Mar 25, 2020
First published in the Atlantic Monthly in 1945, As We May Think is often cited as a watershed moment in computer history. In this essay Vannevar Bush laid out his vision for machines of the future, improved interfaces, and better data handling methods. It's most remembered today for it's description of the Memex, a theoretical device that implemented something close to the internet, complete with hyperlinks, using 1940s technology. However, there is more to this paper than just Memex. Some of the predictions and recommendations made by Bush would be realized in the coming years, others wouldn't come to be for decades.
One of these predictions is not entirely dissimilar to the idea of a computer network, at least if you use a little imagination. A core feature to Memex and some of Bush's earlier works was microfilm, for a long period of time it was the best way to store large amounts of information. Bush devotes a good amount of As We May Think to describing an idea for essentially a melding of a television and fax machine. The device he describes would take images, transfer them over some type of communication lines, and then reproduce them on microfilm. In this way large amounts of data could be requested from a repository of information and then sent to the requester, similar to a UDP request but using microfilm and fax lines.
One major theme in As We May Think, and Vannevar's writing in general, is an apparent aversion to the pen and paper. To quote from Bush:
"To make the record, we now push a pencil or tap a typewriter. Then comes the process of digestion and correction, followed by an intricate process of typesetting, printing, and distribution. To consider the first stage of the procedure, will the author of the future cease writing by hand or typewriter and talk directly to the record?"
To be clear, he thought the later would be the case in the near future. While As We May Think was being written some of the first "talking machines" were starting to show promise. The Voder, a very primitive device that could turn keystrokes into human-like speech, was shows at the 1939 World's Fair. Bush was in attendance and became instantly fascinated by the device, but he went further with the idea. If a machine could produce speech why not make a machine that can also understand speech? For the time that was a revolutionary idea. Speech recognition has only recently started catching on after considerable effort has been made, but to put forward the idea prior to the widespread use of computers was another matter entirely. But Bush had an interesting stance on the matter.
I can't underline enough how Voder only sounded roughly human, it was much more robotic and had a limited range of sounds it could produce. Instead of predicting steady progress towards more human-sounding machines Bush looked in the opposite direction. He posited that in the near future human language would adapt to be better understood by machines, thus rendering human speech more machine-like. In his words:
"Our present languages are not especially adapted to this sort of mechanization, it is true. It is strange that the inventors of universal languages have not seized upon the idea of producing one which better fitted the technique for transmitting and recording speech."
This prediction had at least half come to pass. Today digitized speech and speech recognition have become common place. A large factor in that has not been the adoption of a universal machine-like language spoken by all humans. Instead it has been thanks to a slow, steady progress to machines that can speak and understand existing human language better.
You can read the full text of As We May Think on the Atlantic's website.
And if you want to hear more about Memex and it's connection to the internet, you can find my episode on the matter here: