Augmentation before automation: the founding argument for the interface

The field is usually taught as a list of firsts: a light pen, a mouse, an outline editor, a window. The list is accurate and explains very little. The same twenty years produced a contrary idea, that the machine should take the work over. This unit argues that the interface was founded on the other idea: the founding claim was about who does what, not about hardware.

That claim is a division of labour: the person keeps the goal, the judgement and the criteria, the machine takes the work that reduces to a routine. It rests on limits the next unit examines, and it is contestable.

The argument was about who does the work

Licklider's 1960 paper states the claim most clearly. He called the arrangement man-computer symbiosis and defined it as a close coupling in which the human sets the goals, formulates the hypotheses, determines the criteria and performs the evaluations, while the machine does the work that can be routinised. He named two alternatives: in mechanically extended man the machine extends the arm or the eye and the human supplies the initiative; in the humanly extended machine the system was meant to be automatic and fell short, so the people who remain help it rather than being helped by it. A call centre script that the worker must read aloud belongs to the second kind.

Licklider's evidence was a time-and-motion record of his own work: he counted about 85 per cent of the hours he called thinking time as time spent getting into a position to think. He served as his own subject, so the figure describes one person and not a population.

The paper also carries a qualification that retellings drop. Licklider expected machines to outdo the human brain in due course, citing an Air Force study group estimate that allowed until 1980 for machine thinking of military significance, and gave the partnership five years to develop and fifteen to use.

Bush: the value is in the trail, not in the store

Fifteen years before Licklider, Vannevar Bush described the memex, a desk in which a person keeps books, records and correspondence on microfilm and consults them at speed. The memorable part is not the store but the trail: a reader ties two items together, names the trail and can follow it later. Bush wrote that the process of tying two items together is the important thing. Nelson's 1965 paper set out a file structure for the personal side of that memex, keeping alternatives and variants without disturbing the original, and named it hypertext.

Engelbart: the whole working system, not the machine, is the unit of design

Engelbart's 1962 report turned the claim into a framework, defining the unit of augmentation as a trained human being together with language, artefacts and methodology. He argued that the four have to be redesigned together, because a change in one alters the value of changes in the others.

The report carries a worked case of its own. Engelbart follows an architect at a station with a display about three feet square, a keyboard and a pointer, the computer acting as his clerk. He marks two points on a perspective view, and the system reports the distance and elevation between them. He keeps lists of specifications that grow into an interlinked structure as the design develops. Engelbart labels the passage plainly: he offers it as fiction, not as a description of the system the programme would build. It shows what the joint system was meant to feel like, and it pictures one kind of user, a sighted professional at a large display with a pointer and two steady hands. The user who works through a switch or a screen reader is not in the picture.

The one experiment in the report runs the other way. To show what a change in the artefact does to the whole system, Engelbart's group fastened a pencil to a brick and wrote with it. Against typewriting and ordinary pencil writing they were slower and less precise, had to write larger to write faster, and tired more quickly. Change the tool, and the pace, the precision and the fatigue of the work move with it.

Bardini's history (2000) is the scholarly check. Reading the laboratory record, he argues that the project was co-evolutionary: tools, language and practice were meant to advance together, and industry took the artefacts while discarding the programme.

The demonstration, and what it actually showed

Sutherland's Sketchpad was his doctoral work at MIT in 1963 (Myers, 1998): a person drawing with a light pen and altering the drawing by pointing at it.

Engelbart and English reported the built system in 1968. One time-sharing computer served twelve display work stations, each with a keyboard, a five-key handset giving 31 chords, and, in the paper's words, an SRI cursor device called the mouse. The authors report their own comparison, in which the mouse beat the light pen and the joystick, and describe a conference room where twenty people at a time watched one controlled view.

That last detail is what a clean origin story cannot absorb: the demonstration that the founding argument could be built was a group system.

A medium you write in, not a service you receive

Kay and Goldberg's 1977 paper set out the test that this unit keeps returning to: a medium becomes personal when its owner can author in it. Their report describes the Dynabook, a notebook-sized machine that could be owned by everyone, and the Smalltalk system on interim versions of it, listing programming, an interactive memory, a text editor, and expression through drawing, animation and music. Two histories widen the frame: Waldrop (2001) follows the research funding Licklider organised and argues that interactive and personal computing came from a public research programme, and Markoff (2005) traces the tradition through the laboratories and communities around them.

Contested question: technical progress, or the same social problem in new hardware?

Grudin's 1990 paper is the challenge to this unit's thesis. He surveyed the field from the first commercial systems to his present and argued for continuity in the movement of the interface outward, identifying five levels from hardware to the work group, each requiring different skills, with most development sitting at the third.

His 2017 book puts the position in one sentence: technology has changed rapidly, human nature very little. On that account the field inherits a recurring social problem, and augmentation is a story told about the hardware. Myers (1998) holds the other position: a chain of dated demonstrations runs from research at universities and laboratories to the interfaces commerce later sold.

His own chain also dates the transfer, and the interval is the test. Sutherland demonstrated constraining and manipulating objects with a light pen in 1963, the mouse was built in 1965, overlapping windows appeared in Smalltalk in 1974, and the first systems to use them arrived with the Xerox Star and the Apple Lisa in 1981 and 1982. Eighteen years separate the demonstration from the product, and the laboratories that sold it were not, in most cases, the ones that built it.

The two positions need different evidence. Grudin's is a comparison of problems across decades in comparable settings, not a count. Myers's is a dated list of demonstrations, self-limited to the computer half of the field, which begins the accounting of what reached practice and does not finish it. Neither side has measured what it disputes, and which ideas, funds or markets travelled in the eighteen-year interval is not recorded. The strongest counter to the unit's founding claim is that the argument was not what the field inherited: Bardini (2000) finds the industry taking Engelbart's artefacts and discarding his programme, and Grudin (2017) finds the recurring problems social rather than technical.

A third disagreement sits inside the founding texts. Licklider (1960) treats the partnership as an interim before machine dominance. Engelbart (1962) treats the joint system as the permanent object of design. What would settle it is empirical: in observed work, does the machine's share of the routine extend the person's formulative work, or take the formulation as well? The first pooled answer is unwelcome to both. Vaccaro, Almaatouq and Malone (2024) combined 106 experiments reporting 370 effects: a person working with an AI did worse, on average, than the better of the two working alone, losing on decision tasks and gaining on content creation.

A designer cannot settle that dispute and does not need to: both accounts agree that the interface fixes part of the user's working method, so the same design stands whichever is right. Keep the machine's share of the routine visible and reversible, and keep the user able to author.

Where the argument is heading

Grudin's closing position is the one to watch. Tools built for their users increasingly shape how those users work, sometimes in ways that were planned and sometimes not, and he holds that the past patterns of the field are the best guide to what comes next. If he is right, the test of the founding argument is still Kay and Goldberg's: whether the user keeps the goal and the ability to author.

The sources this unit rests on

• Bush, V. (1945). As We May Think. The Atlantic Monthly, July 1945. https://web.archive.org/web/2024/https://www.theatlantic.com/magazine/archive/1945/07/as-we-may-think/303881/ The essay proposed the memex, a desk of linked microfilm on which a reader follows trails between documents, and named the growth of the scientific record as the problem. A proposal, not a study.
• Licklider, J. C. R. (1960). Man-Computer Symbiosis. IRE Transactions on Human Factors in Electronics, HFE-1(1), 4-11. doi:10.1109/THFE2.1960.4503259 Licklider set out a partnership in which the human keeps the goals, hypotheses, criteria and evaluations while the machine does the routinised work, separating it from the mechanically extended man and the humanly extended machine. Placing it before an expected period of machine dominance makes the paper both the unit's thesis and its sharpest qualification.
• Engelbart, D. C. (1962). Augmenting Human Intellect: A Conceptual Framework. Summary Report AFOSR-3223, Stanford Research Institute. https://www.dougengelbart.org/pubs/augment-3906.html This report defined the unit of augmentation as a trained human together with language, artefacts and methodology, and argued that the four are redesigned together, the founding text for the claim that interface and method change as one system.
• Sutherland, I. E. (1963). Sketchpad: A man-machine graphical communication system. PhD thesis, Massachusetts Institute of Technology. Electronic edition: University of Cambridge Computer Laboratory Technical Report 574, 2003. https://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-574.pdf The thesis built a drawing system in which a person points at the representation and changes it; Myers (1998) records the operations it supported, including constraints.
• Nelson, T. H. (1965). Complex information processing: a file structure for the complex, the changing and the indeterminate. Proceedings of the 20th National Conference of the ACM, 84-100. doi:10.1145/800197.806036 Nelson proposed a linked file allowing intricate arrangements and undecided alternatives, which he named hypertext. Working from Bush's memex, it makes structure, not content, the thing the writer manipulates.
• Engelbart, D. C. and English, W. K. (1968). A research centre for augmenting human intellect. Proceedings of the AFIPS Fall Joint Computer Conference, part I, 395-410. doi:10.1145/1476589.1476645 The paper reports the built system: one time-sharing computer serving twelve work stations with a keyboard, a chord handset and a mouse, and the authors' comparison in which the mouse beat the light pen and the joystick. The conference room it reports, where twenty participants watched one controlled view, makes the evidence that the argument could be built a group system.
• Kay, A. and Goldberg, A. (1977). Personal Dynamic Media. Computer, 10(3), 31-41. doi:10.1109/C-M.1977.217672 The authors describe a notebook-sized personal medium, owned by its user, and the Smalltalk system on interim versions of it, listing programming, an interactive memory, a text editor, drawing, animation and music. Their test of a personal medium is authorship by the owner.
• Grudin, J. (1990). The computer reaches out: the historical continuity of interface design. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '90), 261-268. doi:10.1145/97243.97284 Grudin surveyed interface research and development from the 1950s and argued for continuity in the interface's movement outward, from hardware to software to cognition and finally to social processes, with different skills required at each of five levels.
• Myers, B. A. (1998). A brief history of human-computer interaction technology. Interactions, 5(2), 44-54. doi:10.1145/274430.274436 Myers dated the technical history of interaction, from Sketchpad in 1963 and the mouse in 1965 to tiled windows in 1968, the Bravo editor of 1974 and the commercial systems from 1981, and argued that research produced what commerce sold. His own limit: the paper covers the computer half of the field.
• Bardini, T. (2000). Bootstrapping: Douglas Engelbart, Coevolution, and the Origins of Personal Computing. Stanford University Press. https://archive.org/details/bootstrapping00thie Bardini reads Engelbart's laboratory work as a co-evolutionary project in which tools, language and practice advance together, and argues that industry took the artefacts while discarding the programme.
• Waldrop, M. M. (2001). The Dream Machine: J. C. R. Licklider and the Revolution That Made Computing Personal. Viking. https://archive.org/details/dreammachinejcrl00wald Waldrop follows Licklider and the research funding he organised, arguing that interactive and personal computing came from a deliberate programme built with public research money: evidence that the argument was funded and acted on.
• Markoff, J. (2005). What the Dormouse Said: How the Sixties Counterculture Shaped the Personal Computer Industry. Viking. https://archive.org/details/whatdormousesaid0000mark Markoff traces the tradition through the laboratories and communities around them, opening with a chapter on augmentation, and argues that augmenting the individual carried a claim about who computing was for. It is history, not a study, so it is cited for the argument.
• Grudin, J. (2017). From Tool to Partner: The Evolution of Human-Computer Interaction. Morgan and Claypool. doi:10.2200/S00745ED1V01Y201612HCI035 Grudin gives a chronological history of the fields that contributed to HCI, claiming that technology has changed rapidly while human nature has very little, and that tools built for their users increasingly shape how those users work. It lets the unit state continuity.
• Vaccaro, M., Almaatouq, A. and Malone, T. (2024). When combinations of humans and AI are useful: a systematic review and meta-analysis. Nature Human Behaviour 8(12), 2293-2303. doi:10.1038/s41562-024-02024-1 A preregistered meta-analysis of 106 experiments reporting 370 effects: a person working with an AI did worse on average than the better of the two working alone, losing on decision tasks and gaining on content creation. The unit's first pooled test of its founding claim.