5b. The Memex

Introduction to the Memex

Many of the technologies discussed in this course do not have clear genealogies. For example, in Module Three, we discussed how the telephone and telegraph have very scattered origin stories. Yet when scholars look at hypertext and begin to trace citations and ideas, they continue to come back to the same origin story: Vannevar Bush and the Memex.

The Memex was introduced by Bush in a July 1945 article in The Atlantic Monthly, entitled “As We May Think”. If you haven’t read the article, please do so now.

Figure 1: The Memex.
(Bush, 1945) 

Now that you have thought about the Memex, let’s reflect on a few key characteristics of the device.

• It was based on microfilm to store and record information.
• You can use a desk camera to add more material to the Memex.
• You can tie things together between documents: you can draw links between criteria (for example a bow and an arrow), and the link that you have tied between the two concepts.

We will return to the last point shortly, as it lies at the heart of the Memex’s importance for hypertext. To understand the Memex, we need to understand two constituent parts of it.

Vannevar Bush 

The author of “How We May Think” was Vannevar Bush. His personal biography can help us understand the significance of this idea, as well as why it appeared in the Atlantic Monthly. Bush himself would become an outsized figure in American engineering and science, participating and leading projects that would see outcomes as varied as the atomic bomb, the development of the radio, the establishment of the American National Science Foundation, and even – as we discuss in this lesson – the ancestors of the World Wide Web. 

Figure 2: Vannevar Bush.
(Office for Emergency Management & Library of Congress, c. 1940-1944)

Figure 3: Vannevar Bush timeline.  Image description.
© University of Waterloo

The Memex continues to impact both the fields of information retrieval as well as new media. 

To understand the Memex, we need to understand two major factors that influenced its development.

The Two Factors: Differential Analyzer and Microfilm

The reading that you did for this week by Belinda Barnet argues that the Memex’s foundation can be found in the idea of the differential analyzer. As the reading covers much of this ground, I will not go into detail here. Basically, the differential analyzer is a giant calculator. It solves differential equations by a series of wheels and disks that generate mathematical answers: you would have gear ratios between different gears, and through mechanical revolutions, you would be able to do multiplication, addition, subtraction, and more sophisticated forms of mathematical analysis. Or in other words, the differential analyzer is better at mathematics than your professor is!

Bush and another man, Harold Locke Hazen, created the first general-purpose analyzer at MIT between 1928 and 1931. This is important because it meant that it was a machine that was not created for a specific purpose, but rather could be used in a multiplicity of functions. It allows a user to calculate very complicated things far quicker than ever before.

It is a useful connection to understand as the differential analyzer relies to a large degree on physical computing: wheels and disks spinning. The Memex operates under similar principles.

Yet much of what goes into the Memex relies on microfilm. Microfilm dates back to the early nineteenth century, but it became commercialized and successful by the 1920s. Soon, throughout the 1920s and the 1930s, academics are increasingly interested in what microfilm can do to revolutionize studies and the spread of ideas.

What’s microfilm?

 

Microfilm is scaled-down documents on a film strip. Documents are roughly reduced by to 1/25th of their original size, and are then viewed through microfilm viewers which illuminate and zoom them in. Today, in many academic libraries, they are primarily used to consume newspaper and theses and dissertations. But additional uses have ranged from ways to shrink down written letters from soldiers to send home during the Second World War or even volumes of sheet music and other forms of government records.

If you are on the University of Waterloo campus, you can visit a microfilm room in the basement of Dana Porter Library. Any university library or many large public library systems will have microfilm collections, and the rooms that contain these collections are usually some of the densest information sources! Imagine that a library wanted to keep every copy of the Toronto Star ever published: that would take up rooms if they were full size. But the whole run of a newspaper can fit on a few shelves once they have been condensed to microfilm. 

Figure 4: Microfilm from the Dana Porter Library, University of Waterloo.
© University of Waterloo 

 

Some of the early potential of microfilm can be seen throughout the 1920s and 1930s. The Library of Congress in the United States, for example, microfilms some three million volumes from the British Library and brings them back to Washington. Now instead of having to make the long journey across the Atlantic Ocean to London, England, a researcher could now do research from the comfort in the United States. This process begins to repeat itself around the western world, leading to a massive revolution in information sharing and accessibility.

Microfilm is a beautiful analog system.

It is a moment of massive excitement: the idea that information could be nearly universally accessible.

But there’s a problem.

Microfilm is difficult to use. If you haven’t used a microfilm reel before (you are lucky), try to imagine a long Netflix video that you want to find a particular scene in. However, you can only fast forward and fast rewind. So you might fast forward too quickly, then have to go back, and then forward, ad nauseam, in order to find content. Now, of course, imagine that instead of Netflix scenes, there are thousands of indivdual pages, and you can begin to see how this could become a bit frustrating.

The Memex

One of Vannevar Bush’s first ideas in this area is to rise to the problem of finding information in microfilm. He first proposes the idea of the Selector, which would map microfilm to a series of codes. For example, a user could say that they wanted to access information about “cats,” and the machine would then spin microfilm until it found material relating to our furry friends.

This led to a crucial shortcoming. The problem was that while you could find say, cats here and dogs there, other related concepts would be nearby – fur coats (sorry) here, cat houses there, domestication here, the use of human pets in entertainment there, and beyond. You can see the problem. You might use the selector to find cats. But then you would have to zoom out and find domestication separately. Its quicker and more accessible than a stack of books, but it still suffers from the same problem of not finding the complex interconnections within text.

Bush had a crucial point to make in response to this shortcoming.

Figure 5: Example of a semantic network in the human mind. Image description 
© University of Waterloo

Enter the Memex then, which you have now read about. The Memex was an analog system, based on a series of photographing, microfilming, and information retrieval concepts. Crucially, there would be a hypertext component as well.

As I noted above, the most important part of the Memex is the concept of hypertext. Bush doesn’t call it that – the term “hypertext” itself won’t be defined until 1965 – but the basic building blocks are there. A user would link concepts together, and these links were a critical part of the Memex. There is even a rudimentary machine learning idea here: a user would draw connections between people, places, and other concepts, and the Memex itself might learn how to do it as well.

While the Memex was a private network, these links would form a giant record of your actions.