purpose of the 1957 paper was "to propose a system that seems desirable and reasonable from a psychological point of view, while From a psychologist's point of view of approximate engineering, it is not a completely infeasible system." This paper points out that Licklider proposed the concept of "a network of thinking centers" through SAGE, and added some new Function. The title of Licklider's manuscript suggests that, despite the name, SAGE was not enough to live up to Licklider's expectations, but he did not criticize SAGE harshly. Instead, in the 1950s and early 1960s, he maintained close ties with the Air Force (the developers of SAGE). In addition, he wrote the paper as a special research paper for the National Academy of Sciences-Air Force Research and Development Division (NAS-ARDC). Licklider later acknowledged that Air Force Research and Development (ARDC) was the main source of funding for the 1960 "Human-Machine Symbiosis" paper.

The dissertation consists of 10 sections. In the first part, Licklider wrote that he was concerned with finding a way to "improve the quality or efficiency of scientific and technical thinking." The framework of the paper proposes the effective utilization of computers in networked information center systems. The first four sections explain the goals of the network and provide an overview of the system. The next section, "Steps to Implementation," presents the four steps required for implementation (which were listed as sections in his original outline, but are actually subsections) and describe each step in detail ; The last short section describes the potential of the system.

The use of this system as an information center is described as a general application of SAGE. At each center, "of course there is a large digital computer with very large memory" and "a center of related fields interconnected by telecommunication pipes". There are four man-machine communication modes in the system. First, a human operator can speak into a microphone. Second, there's a combined display and control unit with a large, flat surface on which material can be projected and the operator can write with a pencil or stylus. Third, the operator can enter graphical information (such as curves and functions) into the computer and associate the information thus entered with the graphical information displayed by the computer. Fourth, the operator could use "typing or a typewriter - perhaps also as a computer output device". Of these modes, the last three were achieved through the SAGE console, but Licklider added a new feature of speech recognition that was clearly beyond the level of 50s technology. Later, in his "Human-Machine Symbiosis" paper (1960), Licklider also described in the section "Automatic Speech Production and Recognition" how an "ideal" and "Viable" voice communication.

The calculation method proposed by each center is used online by multiple users based on time-sharing technology:

This computer is time-shared by multiple people. Arrangements for presenting information to people are highly developed. They include digital-to-analog converters, curve plotters, large-screen cathode ray tubes, automatic typing or printing presses, and speakers.

It is worth noting that he mentions time-sharing several times here, as Licklider's 1957 paper is one of the first written records to mention time-sharing in this sense. While the concept of time-sharing systems is often associated with John McCarthy, who specifically referred to time-sharing systems in a 1959 MIT memo, the "True SAGE System" paper states that benefit Creed had been using the time-sharing system as early as the summer of 1957. It is possible that Licklider came across the term "time-sharing" in the SAGE project, although there it was used to describe the round-robin scheduling process of the SAGE program. But at least Licklider's intentions to advance its development are clear in his 1957 paper. For Licklider, time-sharing is a technology that allows users to have multiple simultaneous accesses, as well as a technology needed to share information stored in the mainframe computers in the Think Center.

The key point is that whether it is a SAGE or a thinking center network, computers are networked for data processing and storage, and the computing method is online. Like SAGE, Licklider's description of a "thinking center" could "easily be extended into a network of such centers."

Against "artificial intelligence"

As we have seen, Licklider's network of thought centers was inspired by SAGE. But what about the automation trends seen in SAGE?

The title of his 1957 dissertation indicated that Licklider wanted a truly semi-automatic system with a human element. In the paper, he identified four categories of people in the system: scientists, engineers, maintainers, and programmers. Machines need humans to help them complete their programs.

The designers of SAGE attempted to automate by gathering information from a variety of sources and using this data as a composite map of the development of the air situation, but it was semi-automatic in many senses. It requires a lot of operators. In SAGE, there has been a transition from manual systems of the past to transfer data by hand drawing or voice communication to the use of automation. In 1958, Licklider criticized SAGE for its inadequacy in achieving the level of automation, stating:

The SAGE system provides our primary experience with situation analysis and control in large human-machine systems. But it also doesn't help convey the concepts in our minds. The SAGE system was initially thought to be very automated. A large number of human operators were introduced into SAGE, mainly to handle tasks that were not feasible for computers. So let's get a first look at the Air Force's information processing and control systems, not so much a human helping the machine, but a true human-machine symbiosis.

This criticism also appeared in his 1960 paper, which outlined the concept of human-machine symbiosis in terms of its specific application to Air Force systems. Thus, Licklider does not deny the need for automated functions to help computer programs, and even encourages its use in future human-machine systems. But in SAGE, there are thinkers: commanders and weapons commanders. Licklider argues that, in his "true SAGE system", the most important components of humans are thinkers:

Scientists and engineers combine their brain data processing with machine facilities to form a system that is more efficient than systems made by humans or mechanical parts alone.

At the time, SAGE had another general application, developed by IBM: the Semi-Automatic Business Research Environment (Sabre), American Airlines' online flight reservation network. Sabre is basically an advanced automation of previously manually operated information network systems. All users of the Sabre system follow the same preset procedure, although in a network of thought centers, users apply information in many different ways. In this sense, Sabre is a commercial application that makes SAGE simpler and more stable through "IBM's proven and experienced engineering approach derived from its SAGE experience". In contrast, the network of thinking centers will be more complex and advanced than SAGE. In Sabre, the user is like a radar in SAGE, entering a piece of information into a pre-set process. Sabre basically automates the equivalent of a SAGE commander, deciding which seat should be reserved or removed. But in a network of think centers, there will be many thinkers who enter some data and use the system information in their thinking process, just like the commanders in the SAGE system do. Also, thinkers at the Licklider Thinking Center can use the entire system simultaneously in different ways, whereas SAGE has only a single air defense program. In this sense, Licklider does not lean toward full automation, including the decision-making process, as Sabre does. In fact, Sabre's direction seemed promising at the time, as artificial intelligence was considered a key technology for automating complex human-machine systems. Licklider disagrees; in fact, he uses "human-machine symbiosis" to express his opposition to automation.

Man–computer symbiosis

Computers have been compared to the human brain long before the term artificial intelligence was coined. Even the use of the term computer implies the anthropomorphism of machines, as the old usage of the term refers to people, not machines. Fittingly, throughout the 1950s and 1960s, expressions such as "giant brain" or "electronic brain" were frequently used in popular media.

Licklider expressed sympathy for research in this area in the "Artificial Intelligence" section of his 1957 manuscript:

There is little reason to limit a man or a machine to what at first seems to be its own domain. In fact, if machines could develop partially in the image of humans, the coupling problem between humans and machines would be greatly simplified. Therefore, it is advisable to explore the possibilities of research areas currently designated by the terms artificial intelligence and self-organizing automata. Such research and the study of human thinking and perception should be mutually reinforcing.

In fact, Licklider did communicate with participants in the Dartmouth Summer Artificial Intelligence Research Program in 1956, when the term "artificial intelligence" appeared to be officially used for the first time. For example, Marvin Minsky, one of the original members of the summer program, lists Licklider as one of his mentors. In the late 1940s, Minsky was one of the students in the Licklider Seminar, which was affiliated with the Harvard Psychoacoustic Laboratory. They often build electronic devices together to conduct experiments in learning theory. According to Minsky, in the early 1950s Licklider built "a wonderful machine" that "recognized whoever said the word 'watermelon' in what sentence". At the 7th Massey Conference, when Licklider presented on "The Feedback Mechanisms and Circular Causal Systems in Biological and Social Systems" on the psychology of acoustics Claude Shannon, who was also a member of the Dartmouth Project at the time of the research presentation, was a speaker after Licklider, and they all participated in discussions together at the conference.

Other researchers associated with the Dartmouth project include Oliver Selfridge, a former assistant to famed mathematician Norbert Wiener, who was at the Massachusetts Institute of Technology in Lincoln. The laboratory conducts pattern recognition research, where Licklider worked with him for a while. Another participant, IBM's Nathaniel Rochester, programmed a neural network model on a computer. The work is said to apply Donald O. Hebb's The Organization of Behavior (1949 ), Rochester was inspired by Licklider's 1952 lecture on Hebb's theory inspiration. In fact, in the preface to his book, Hebb noted that Licklider was one of the scholars he was grateful for "for his nuanced criticism of much of the manuscript, improving both content and style".

In his "Human-Machine Symbiosis" paper, Licklider lists 13 papers on "Theorem Proving, Problem Solving, Chess, and Pattern Recognition Programs" as the best papers in AI, including research he recognizes staff papers. He went on to say:

In short, it seems worthwhile to avoid arguing with (other) AI enthusiasts who see only machine dominance in the distant future. During this period, however, major intellectual advancements will be made by closely collaborating humans and computers, and this will be a fairly long transition period.

Clearly, Licklider is trying to differentiate his concept of human-machine symbiosis from artificial intelligence, considering the possibility of fully automated human-machine systems, including a heuristic human-brain program. This point was at the heart of his 1960 paper.

"Human-Machine Symbiosis" was the first article in the inaugural issue of the March 1960 issue of IRE Transactions on Human Factors in Electronics. The journal was one of the newly established human factors research journals at the time. In the 1950s, an emerging area of research involved designing complex systems that included humans as elements. Such systems are often referred to as human-machine systems. The main human-machine systems are for military use, but there are also requirements for "commodity production or facility construction" in industry.

As described in the official history of the Office of Scientific Research and Development (OSRD), during World War II, a professional field focused on the problems that occurred between man and machine emerged. By 1960, new interdisciplinary fields dealing with problems involving "people as part of a system" were given various names such as human factors in engineering, ergonomics, engineering psychology, biomechanics, applied experimentation Psychology and ergonomics, etc., as mentioned in the preface to McCormick's famous textbook Human Engineering. As he wrote in this 1957 book, McCormick noticed that many people disliked the term "ergonomics," but he adopted it because it was the most popular expression, although "There is not enough description of the field". In 1957, the American Psychological Association announced the formation of a new chapter of the Association of Engineering Psychologists to form an academic field to collect their research on human factors in human-machine systems.

Passionate about the establishment of the academic field, Licklider served on the editorial board of another newly published journal, Human Factors, in the late 1950s and became an engineering psychologist in 1961 The fourth president of the association. Thus, Licklider's "Human-Machine Symbiosis" clearly reflects important elements of the new field of human factors research.

The "Human-Machine Symbiosis" paper proposes that computers can be ideal partners for human "formulative thinking" , which is better than fully automated human-machine systems. The paper then provides the necessary research agenda to achieve the goal of interaction between humans and computers. At the time, researchers saw automation as an important goal of human-machine systems. If such a system were to be fully automated, humans would disappear entirely, replaced by computers performing all tasks, even intellectual tasks previously thought to be reserved for humans. Humans will no longer be part of the human-machine system. Licklider is trying to change this trend by using the term "human-machine symbiosis".

There is also an interesting paper in Licklider's archive of papers titled: "Human-Machine Symbiosis: Part of the 1958 NAS-ARDC Special Study Oral Report, Submitted on behalf of the Committee on the Role of Humans in Future Air Force Systems, November 1958 (Man–Computer Symbiosis: Part of the Oral Report of the 1958 NAS-ARDC Special Study, Presented on Behalf of the Committee on the Roles of Men in Future Air Force Systems, November 1958) This seems to be Licklider's first The term "human-machine symbiosis" is used in the paper for the first time. It shows how Licklider explained the importance of human-machine symbiosis research to the U.S. Air Force, the main agency funding this type of research. In his 1958 "Man-Machine Symbiosis" report, Licklider referred to another report: "Report of the Committee on the Roles of Men in Future Air Force." Systems), he wrote: "Since I don't have time to report on all of these topics, I'll just pick one -- a sub-topic, actually -- and focus on it. We've been calling it human-machine symbiosis."

From this, we know that Licklider coined the term human-machine symbiosis in committee discussions and has been using it in the context of future military applications of human-machine systems, at least as early as the summer of 1958.

Licklider began his report in the style of popular science fiction in early 1958: "The president pressed the golden button. The wheels whirled, the lights flickered, the display flashed, and the computer announced: 'I'm glad you asked I'm the problem.'" Just after this humorous start, Licklider rejected the image of complete automation through computers. He doubts that the image of a self-thinking machine created by the general understanding of artificial intelligence is sufficient. At the time, the phrase "human-machine symbiosis" was a valid phrase to call attention to his argument against full automation: there should be a partnership between humans and computers. He denied early full automation in "intelligence data processing, mission planning, combat control and assessment." Licklider believes that "it will take 10 to 20 years for the development of artificial intelligence to enable machines to think or solve problems of military significance" on their own. And he also confirmed the necessity of "human-machine communication" research, because in the near future, there will still be a human element in the human-machine system. This estimate was also referenced in his 1960 paper, and Licklider even mentioned that it might take much longer, 500 years, to develop an applicable AI.

Once he started emphasizing the importance of human-machine symbiosis, he had to defend his research agenda from the source of funding—the U.S. Air Force. In 1961, Licklider wrote "The Cerebral Frontier" as "a draft report of the Committee on Artificial Intelligence, Bionics, and Human-Machine Symbiosis to the Basic Research Group of the Air Force Scientific Advisory Board." The Air Force cares about the future of human-machine systems and has at least three agendas to advance them: artificial intelligence, bionics, and human-machine symbiosis. In this competitive situation, he tries to emphasize that his research agenda is the most practical. Acknowledging that there is some overlap between the three agendas, Licklider recommends funding both human-machine symbiosis and artificial intelligence because there are "smart and dedicated young researchers who are professionally trained and qualified for the task, They are eager to explore and assimilate the frontiers of the brain." By "young researchers" he was apparently referring to Ske and McCarthy, and others who had helped him research BBN's future library.

Guiding IPTO

When Licklider became the IPTO's first director, the fundamental consideration for AI in the military was its potential application to automated command and control systems. The terms automated and automated are frequently used in the IDA report that spawned the IPTO in 1961, and the authors of the report looked at artificial intelligence as automated tasks such as problem solving, self-modifying or optimizing systems, and decision-making, etc.) the potential of key technologies. Although the field of artificial intelligence is different from his human-machine systems program, it is a great opportunity for Licklider, who has been struggling to find funding for his "young researchers" - artificial intelligence Possibly a project funded by the IPTO.

Licklider, however, was careful not to commit to early direct applications for automated command and control systems from the funding. In his 1964 paper " Artificial Intelligence, Military Intelligence, and Command and Control ," Licklider didn't say that AI would be the key to automating military systems, but it would. "Solving some fundamental problems in command and control". His reasons are as follows:

There is a clear prospect of substantial progress in AI research, as dynamic representation tools exist, it can evolve significantly beyond its current state, and research acknowledges its value. The prospect that their work could have practical implications for command and control as well as military intelligence is favorable, as researchers in the field of artificial intelligence are working on related fundamental problems, and they are developing tools that are expected to play a role in military information processing.

In 1963, four of the nine organizations Licklider originally funded received budgets for artificial intelligence: Allen Newell's group at Carnegie Institute of Technology, McCarthy's group at Stanford University, heuristic programming and Systems Development Corp. of Computation Theory, and the Minsky Group at MIT for recursive functions, symbolic manipulation, heuristics, and problem solving in Project MAC. Note that all organizations are doing time-sharing at the same time, and Licklider-funded AI is not outside of his Think Center plans; instead, it may support his ideas through console improvements in human-computer interaction.

We can conclude that Licklider's tendency to develop a network of thought centres played a key, albeit informal, role in his overall agenda when deciding what the IPTO should fund. Licklider directed the IPTO to officially fund time-sharing projects (which could serve as the basis for thinking-centre networks) and artificial intelligence (which could improve the human-machine communication needed for these networks). At one level, Licklider established the IPTO agenda to fund projects to meet the military's need for improved command and control systems, but at another level he continued to advance the research agenda of his personal network of thought centers.

IPTO after Licklider's departure

The direction Licklider initially set for the IPTO was strongly influenced by his own research interests. However, as Arthur Norberg and Judy O'Neill write in Transforming Computer Technology, important IPTO priorities include graphics and networking . The latter two initiatives have not been finalized during Licklider's tenure as director. In recent years, as more attention has been paid to the origins of the concept of computer networking, historians and writers have tended to point to Licklider as the inspiration for the Arpanet. But, in fact, neither Licklider's vision for the Thinking Center, nor his call for an interplanetary computer network, have directly shaped the IPTO-funded agenda for networking projects, as we shall see.

Licklider directed the IPTO for two years (1962-1964). Although his research focus at IPTO initially leaned towards time-sharing systems rather than human-computer interaction, Licklider recognized the need to fund human-computer interaction research. Because he spoke highly of Ivan Sutherland's work at MIT on human-computer communication through graphics, he invited Sutherland to become the IPTO's second director. As the IDA report originally explained, the development of "human-machine communications" is critical to the future command and control system of the U.S. military, so it is logical for the IPTO to stress its importance.

Through Sutherland, the IPTO has become more focused on the development of human-machine communications, especially in the field of graphics, which is sorely needed for future military command and control. During his tenure as director (1964-1966), Sutherland set an agenda that firmly established computer graphics as a research focus. So, with research funding from the IPTO, Larry Roberts, who had worked in computer graphics at MIT's Lincoln Laboratory, started working on graphics-related time-sharing systems. In a 1965 paper titled "Graphical Communication in a Time-Sharing Environment" Roberts wrote about the quality of displays connected to time-sharing systems:

Today, it is generally accepted that display devices are an important part of computer systems that facilitate human-computer communication. The advent of large-scale time-sharing operations made previously inefficient online operations economically viable for the first time. This has led to renewed interest in low-cost display methods to improve the ease and speed of interaction. While there's considerable consensus on the value of graphics technology, there's a lot of disagreement over the approach to designing display consoles. This difference reflects not only the necessary differences in application requirements, but also the limited operating experience of the designers. In particular, the extensive experience with online graphics on the Lincoln Laboratory TX-2 computer shows that the general tendency to sacrifice performance for cost reduction is dangerous.

Even in 1965, the contradiction between the concept of time-sharing systems and high-quality human-machine communication was evident. But through the efforts of Robert and others, the two research projects have coexisted to some extent. In directing IPTO funding for Robert's research, Sutherland integrated two previously separate funding priorities: time-sharing systems and computer graphics.

Although both Licklider and Sutherland saw the potential of human-machine communication, Sutherland was not as skeptical of the full automation of human-machine systems as Licklider. As IPTO's second director, Sutherland believes:

The future of online systems depends heavily on the future of offline systems. There's been a lot of talk lately about semi-automated math labs, where mathematicians can prove theorems they couldn't prove without computer assistance. How about letting the computer prove these theorems by itself? Suppose that the artificial intelligence person builds a machine that, in fact, can prove new theorems on its own. So what will our semi-automated math labs look like? It's useless. …I predict that, in the distant future, today’s interest in systems in which people and machines are connected will be replaced by systems in which computers are directly connected to the real world, sensing and interacting directly with the real world through sensors. Of course, the "real world" that interacts with such a system would include humans.

Sutherland dismissed the concept of a "symbiotic" system, which was supposed to be nearly, but not fully, automated. Instead, he tried to change the IPTO funding agenda in favor of advanced hardware design projects, as the advanced design of the TX-2 spurred him to make Sketchpad. In fact, support for hardware research was recommended in the IDA report that originally established the IPTO. Although historians saw this direction as "almost an off-topic direction" for the IPTO, at the time it was a reasonable adjustment.

In computer networking, Sutherland directed and funded a networking experiment at UCLA where he "noted the effect of time-sharing systems on the sharing of programs and other files by users of the same system." Around the same time, Sutherland Lan wrote:

We are just beginning to use computers as a medium of communication between people. Today, by connecting to remote sites, we can allow one person to see another person "over the shoulder" through a computer. We have not yet combined the capabilities of the design system and the query system. Being able to know exactly what's going on, with many people being far apart, has proven useful in airline booking systems. It had to be incorporated into our computer-aided design system.

Apparently, Sutherland highly valued other trends outside IPTO, such as Sabre, who described the networking of computers, including these attempts, as "a medium of communication between people".

During Sutherland's tenure as director, Licklider's earlier research agenda for time-sharing systems is now beginning to produce some complex large-scale systems. However, these IPTO-funded systems are complex, unwieldy, and have many problems. Around the same time, the relatively simple time-sharing system was first marketed outside the IPTO. As Arthur Norberg and Judy O'Neill point out, "By 1965, most computer manufacturers had announced and were developing time-sharing systems." From the 1960s Since the mid-1990s, the business field of time-sharing services has developed rapidly. As Thomas Hughes points out in Rescuing Prometheus , this trend—the distribution of computing power—is analogous to the evolution of electricity distribution. At the same time, however, time-sharing systems are not necessarily favored by all computer users. For example, when minicomputers began to appear in the 1960s, there was conflict in which approach companies should take when choosing computing equipment, as discussed in popular magazines such as Computers and Automation. manifested.

The IPTO appointed a third director in 1966. At that time, the drawbacks of the time-sharing system began to be discussed publicly, and Director Robert Taylor attempted to continue funding the time-sharing system. He noted in 1967:

All time-sharing systems have the potential to gain economic advantage by serving more than one online user through a single computer system. Furthermore, a computer system sufficient to serve multiple users simultaneously provides any one user with more storage space, computing power, and convenience than he can afford on his own stand-alone machine.

For the same reason, he began to emphasize the virtues of the time-sharing system: a community of users communicating through a computer console. In pursuit of this potential, he initiated a networking project that led to the Arpanet. Back then, computers and telecommunications were often discussed together, and computer networks were constructed by connecting computers of the same kind. But Taylor hopes the project will connect the community and make IPTO's case for supporting timesharing more meaningful, independent of Licklider's earlier thinking-centre networking agenda. So the IPTO wanted to conduct a network experiment between the most geographically distant organizations it funded: MIT Lincoln Laboratory (Massachusetts) and SDC (California), and asked Robert to take charge because he was in the Lincoln Experiment Room management TX-2.

According to Robert, he was well aware of Licklider's passion for network time-sharing systems, but when the IPTO sought his expertise in 1966, Robert declined; he was more interested in computer graphics. Ultimately, however, Robert was forced to join the IPTO to direct the networking project, and the IPTO's computer networking agenda was finally launched.

Robert's approach to network experiments differed from Licklider's original vision. Unlike Licklider's network concept, which was a centralized time-sharing system at its core, the Arpanet was designed as a distributed network, and Roberts consolidated the agenda of the previous directors' vision into a resource-sharing network.

We can conclude that the IPTO's network agenda is supported by the time-sharing agenda set by Licklider, although it does not quite follow the network of thought centers he describes. As we all know, the Arpanet became the foundation of today's Internet. Of course, the Internet has become the driving force behind the integration of multiple independent networks into a single network system, so we can't just point to one source. But we can say that Licklider's original vision of sharing knowledge resources through a network of online databases stored in digital computers has been realized on the Internet today.

Compiled from: JCR Licklider's vision for the IPTO