Project MAC and Multics

Welcome to the history of computing podcast. Today we’re going to cover a cold war-era project called Project MAC that bridged MIT with GE and Bell Labs. The Russians beat the US to space when they launched Sputnik in 1958. Many in the US felt the nation was falling behind and so later that year president Dwight D. Eisenhower appointed then president of MIT James Killian as the Presidential Assistant for Science and created ARPA. The office was lean and funded a few projects without much oversight. One was Project MAC at MIT, which helped cement the university as one of the top in the field of computing as it grew. Project MAC, short for Project on Mathematics and Computation, was a 1960s collaborative endeavor to develop a workable timesharing system. The concept of timesharing initially emerged during the late 1950s. Scientists and Researchers finally went beyond batch processing with Whirlwind and its spiritual predecessors, the TX-0 through TX-2 computers at MIT. We had computer memory now and so had interactive computing. That meant we could explore different ways to connect directly with the machine. In 1959, British mathematician Christopher Strachey presented the first public presentation on timesharing at a UNESCO meeting, and John McCarthy distributed an internal letter regarding timesharing at MIT. Timesharing was initially demonstrated at the MIT Computational Center in November 1961, under the supervision of Fernando Corbato, an MIT professor. J.C.R. Licklider at ARPA had been involved with MIT for most of his career in one way or another and helped provide vision and funding along with contacts and guidance, including getting the team to work with Bolt, Beranek & Newman (BBN). Yuri Alekseyevich Gagarin went to space in 1961. The Russians were still lapping us. Money. Governments spend money. Let’s do that. Licklider assisted in the development of Project MAC, machine-assisted cognition, led by Professor Robert M. Fano. He then funded the project with $3 million per year. That would become the most prominent initiative in timesharing. In 1967, the Information Processing Techniques Office invested more than $12 million in over a dozen timesharing programs at colleges and research institutions. Timesharing then enabled the development of new software and hardware separate from that used for batch processing. Thus, one of the most important innovations to come out of the project was an operating system capable of supporting multiple parallel users - all of whom could have complete control of the machine. The operating system they created would be known as Multics, short for Multiplexed Information and Computing Service. It was created for a GE 645 computer but modular in nature and could be ported to other computers. The project was a collaborative effort between MIT, GE, and Bell Labs. Multics was the first time we really split files away from objects read in memory and wrote them into memory for processing then back to disk. They developed the concepts of dynamic linking, daemons, procedural calls, hierarchical file systems, process stacks, a split between user land and the system, and much more. By the end of six months after Project MAC was created, 200 users in 10 different MIT departments had secured access to the system. The Project MAC laboratory was apart from its former Department of Electrical Engineering by 1967 and evolved into its interdepartmental laboratory. Multics progressed from computer timesharing to a networked computer system, integrating file sharing and administration capabilities and security mechanisms into its architecture. The sophisticated design, which could serve 300 daily active users on 1,000 MIT terminal computers within a couple more years, inspired engineers Ken Thompson and Dennis Ritchie to create their own at Bell Labs, which evolved into the C programming language and the Unix operating system. See, all the stakeholders with all the things they wanted in the operating system had built something slow and fragile. Solo developers don’t tend to build amazing systems, but neither do large intracompany bureaucracies. GE never did commercialize Multics because they ended their computer hardware business in 1970. Bell Labs dropped out of the project as well. So Honeywell acquired the General Electric computer division and so rights to the Multics project. In addition, Honeywell possessed several other operating systems, each supported by its internal organizations. In 1976, Project MAC was renamed the Laboratory for Computer Science (LCS) at MIT, broadening its scope. Michael L. Dertouzos, the lab's director, advocated developing intelligent computer programs. To increase computer use, the laboratory analyzed how to construct cost-effective, user-friendly systems and the theoretical underpinnings of computer science to recognize space and time constraints. Some of their project ran for decades afterwards. In 2000, several Multics sites were shut down. The concept of buying corporate “computer utilities” was a large area of research in the late 60s to 70s. Scientists bought time on computers that universities purchased. Companies did the same. The pace of research at both increased dramatically. Companies like Tymeshare and IBM made money selling time or processing credits, and then after an anti-trust case, IBM handed that business over to Control Data Corporation, who developed training centers to teach people how to lease time. These helped prepare a generation of programmers when the microcomputers came along, often taking people who had spent their whole careers on CDC Cybers or Burroughs mainframes by surprise. That seems to happen with the rapid changes in computing. But it was good to those who invested in the concept early. And the lessons learned about scalable architectures were skills that transitioned nicely into a microcomputer world. In fact, many environments still run on applications built in this era. The Laboratory for Computer Science (LCS) accomplished other ground-breaking work, including playing a critical role in advancing the Internet. It was often larger but less opulent than the AI lab at MIT. And their role in developing applications that would facilitate online processing and evaluation across various academic fields, such as engineering, medical, and library sciences led to advances in each. In 2004, LCS merged with MIT's AI laboratory to establish the Computer Science and Artificial Intelligence Laboratory (CSAIL), one of the flagship research labs at MIT. And in the meantime countless computer scientists who contributed at every level of the field flowed through MIT - some because of the name made in those early days. And the royalties from patents have certainly helped the universities endowment. The Cold War thawed. The US reduced ARPA spending after the Mansfield Amendment was passed in 1969. The MIT hackers flowed out to the world, changing not only how people thought of automating business processes, but how they thought of work and collaboration. And those hackers were happy to circumvent all the security precautions put on Multics, and so cultural movements evolved from there. And the legacy of Multics lived on in Unix, which evolved to influence Linux and is in some way now a part of iOS, Mac OS, Android, and Chrome OS.