In the spring of 1993, my experiences as a hardware and software engineer collided with the art world as an installation in Kyoto benefitting from the educational direction of edutech pioneer Nobuyuki Ueda. Having worked for too long in the abstract world of electricity and cyberspace, I hungered to see it made real, in order to understand it better. Thus was born the Human Powered Computer Experiment, which was “performed” by Dr. Ueda’s graduate students in a two-day long workshop where human beings embodied the working components of a computer.
At the time computers still had disk drives, so the entire system started up by turning on a power switch, and then placing a giant cardboard disk into a slot through a partition. On one side of the divider sat the computer user; the other half was for the human-powered computation to happen. The performance area spanned two floors of empty space so that visitors could view what was occurring on both sides at the same time.
When the power switch (made out of paper) was flipped to “ON,” the corresponding person playing the “power manager” role was watching and followed a handwritten script:
She was waiting for the disk to be placed into the slot so that information could be read off of the disk’s platter through a window — just like the way a computer would read information from a floppy disk.
As information was being read off of the disk, the bus would then deliver the data to the relevant component — CPU, memory, mouse driver, GPU, etc — to do the work of “booting” the computer to become operational.
A Human-Powered Computer Is Slooooow
The computer was programmed to track the position of the mouse (a cardboard box on a grid) and to display the corresponding location on the computer screen. That means when you moved the mouse, the cursor graphic would move on the video screen in a corresponding way — which happens fluidly on a regular computer. Writing the program was a matter of taping instructions onto the cardboard platter of the disk, and then inserting it into the partition from the user side to the computation side. The disk drive picked up the disk, wore it, and then proceeded to spin the disk platter to the first sector.
The bus took the first instruction packet and then delivered it to the right destinations, and after three hours of delivering subsequent cards of information, it finally went to read an x-y coordinate from the mouse manager. The bus carried that information over to the video manager, and subsequently the screen was updated with the new position. One can imagine how unfavorable it would be for their computer to take three hours to move the position of their mouse on screen. You can watch a reenactment of those three hours — condensed to an artistically reinterpreted three minutes — in this video by Dr. Ueda’s team.
What learnings came from the experiment?
The big takeaway was that computers are extremely fast, and when you try to replicate that with human beings, you see how slow we move by comparison. Another thing we noticed was when one of the first computer programs I wrote “crashed.” But it was okay because we had a good excuse to take a lunch break to refresh ourselves, and then noted how the real computer didn’t need any rest at all.
One outcome that I didn’t expect was how insightfully Professor Ueda’s students — all pursuing Education degrees — could articulate how to improve the computer’s operation with suggestions like this:
- Couldn’t there be two buses instead of just one to speed things up?
- Why wasn’t the video manager closer to the CPU so it waits less often?
They were of course describing technological advances that we’ve come to expect in modern computing architectures, and exhibited their new competency to speak machine. If you wonder why I became so interested in enabling more people to learn how to speak machine, this experiment is how I really began this work.
The underlying motivation for running this experiment came from learning about the Bauhaus and its approach to finding new ways to leverage new electrical, chemical, and mechanical technologies during the 1st and 2nd Industrial Revolutions in service of human beings. The 3rd and 4th Industrial Revolutions are about computing and the need for more people who can speak machine (and human). What are these four Industrial Revolutions? I’ll cover that in my next post.