Events and Sightings

MIT 150th Anniversary and MIT Museum Project Whirlwind Presentation

The Massachusetts Institute of Technology (MIT) celebrated its 150th anniversary in 2011 (http://mit150.mit.edu). The celebration scheduled between 7 January and 5 June 2011 included hundreds of special events, symposia, publications, and a major exhibition at the MIT Museum.

The MIT 150 Exhibition (http://museum.mit.edu/150) opening at the MIT Museum was the inaugural (7 January) event of the institute's celebration. Running through 31 December 2011, the exhibition consisted of 150 objects representative of MIT's history and ambitions for the future. The exhibition curator, Deborah Douglas, noted that the list of 150 objects was determined by "crowd sourcing" from the MIT community rather than the typical curator-driven approach. Naturally, a number of the objects fell under the exhibition's Analog/Digital MIT theme.

Throughout the year, the MIT Museum also had many presentations on a variety of topics, including the 15 October public event that I attended, entitled "Project Whirlwind, Sage, and Pioneering MIT Computer Projects." A couple dozen participants from the Whirlwind development era in the late 1940s and 1950s were present for the presentation, making the event also sort of a Project Whirlwind reunion.

The primary focus of the event was a panel session with Project Whirlwind leaders Jay Forrester and Robert Everett (see Figure 1), moderated by museum director John Durant.

Durant's first question for Forrester was, "What was Project Whirlwind?" Forrester explained that Whirlwind had a varied history. At first it was to be an analog computer for predicting the controllability of future airplanes— unlike the Link Trainer, which was for training pilots of an existing airplane. After a year or so, they concluded the future-airplane task was not possible with an analog computer. At that point Perry Crawford, who was with the US Navy's Special Devices Division, which was in charge of project oversight, suggested they switch to digital computing. Bob Everett remembered that one day Forrester came by and said, "'We are now working on a digital computer,' and I said, 'What's that?'"

Forrester talked about the risks inherent in the project. Whirlwind was transferred to the Office of Naval Research, which generally spent only enough money on a project to support one mathematician and an assistant. The Whirlwind project needed 100 times as much money, so funding involved an "annual inquisition." Some people in the Electrical Engineering Department thought computers needed to do decimal arithmetic, and the Whirlwind team had to argue for the efficiencies of binary arithmetic. The average life of a vacuum tube was 500 hours, and their machine would have tens of thousands of vacuum tubes, which if you do the arithmetic, meant major reliability problems.

Some useful prior efforts existed, however. Coming out of World War II, the MIT Radiation Laboratory had knowledge of pulse circuits and vacuum tubes. The Whirlwind team knew of the early computer work at Harvard and of work with EDVAC at the University of Pennsylvania where John von Neumann pushed the idea that a computer could run on a program stored in its own memory—"a very big breakthrough," said Forrester. (The EDVAC was a serial system, but Whirlwind needed a parallel computer system to handle the speed required for the real-time work they had in mind.)

Mercury delays lines were one memory possibility, but it took a millisecond for a bit of data to cycle from one end of the delay line to the other. A Williams tube using a 2D grid on a cathode-ray tube was another possibility, but they were unreliable. Forrester said that he wanted a 3D storage system and first thought about using glow discharge tubes, but he gave that up as impractical. Forrester went through another idea or two before settling on ferrite cores (little doughnuts of ceramic magnetic material) arranged in an a 2D array with interconnecting wires so any individual core was instantly accessible. (Multiple such core planes provided access to the bits of a word of computer memory.) Forrester explained that a guy in New Jersey could occasionally produce a ferrite core that...