One Small Cubic Foot

agc

It is often said that man’s ability to dream exceeds his grasp. But how far would people go chasing after their dreams? In the 1960’s such an extraordinarily ambitious dream pushed a brightest band of technological genius at Massachusetts Institute of Technology (MIT), to the edges where reality and science fiction overlapped, and led them to rediscover the Computing as it was known.

In September 12, 1962, the late U.S. President John F. Kennedy delivered a speech at Rice University in Texas. A speech that inspired all Americans on a singular goal of “Landing a man on the moon and returning him safely to the earth”. It immediately gripped the people by a nationalistic fervour intensified by the consistent drawbacks of the U.S Space Program against its Soviet counterpart. As the U.S space engineers were still struggling to get even the most primitive rockets off the launchpad, Kennedy’s challenge seemed daunting and even near impossible for most of them.

One such group who underwent an enormous amount of pressure was the staff of Instrumentation Laboratory at MIT. Headed by the charismatic and influential scientist Charles Stark Draper, MIT has already revolutionised the aerial navigation technology by inventing a revolutionary method called Inertial Navigation System which out performed all the other guidance systems at the time. NASA, recognizing the need for an extremely precise and accurate navigational platform over the span of 300,000 kilometers between Earth and Moon, decided to handover the contract to Draper’s Lab. The cruciality of the task lay on the fact that the lives of astronauts hinged on it. To make things even more complicated NASA required the entire guidance platform to be based on an almost dream machine, a digital computer called the Apollo Guidance Computer (AGC).

Back in the 60’s computers were humongous machines taking up entire floors of buildings based on Vacuum Tube technology as the key component of their logic circuit design. The imposing challenge for the engineers working on Apollo Guidance Program was to shrink the entirety of this into something that would fit inside a capsule no larger than a couple of telephone booths. As the solution they turned toward a newly emerging technology of unknown capability, the Integrated Circuits (IC). Back then, the process of manufacturing ICs were exhaustively difficult and expensive. Even so, it delivered so little reassurance as to its reliance. The experimental technology would allow the AGC to be compacted into a “One cubic foot unit”, as the technical leader of MIT instrumentation lab, Richard Battin, used to called it. This eventually overcame the issue of size but spawned another hurdle on its own, low processing power. The AGC was becoming extremely slow and cumbersome to perform the complex calculations required for space flight. Though the origin rested partially on the limited ability of ICs, the methodology computers were scheduling their tasks itself was also playing a crucial role. The way computers were operating back then was a technology called Time Sharing in which the computer was given equal slices of time of a system clock to perform various tasks. The fault lay in there. When the computer was handed too many tasks to be completed simultaneously this technique led to a ground still, halting the computer. The simple but elegant solution came from the brilliance of the engineer Halcombe Laning who proposed that tasks should be prioritised. So for instance crucial tasks like engaging thrusters to control the attitude of the spacecraft could be of high priority than updating the lights of an indicator panel. The cleverness of the technique was then, when an overload occurs the computer would drop lower priority tasks and proceed on with high priority ones.  

While the hardware of the AGC was taking shape, software which would run on them became a growing concern. In an age when even the word Software was freshly invented, a pioneering breed of Software Engineers at Drapers lab headed by Margaret Hamilton took on the task of writing software routines, in a form of assembly language, for a machine which has not even being built yet. Once the code has been written it was transferred to a set of Punch Cards that would be fed into a computer to generate test outputs. If the routine passes the test, the software would be written into what is called a Core Rope Memory to be stored on the AGC. Core rope memory was another invention that was used to store binarized code which consisted of an array of ferrite loops and hundreds of threads of copper wire. If the wire goes through the loop it represented one, if it wounds around it that was defined as zero. This extremely crude representation required a bizarre method of “writing” it too. Once a code was tested it was handed down to tuplets of women who would sit face-to-face with the loop frame between them. Then, they would literally weave copper wires with needles through and around ferrite loops to sew the program into memory. This approach seems bizarre and over complicated when compared with modern standards and in fact it was indeed a nightmare when the programmers had to trace through an entangled copper wilderness to find out what caused the routine to crash. But as it turned out, the method worked. It should also be noteworthy to mention the tireless effort of these programmers who had to cope with unwielding deadlines and restrictions.

After a relentless struggle with deadlines, technological pitfalls and even personal conflicts the Apollo Guidance Computer was finally tested on Earth orbit on board Apollo 7 in February, 1968. The system worked flawlessly. In December, the same year it was flown on board Apollo 8 around Lunar orbit. During the flight, AGC worked perfectly providing accurate navigational calculations when the spaceship was on the dark side of the moon when ground based radio communication could not be reached. Both of this flights extensively tested all the routines of AGC except for one that was nevertheless crucial, the routine to land on the moon. It was in July, 1969 on the historic descent to the lunar surface of lunar module Eagle, on Apollo 11 that the robustness of AGC was finally ratified. While descending the computer unexpectedly started spewing of error messages with codes 1201 and 1202. Even for the controllers on ground staff the codes remained a mystery until one Jack Garman, who happened to have had a list of error messages handwritten with him, found it was a temporary overload issue of the computer with non fatal consequences to the mission. Owing to the ingenious design of prioritized time sharing the computer was simply dumping a data stream from a mistakenly activated sensor in favor of proceeding with the complications of landing. This minor fiasco acted as a catalyst on boosting the faith on AGC and for subsequent landings, the unit performed exceptionally well inspite of minor glitches.

For someone with technical interest AGC might seem like many things. One of the first digital computers, the first ever computer to be fully based on ICs, radical operating system based on prioritized time sharing and likewise. But for those people who dreamt of her, designed her and endured the pains in building her, she was far more than a machine. For them it was the product, the dream child of collective ingenuity, passion and imagination. An unfathomably deep feeling of accomplishment and pride, when looking at the moon nested in heavens, in feeling that some part of their life and soul still rests on that surface 384,000 kilometers away.