Augmenting Human Intellect and Douglas Engelbart

I've been learning tools for thought and intelligence amplification/augmentation¹ recently. After researching a bit more, I found myself frequently going back to Douglas Engelbart's Augmenting Human Intellect (AHI), the visionary manifesto on how computers can improve human intellectual effectiveness, and the paper behind the creation of computer mouse, networked computers, hyperlink, video conferencing, collaborative real-time editors and many more.

So, in this "Revisiting Augmenting Human Intellect" series, I want to break down AHI in an easier-to-understand way and draw connections to other things I've learned in adjacent fields like semiotics, cognitive science, machine learning, linguistics, information processing theory, and some philosophy.

My Interpretation of Augmenting Human Intellect

Because my interpretations are based on my understanding of the world and how I apply AHI in my context, this series might not be a direct translation or summary of AHI.

And remember Wittgenstein's Ruler: if you’re unfamiliar with my background, my interpretations of these works may say more about me than the works themselves,

Because this series is still evolving, I'd love to hear your suggestions, feedback, or questions! Feel free to send me a message via twitter, facebook, or email.

You can also subscribe to get new posts when they come out.

Table of Contents

• My Interpretation of Augmenting Human Intellect
• Douglas Engelbart aka the Doug
• Timeline of Important Work in Intelligence Amplification
• Goal of Augmenting Human Intellect

Four stages where human intellectual capabilities evolve

Before we dive into Douglas Engelbart's background and Augmenting Human Intellect, let's think about the following three questions.

Three questions representing the three leaps in our intellectual capabilities

Question 1: Can you determine which side of the fence has more sheep without using any mathematical concepts to count them?

It's challenging, but by using symbols such as the Hindu–Arabic numeral system, we know that the left side has 26 sheep while the right has only 25. Because 26 is greater than 25, the problem is solved.

Question 2: Now that we know the Hindu–Arabic numeral system, can you calculate 213 times 53 in your head?

It's not too easy, but if we write it down on paper, we can visualize the carry and calculate to find 213 x 53 = 11289.

Question 3: Now that we know how to manually calculate using external tools like pen and paper, can you calculate what's 42 to the power of 42?

It's almost impossible. But if we use a computer, we can easily input the number and get the answer automatically. 42^42 = 1.5013094e+68.

The above three questions roughly illustrate the three leaps in the historical progression of our intellectual capabilities, which evolve in the following four stages.

Stage 1: Concept Manipulation

This stage is when we can form and manipulate concepts in our minds without using words or symbols. It's about understanding relationships between objects and events, making predictions, and associating different thoughts.

For example, we can instinctively associate danger when we see a human running away from a chasing tiger without using any words.

And We don't need to know any language to know that the milk below is going to spill and that we would want to hold the glass.

At the Concept manipulation stage, it's like having a mental toolbox filled with unlabeled tools. We understand what each tool does, how they look, and how they interact with others, but we don't know their names. This makes it difficult to communicate with each other and remember things effectively.

Imagine if we had to herd 42 sheep without using any symbols for counting. We would need to remember the exact appearance of each sheep so that if the flock seemed smaller than expected, we could visualize each individual sheep to determine if any were missing. It would be even more difficult if we needed to communicate this to the next person in charge of herding these sheep.

Stage 2: Symbol Manipulation

In Stage 2, we learn how to represent concepts using symbols such as words or numbers. This makes thinking more efficient as it allows us to manipulate concepts more easily.

At the Symbol manipulation stage, it is like giving each tool in our mental toolbox a label. This enables us to organize the tools better and communicate their usage to others.

So now, if we need to herd 42 sheep, we can just remember the number 42 and count them using math. We don't have to remember the appearance of each sheep anymore to know if any were missing.

Stage 3: External Manual Symbol Manipulation

After learning to use symbols to represent concepts, the next stage involves learning to use physical tools and materials to represent and manipulate symbols outside our minds. This significantly improves our ability to solve more complex problems by externalizing our working memory and visualizing concepts with relationships.

At the External Manual Symbol Manipulation stage, it is like taking the labeled tools from our mental toolbox to the physical world so we don't have to remember every single tool in our brain all the time. Because we externalize part of our memory, we don't have to worry about losing our tool just because we forget where we put it in our brain or the information exceeds our working memory capacity.

Therefore, we can now draw a map on the sand with a stick to communicate the relative position of different locations or write a mathematical equation on paper to solve more complex math problems.

Stage 4: Automated External Symbol Manipulation

As we progress, we eventually reach stage 4, where we utilize technology to automate the manipulation of symbols. This allows us to instruct machines, such as computers, to manipulate symbols on our behalf.

At this Automated External Symbol Manipulation stage, it is like working with a robotic assistant who can use the tools with incredible speed and precision, following our commands.

For instance, we can use a calculator to perform complex calculations, a word processor to write and edit a document, or a spreadsheet to analyze data.

This stage has offered, and will continue to offer, countless opportunities for research on enhancing our intellectual capabilities with machines, potentially paving the way for the next phase of human intellectual evolution. And this is where Douglas Engelbart entered the scene and made his mark.

Douglas Engelbart aka the Doug

In the following text, we will refer to Douglas Engelbart by Doug for short as a convention like what I did in HyperLogLog, though I'm not close enough with him on a first-name basis. Hope you (and Mr. Engelbart) won't mind.

Doug was born in 1925. At the end of World War II, 20-year-old Doug was a radar technician in the U.S. Navy, waiting to return home from the Philippines. That year, he read Vannevar Bush's article "As We May Think" in a Red Cross library, which significantly impacted his future work (see his blurry notes on As We May Think).

In 1951, he began designing computer-based problem-solving systems. Doug earned his PhD from the University of California, Berkeley, in 1955 and started a research position at Stanford Research Institute in 1957. Initially, he kept his ideas about enhancing human intellect to himself, fearing they might seem too radical. However, he began to explore these ideas seriously after receiving a small grant from the U.S. Air Force Office of Scientific Research.

"It was lonely work, not having anybody to bounce the ideas off," he said, "but I finally got it written down in a paper I finished in 1962 and published in 1963." He published "Augmenting Human Intellect" as part of his Program On Human Effectiveness in 1963 and demonstrated the "Mother of All Demos" in 1968.

Doug was a pioneer in computer and human-computer interaction (HCI). He either directly created or indirectly inspired the creation of the computer mouse, hypertext, networked computers, version control, word processing, video conferencing, collaborative real-time editors, graphical user interface (GUI), and more.

His work laid the foundation for the entire research field of tools for thought and HCI, inspiring people and companies such as Notion, Bret Victor, Alan Kay, Heptabase, and Logitech to work in this field².

Timeline of Important Work in Intelligence Amplification

I believe that one of the best ways to learn a new field is to study the timeline of the important works because science is built on top of each other³. With the timeline, we learn the context of how each work is related and how progress in other fields might contribute to the success of this field.

To better understand the context, here's a non-exhaustive timeline of some important works in the field of tools for thought and intelligence amplification. More can be found in the work-in-progress History of Intelligence Amplification note.⁴

1. In 1936, Alan Turing published "On Computable Numbers, with an Application to the Entscheidungsproblem", introducing the Turing Machine and halting problem.
2. In 1945, Vannevar Bush published "As We May Think" in The Atlantic Monthly, proposing the Memex, a device that could store and retrieve information similar to how the human brain works.
3. In 1945, John von Neumann co-authored the "First Draft of a Report on the EDVAC", outlining the von Neumann architecture.
4. In 1948, Claude Shannon published "A Mathematical Theory of Communication", laying the groundwork for modern information theory.
5. In 1951, Marvin Minsky designed the Stochastic Neural Analog Reinforcement Calculator (SNARC), the first artificial neural network (ANN) as an early attempt at simulating a network that could learn and potentially enhance intelligence
6. In 1955, John McCarthy coined the term "artificial intelligence" for the Dartmouth Summer Research Project on Artificial Intelligence, which is considered the founding event of AI as a field.
7. In 1956, Allen Newell, Herbert A. Simon, and Cliff Shaw developed the Logic Theorist, which proved theorems in symbolic logic and was described as the first artificial intelligence program.
8. In 1956, William Ross Ashby published "An Introduction to Cybernetics", introducing the term intelligence amplification (IA).
9. In 1960, J.C.R. Licklider published "Man-Computer Symbiosis," proposing a partnership between humans and computers where humans provide the initiative, direction, and integration while machines extend their capabilities.
10. In 1963, Douglas Engelbart published "Augmenting Human Intellect: A Conceptual Framework.", outlining the vision for technology that enhances human abilities to solve complex problems.
11. In 1963, Ivan Sutherland developed Sketchpad in his Ph.D. thesis, "Sketchpad: A Man-Machine Graphical Communication System", as one of the earliest program to utilize a complete graphical interface for interacting with computers.
12. In 1965, Ted Nelson published "hypertext", a non-sequential way of linking information.
13. In 1968, Douglas Engelbart demonstrated "The Mother of All Demos", introducing a complete computer hardware and software system called the oN-Line System (NLS).
14. In 1969, Allen Newell and Herbert A. Simon published "Human Problem Solving", explaining how humans think and solve problems with the information processing theory of human reasoning.
15. In 1972, Alan Kay developed the Smalltalk programming language and proposed the Dynabook concept in "A personal computer for children of all ages."
16. In 1973, Xerox PARC developed Xerox Alto, one of the first personal computers with GUI, WYSIWYG (What You See Is What You Get) text editor, and desktop metaphor.
17. In 1976, Steve Wozniak and Steve Jobs released the Apple I.
18. In 1980, Seymour Papert published "Mindstorms: Children, Computers, and Powerful Ideas", advocating for computer literacy in education and introducing the Logo programming language.
19. In 1982, Autodesk launched AutoCAD, later became an essential tool for computer-aided design (CAD), used widely in architecture, engineering, and other fields.
20. In 1985, Howard Rheingold published "Tools for Thought", popularizing the term and exploring the history of computing for augmenting human thinking.
21. In 1989, Tim Berners-Lee proposed the World Wide Web.
22. In 1993, Kieran Egan published "The Educated Mind: How Cognitive Tools Shape Our Understanding", exploring the historical development of cognitive tools and their impact on human thinking and learning.

Goal of Augmenting Human Intellect

In 1962, Doug sketched out AHI based on the following observations and assumptions:

1. Human population and economic growth are increasing rapidly, but the complexity of problems is growing even faster.
2. Throughout history, complex problems have been solved by groups of people working together.
3. Human intellectual capabilities can be systematically improved by computers and digital tools.

Based on these observations and the assumption, Doug concluded that:

We must use computers to enhance humans' ability to solve more complex problems together.

To address this issue, Doug outlined AHI as a conceptual framework that advocates for further exploration and research to evolve the framework. In other words, it's research for more research. He particularly wants future research programs to achieve the following two goals:

1. Identify what limits the effectiveness of individuals' basic information-handling capabilities in meeting the various needs of society for problem-solving.
2. Develop new techniques, procedures, and systems to improve these basic capabilities for solving the latest problems in a progressing society.

Augmenting Human Intellect is rudimentary. It doesn't provide concrete solutions. Instead, it points out a direction and provides the mental model needed. Therefore, for potential example solutions in the following posts in this series, we will mainly brainstorm with Doug using connections to other great works.

In the next post, I will introduce the intelligence augmentation framework. We will explore the source of intelligence in any system, how Doug approaches the intelligence augmentation problem, the four elements of enhanced thinking, and the four classes of tools that help us think better. Stay tuned!

Thanks to Angelica Kosasih, and Alan Chan for reading the draft of this and giving feedback.