Box 4.2

Project MAC and Computer Time-sharing

The development of computer time-sharing and the advent of minicomputers set the technological stage for the 1970s. Time-sharing systems divide computation power cyclically between many users over a network. Properly designed time-sharing computers can switch among processes quickly enough so that users do not recognize any delay, making it appear as though each user has the computer's full attention. Such systems took advantage of design and manufacturing peculiarities of mainframes that resulted in the power of a mainframe computer varying as the square of cost of the computer.¹ Therefore, building one computer for twice the cost of a smaller machine created four times the power. Time-sharing systems took advantage of this phenomena by allowing several users to share a single larger computer instead of several smaller machines. Development of such systems emerged from the complementary efforts of industry, universities, and government. Key to these efforts were Project MAC and its predecessors, funded by the Advanced Research Projects Agency and the National Science Foundation (NSF). While Project MAC was not responsible for the first time-sharing system, it played a significant role in the technology's development.

Project MAC was started by IPTO in 1963, with funding going to the Massachusetts Institute of Technology (MIT). MAC stood for Man and Computer, Machine-Aided Cognition, and Multi-Access Computer. J.C.R. Licklider chose MIT as the site for Project MAC because of the large variety of computer disciplines being studied at MIT. Project MAC brought together, for example, Marvin Minsky's artificial intelligence work, Douglas Ross's computer-aided design systems, Herbert Teager's studies in languages and devices, and Martin Greenberger's work with human-machine systems. While the program was justified to the military as a command-and-control program, Licklider's goal was much broader. He sought "the possibility of a profound advance, which will be almost literally an advance in the way of thinking about computing." In an interview with the Charles Babbage Institute, Licklider said, "I wanted interactive computing, I wanted time-sharing. I wanted themes like: computers are as much for communication as they are for calculation" (Norberg and O'Neil, 1996, pp. 97-98). Project MAC would eventually receive $25 million in total from 1963 to 1970 (Reed et al., 1990, Chapter 19, p. 14).

The core of Project MAC involved the design of a time-sharing computer system. Project MAC was not the first time-sharing initiative, but it significantly pushed the state of the art. Time-sharing systems had previously been developed in the MIT Computation Center, at System Development Corporation, and at Bolt, Beranek and Newman. At first, Project MAC used the MIT Computation Center's Compatible Time-Sharing System (CTSS), which had been designed under a grant from NSF. The system was built on an IBM 7090/94 and became operational in 1961. This was the first system enabling users to write their own programs online (Reed et al., 1990, pp. 19-2 to 19-3). In 1964, CTSS was connected to 24 terminals across the MIT campus. Eventually, 160 terminals were in place and 30 could be in use at one time. However, the CTSS still could not provide as much power as researchers desired, and it lacked necessary data access security.

Beginning in 1965, Project MAC began to create a second system with the help of General Electric and Bell Laboratories: MULTICS (Multiplexed Information and Computing Service), was completed in 1969 and would eventually support 1,000 terminals at MIT with 300 in use at any one time (Campbell-Kelly and Aspray, 1996, pp. 214-215). MULTICS also incorporated a multiuser file system and a complex virtual-memory system that allowed application programs to function as if available memory were much larger than the memory actually attached to the processor. It featured an automatically managed three-level memory system, controlled sharing and protection of data and programs accessed by multiple users, and the ability to reallocate its resources dyamically without interruption. MULTICS had a multiuser file system that allowed each user to work as if on an independent computer (Flamm, 1987, p. 58).

Project MAC led to many advances beyond time-sharing. MIT's Artificial Intelligence Laboratory received $1 million in funding through Project MAC for work to further the objectives of interactive computing (of which time-sharing was an integral part) and intelligent assistance (Norberg and O'Neill, 1996). Funds also went toward research in input/output devices. One of the earliest computer-aided design systems, KLUDGE, was developed through Project MAC. Project MAC's ability to compose and edit programs and documents online laid the groundwork for word processors and interactive programming. The idea for the spreadsheet, later popularized by Lotus 123 and subsequently Microsoft's Excel, also came from two students who worked on Project MAC. This idea spurred development of the first spreadsheet on the personal computer, VisiCalc, from Software Arts. The first real networking of the personal computer (the first version of Internet protocols for the PC) also came from MIT's Project MAC (renamed the Laboratory for Computer Science by then), which led to the company called FTP Software. FTP sold the first Internet protocol suite for DOS.

Another lasting spin-off from Project MAC was the popular operating system, Unix. The difficulty that Bell Laboratories had in developing the MULTICS operating system led to a new philosophy of software design stressing simplicity and elegance. In 1969, when Bell Laboratories realized that a commercial product was still many years away, it withdrew from Project MAC. Over the next 5 years, Bell researchers Kenneth Thompson and Dennis Ritchie, along with others who had been working with MAC and had become frustrated with MULTICS's complexity, developed Unix, which was based on MULTICS but was much simpler. It offered quick responses, had minimal system overhead, and ran on minicomputers instead of more expensive mainframes with special memory management systems.

Beyond the technical advances in time-sharing, Project MAC influenced an industrywide movement toward developing time-sharing computers. When searching for a contractor to supply the hardware for MULTICS, MIT turned down its traditional supplier, IBM, and hired General Electric (GE) because of IBM's unwillingness to modify their machines for the project. The early results of Project MAC, though, convinced IBM and other manufacturers that they would have to pursue time-sharing (Campbell-Kelly and Aspray, 1996, p. 215). By 1967, 20 firms were competing for a $20 million industry providing time-shared computer services to businesses across the nation including GE, Telcomp, Tymshare, Keydata, and University Computing Company. By the mid-1970s, almost every mainframe computer sold incorporated time-sharing technology (Reed et al., 1990, pp. 9-14).

Project MAC was largely responsible for bringing the computer out of the laboratory and business and leading it to the home. Licklider's desire to create a "new way of thinking" about computers succeeded. Project MAC developed technology and ideas that allowed interactive computing to become a reality. . . ." As a result of Project MAC and other computer time-sharing research programs in the late 1960s, the concept of computer utilities became widely accepted in the computer and business world. In 1964, only one year after Project MAC began, Martin Greenberger wrote, "Barring unforeseen obstacles, an on-line interactive computer service, provided commercially by an information utility, may be as commonplace by 2000 A.D. as the telephone service is today" (Campbell-Kelly and Aspray, 1996, p. 217). The image Greenberger described is remarkably similar to the Internet. Before time-sharing became a reality, computing remained available only to large businesses, academic institutions, and the government. However, as more users could simultaneously use a single machine, the cost of computing dramatically decreased, and usage increased accordingly. Project MAC played a large role in the public's change of philosophy regarding the use of computers.