Archive: Computer Collector Newsletter / Technology Rewind, Jan. 2004 - March 2006
Click here to return to archive
My Adventures at SAGE
by Robert Morrisette
San Jose, CA - In 1957, I was one of the army of Boeing engineers working on the initial models of the 707 and KC-135 airplanes. At that time, Boeing was run like an army. Break times were strictly enforced, even for engineers, and were signaled by whistles. Everyone had to line up to punch a time card in and out. I was not happy. One Sunday morning I saw a newspaper ad for computer programmers by Rand Corp., offering free training to learn computer programming for the SAGE project. I hardly knew what a computer was or what SAGE was. I scheduled an interview, passed an aptitude test, sold my house, and moved to Santa Monica, CA. My wife was very nervous, because the new job required a pay cut. We had one son and another child on the way.
SAGE Programming Training
At Santa Monica, the first challenge was to take a assembly language programming course on the SAGE computer, the AN/FSQ-7. SAGE stands for Semi-Automatic Ground Environment. When SAGE started out in the mid 50s, there weren't many programmers around. Rand recruited schoolteachers, taxi drivers, and anyone who would take their aptitude test. Those who passed the test received job offers. The project was responsible for training more than 10,000 programmers in the 1950s.
The training was not easy. About half the students flunked or resigned. Programming was much different in those days. There were several steps necessary to write and run a program:
1. Write the assembly language code on paper coding sheets and submit to key punchers who punch the code on IBM source cards
2. Receive source cards and listing of the cards and verify the listing vs. the coding sheets
3. Submit the program for compiling on the computer
4. Receive the compiled output with error list and correct coding errors (there are always errors)
5. Correct the source deck on the 029 keypunch and submit for recompile. Continue.
6. Go until step three until program runs correctly
The SAGE System
The purpose of SAGE was to detect Soviet bombers and decide what fighters to scramble to intercept them or which missiles to fire. SAGE also controlled the Nike and Bomarc missile. The DEW line sent radar inputs to SAGE. SAGE became obsolete when Russia developed missiles.
The SAGE computer was impressive, especially for mid-1950s hardware. Every computer I worked on after that seems puny in comparison. Later computers were many times faster, but much smaller in size. IBM built the AN/FSQ-7 computer, a military-grade version of the Whirlwind computer. Each AN/FSQ-7 weighed 250 tons, required a 3,000 kW power supply, and needed more than 50,000 vacuum tubes to operate. We tested the vacuum tubes in the standby computer by running higher-than-normal currents through them. The tubes almost ready to burn out would fail and be replaced. The phone bill was around $1 million per month.
Each SAGE site contained two AN/FSQ-7s, one always on standby for quick (13 seconds) switchover. The building housing the computers was a cube three stories high. The radar data came in on phone lines and was stored on 12 150K-word drums. SAGE had a core memory of 65,536 words with a 6-microsecond cycle time that constantly cycled the SAGE program.
SAGE was a 32-bit word machine. The left half was the ops code, and the right half the address. The program was stored on magnetic tape. In those days, and well into the 70s, even after memory was solid-state, computer professionals referred to memory as "core."
SAGE ran 500,000 lines of code. It took six years to develop the SAGE system using 7,000 person years of programming and $61 billion. The last SAGE installation didn't close until 1983.
The project was good for IBM, which built 56 computers for SAGE, earning more than $500 million. More than 7,000 IBM employees worked on the project at one time.
All the SAGE sites were completed in 1961. Each site needed only 15 programmers. There were 20 Direction Center sites, four Combat Center sites, and one combination Direction Center and Combat Center site, in North Bay, Canada.
Each SAGE center could track up to 400 airplanes, and differentiate the enemy from friendly planes by comparing the unknown flight plan coming in on the radar input to flight plans in the system. Many unknowns were bush pilots flying around at about 200 knots.
SAGE Site Installation
After completing the programming course, I was assigned to the site installation and test group. This was very desirable, because we received a 40 percent bonus while on site. We wrote tests to input to SAGE. Most tests consisted of creating dummy radar tracks and flight plans. The site programmers made the necessary fixes, but they were not allowed to change the master tape that came from home base. We tested their changes, and then submitted the changes back to home base to be integrated into the whole system. If it was not an emergency fix, they would accumulate the fixes and then produce a new mod of the software to go out to the sites.
We used a Polaroid camera mounted in front of the console to take a picture of the lights on the console if the system crashed. You could monitor the system from the telephone while your program was running. You dialed a number to listen to the computer and after a while learned how to tell if things were OK or not by the sound of the computer.
About this time Rand created the System Development Corp. (SDC) to run the SAGE project. My first assignment was to install and test SAGE at Truax AFB in Madison, WI. Madison is hot and humid in the summer, and the winters were cold and snowy. Our site bonus started. SDC paid all moving expenses, even including installing TV antennas.
While at Truax, I became the first person to become lost inside a computer. I was wondering around one day and was totally lost and could not find an exit. After about an hour, an IBM engineer led me to safety.
(In those days, all IBM male employees were required to wear a dark suit, a white shirt, and tie. An IBM employee in my car pool forget his tie one morning. He had us drop him off at a store to buy a tie. He was afraid to show up for work without a tie.)
My next assignment was at McChord AFB in Tacoma, WA. After McChord, I had the choice of going to North Bay in Canada, which is inaccessible by road or airplane in the winter, or back to the Santa Monica home office. You can guess which I chose.
Programmers with SAGE experience were in great demand, so I did not stay with SDC very long. After leaving SDC, I worked for more companies than I should have, including Sperry Rand, Control Data, Santa Clara County, IBM, Digital Research, Motorola, and Sun. I spent the final 20 years of my computer career as a writer and editor. The stories I can tell!
For many years, the word around the industry was, "It was first done in SAGE." SAGE was the first real-time, command and control computer-based system with capability so advanced that 50 years later, today, some of that capability can still be called state-of-the-art.
Some of the concepts pioneered by SAGE:
- Forrester's magnetic core memory
- Forrester's highly improved vacuum tubes
- Duplexing (linking two computers to share data, one could go down)
- Digital data transmission over phone lines
- Modems and long-distance data communication
- Interactive video displays with light guns
- Graphic display techniques
- Real-time digital control
- Real-time database management
- Distributed processing
- Digitized radar data
- Long-distance data communications via land lines
- Ground-to-air radio from up to 100 radar and observation stations
SAGE and the Internet
SAGE was a precursor to the Internet. It is commonly acknowledged that the Internet Age began in 1969 when the U.S. Department of Defense implemented the Advanced Research Projects Agency Network (ARPANET), whose purpose was to demonstrate how communications between computers could promote cooperative research among scientists. But the foundation for ARPANET, and the Internet Age, actually began forming in 1954 with the development of SAGE.
A very important spin-off of SAGE was the SABRE airline reservation system.
If you want to learn more about SAGE, here are some references:
- Google or other search engines have many references to SAGE
- "On Guard, the Story of SAGE" is a movie made by IBM
- "From Whirlwind to Mitre: The R & D Story of SAGE" by Redmond & Smith is an excellent book. Published by MIT Press.
- "The SAGE Air Defense System," by John Jacobs, published by Mitre