Joseph "Lick" Liklider: "The symbiosis of man and computer" (1960), part 1

Alan Kay, Douglas Engelbart and other pioneers of the computer age declare that “the information revolution has not yet begun” [ 1 ], [ 2 ]. In the sense of? What do they mean? It seems that they compare the current state of affairs with their internal picture, as it could be, “if you did everything exactly what we were talking about, not the 5% that you could barely realize.”

I invite everyone to take part in collective translation and discussions of what could be (could be) the "information world", how to strengthen the intellect of humanity and how to build an image of the future in which I would like to live.
')

“Anyone and any company using a computer interactively should be grateful for Lika”
- Bob Taylor, founder of Xerox PARC and founder of the DEC research lab

Joseph "Lick" Liclider "invented" the computer era, was the ideological inspirer of Marvin Minsky , John McCarthy , Douglas Engelbart , Ivan Sutherland , Robert Taylor . (Biography here )

I present to you the first part of the article, thanks to which Liclayder was invited to ARPA to implement his projects.




D.K. R. Lyclayder
IRE Proceedings on the Human Factor in Electronics,
HFE-1 Issue, Pages 4-11, March 1960

Short description

The symbiosis of man and computer is the expected stage of development of interaction between people and machines. He organizes a close relationship between people and electronic participants in business collaboration. The main objectives are:

1. give computers the opportunity to simplify the process of formalized thinking as today they simplify work with algorithmic tasks, and
2. allow people and computers to make joint decisions and manage difficult situations without hard dependencies on pre-installed programs.

The expected symbiotic cooperation implies that the person will set goals, formulate hypotheses, define criteria, and give an assessment, and computers will perform important routine work that will lead us to understanding, and will also push solutions in the field of science and technology.

A preliminary analysis showed that in symbiotic cooperation, intellectual operations will be performed much more efficiently than humans do. To achieve such an effective alliance, it is necessary to develop a time sharing system, memory components and its organization, programming languages, and input / output devices.

1. Introduction

1.1 Symbiosis

The fig tree is pollinated only by the insect Blastophaga grossorun, the larva of which lives in the ovary of the tree, where it receives food. Thus, the tree and the insect are in an interdependent position: the tree cannot carry out the process of reproduction without an insect; an insect without a tree cannot find food; together they organize sustainable, productive and thriving collaboration. Such an interaction of “living together in close connection of two organisms of different species” is called symbiosis [27].

“The symbiosis of man and computer is a subclass of man-machine systems, of which there are many. However, at the moment there is no symbiosis system between man and computer.

The purpose of this article is to familiarize with the concept and stimulate the development of human-computer symbiosis through analyzing some of the problems of human-computer interaction, paying attention to suitable principles for designing human-computer systems, and determining a certain number of questions that need to be answered after preliminary research. .

It is hoped that in the not too distant future, the human brain and the computer will be very closely interconnected, and the result of such cooperation will be thought in a way that no human brain has done before, and operate on these data in a way different from that use machines known to us today for information processing.

1.2 Between "Mechanically Improved Man" and "Artificial Intelligence"

Man-machine symbiosis as a concept is significantly different from what North [21] called "a mechanically improved person." In the man-machine systems of the past, the operator was responsible for taking initiative, managing, integrating, and setting evaluation criteria.

The mechanical parts of the system acted only as additions, first for human hands, and then for the eyes. Of course, these systems did not include "organisms of different species living together." There was only one type of organism present - man, and everything else was needed only to help him.

Of course, in a sense, any system created by man is designed to help him or other people outside this system. If we consider a human operator within the system, we will see that in the last few years fantastic changes have occurred in some areas of technology. "Mechanical addition" led to the substitution of man, to automation, and those people who remain, for the most part, are beneficial and do not need help.

In some cases, especially in large computerized information systems and control systems, operators perform mainly those functions that cannot be automated.

Such systems (North would call them "machines supplemented by human capabilities") are not symbiotic. They are "semi-automated", i.e. these are systems that were conceived as fully automated, but which could not achieve the goal.

Perhaps the symbiosis of humans and computers is not the main paradigm for complex technological systems. It is quite probable that at a certain stage electronic or chemical “machines” will surpass the human brain in most functions, which we now consider exclusively its prerogative.

Even now, the Gelernter program for proving theorems on planimetry on the IBM-704 is progressing at about the same speed as students at the Brooklyn school do, and makes similar mistakes. [12]

In fact, there are several programs for proving theorems, making decisions, for playing chess and recognizing images (there are too many of them to list everything [1, 2, 5, 8, 11, 13, 17, 18, 19, 22, 23, 25]), able to compete with human intellectual capabilities in a limited number of areas. And the “universal problem solver” created by Newell, Simon and Shaw [20] can remove some limitations. In general, one should avoid discussions with (other) ardent fans of artificial intelligence about the fact that in the distant future we will give up the palm in the field of mental activity to machines. However, quite a lot of time will pass during which the main intellectual discoveries will be made during the course of collaboration between people and computers.

A multifunctional research group studying future tasks to be solved by the air force in research and design activities has calculated that a level of artificial intelligence development in which the machines will be able to independently reason or make decisions of military significance will come no earlier than 1980. We have, say, 5 years to develop human-machine symbiosis and 15 years to use it. It may be 15, or 10, or 500, but these years in intellectual terms will be the most creative and exciting in the history of mankind.

2. The goals of man-machine symbiosis

Today's computers are designed primarily for solving previously formulated tasks or for processing data in accordance with predefined programs.

The sequence of calculations may vary depending on the results obtained in the course of calculations, but all possible options must be determined in advance. (If an unforeseen situation arises, the whole process stops in anticipation of the necessary additional program) The requirement of a preliminary formulation or definition is sometimes not such a significant drawback.

It is often said that programming for a computer forces one to clearly build thoughts that it disciplines the thinking process itself. If the user can carefully consider his task in advance, then a symbiotic connection with the computer is not needed.

However, many tasks that could have been thought out in advance are too complex for this. They would be easier and faster to solve through an intuitive trial and error method, in which, in collaboration with a computer, errors in reasoning would be revealed or unexpected changes would be detected in the search for a solution. Other tasks are simply impossible to formalize without the help of computer technology.

Poincare predicted frustration for an important group of potential computer users when he said: “The question is not what the answer is, but what the question is.” One of the main goals of man-machine symbiosis is the effective implementation of computers in the algorithmic parts of technical problems.

The other main goal is closely related to the first. It consists in effectively integrating computers into the process of thinking that should occur in the “real time” mode, i.e. time that flows too fast to be able to use computers in the usual way.

Imagine, for example, that you are trying to wage a battle using a computer in the following scenario. Today you are formulating a task. Tomorrow you explain it to the programmer. Next week, the computer spends 5 minutes building your program and 47 seconds calculating the answer. You get a sheet of paper 20 feet long, all filled with numbers, which, instead of presenting a final solution, only offer some tactics that need to be checked through modeling.

Obviously, the battle will end before you take the next step. Reasoning when interacting with a computer as well as reflections with a colleague, whose knowledge complements yours, will require a much closer connection between a person and a machine than is implied in the example, and than is possible today.

3. The need for computer participation in the formulation of tasks and the thinking process in real time

In the previous part, it was implicitly assumed that, if they could be effectively incorporated into the thought process, the functions that data processing machines could perform would significantly improve or complement the process of thinking and decision making. Such an assumption may require justification.

3.1 Preliminary and informal analysis of the performance of technical thinking

Despite the fact that there are a large number of books on the subject of thinking and decision making, including a comprehensive study of the process of invention, I could not find anything similar to the analysis of the performance of the mental work of a person taking part in a scientific or technical project.

Therefore, in the spring and summer of 1957, I dedicated to torture to track down what a person with an average level of involvement in technical details actually does during all working hours. Although I was aware of the insufficiency of my sample, I decided to become my own subject.

It soon became apparent that the only thing I did was to keep records, and the project would turn into an endless movement backward if the reporting was carried out with all the details provided for in the original plan. But this was not the case. However, I saw a picture of my activity that made me think. Perhaps my range is not typical - I hope so, but I'm afraid that it is not.

About 85% of the time “thinking” was spent on finding a position that was convenient for reflection, on making decisions, on learning what I needed to know. Much more time was spent searching for or extracting information than comprehending it. A lot of hours went on the construction of graphs, and another pile on instructing an assistant on how to do it.

When the charts were finished, the dependencies immediately became obvious, but in order for these to become possible, you had to build graphs. At some point, it was necessary to compare 6 experimentally derived definitions of a function that relates clarity of speech to a speech-to-noise ratio.

None of the experimenters used the same definition or measured speech / noise ratio. It took several hours of calculations to present the data in a convenient form for comparison. When this form was received, it took only a few seconds to select what I needed.

If I briefly describe the results of my research, it turned out that my time for “thinking” was spent mainly on clerical or mechanical activities: searching, calculating, plotting, transforming, determining logical or dynamic conclusions from a number of assumptions or hypotheses, preparing for making decisions or understanding.

Moreover, my choice of why to make efforts, and why not, was largely determined by the degree of feasibility of clerical tasks, and not by intellectual possibilities.

The main assumption expressed in the above results is that operations that take up most of the time, supposedly aimed at thinking about technical details, are operations that machines can perform better than people.

The fact that these operations should be carried out on different variables and based on an unpredictable and constantly changing sequence of actions sets us serious challenges. And if they can be solved by creating a symbiotic connection between a person and a machine with fast information retrieval and data processing, then it becomes obvious that such an interaction will significantly improve the thinking process.

Here it is appropriate to recognize that we use the term “computer” to refer to a wide class of machines: computing, processing data, as well as information storage and retrieval. The capabilities of this class of equipment are increasing every day. And therefore, making general statements about the possibilities of a class is quite dangerous.

It is possible that making general statements about people's opportunities is just as dangerous. However, certain genotypic differences in the abilities of people and computers really stand out, and they are related to the nature of the possible symbiosis of man and computer and to the potential value of its achievement.

As already mentioned, people are noisy, narrow-band devices, but their nervous system has a large number of parallel and simultaneously operating channels. With respect to people, computers work very quickly and accurately, but they can only perform one or a few basic operations at a time.

People in this regard are more flexible and able to “program themselves according to circumstances” on the basis of recently received information. Computers are set to perform one specific goal, and are limited to their "pre-compiled program."

It is natural for people to speak redundant languages ​​based on unitary objects and coherent actions using 20-60 elementary symbols.

For computers, it is “natural” to speak non-redundant languages, usually with only two elementary symbols, without the ability to distinguish between unitary objects or coherent actions.

If we are to be completely correct, then these characteristics would have to include many parameters. However, the differences (and, therefore, possible additions) that these characteristics describe are indisputable.

Computers can perform easily, quickly and well a lot of things that are difficult or impossible for a person, but people are able to easily and well, although not so quickly, perform many tasks, difficult or impossible for a computer.

This leads us to the fact that a symbiotic interaction, provided that the positive characteristics of people and machines are successfully integrated, would be of tremendous value to us. Of course, at the same time we need to overcome the obstacle in the form of a difference in speed and language.

To be continued...

Candidates for translation

Vannevar Bush
As We May Think (the first part is translated by us on Habré )
Letter to Roosevelt: Science The Endless Frontier
Science Is Not Enough

Joseph "Lick" Liklider
"Man-Computer Symbiosis" (translated: Part 1)
Memorandum for Members and Affiliates of the Intergalactic Computer Network (translated)
"The Computer as a Communication Device" ? 1968

Douglas Engelbart
Augmenting Human Intellect: A Conceptual Framework , 1962
Improving our abilities: A new future , 2002

Alan Kay
The Power Of The Context
The Real Computer Revolution Hasn't Happened Yet

What to read more in Russian

Especially for Habr: interview with Alan Kay
Alan Kay: The future of "reading" depends on the future of "learning difficult to understand things"
Vanivar Bush: “As We May Think” (As We May Think)
Douglas Engelbart: "The Mother of All Demos". Part 1