Home Page
Description of JOSS

"1.  Implementing an Ideal
Shortly after the Rand Corporation moved to its present location in Santa Monica, California, in early 1953, the JOHNNIAC was completed and installed as Rand's first stored-program computer. Eight years later, this Princeton-type computing machine (named for mathematician John van Neumann) became the basis for JOSS, the JOHNNIAC Open-Shop System.

The first truly simple on-line system, JOSS represents a milestone in the history of conversational timesharing. In the years of its operation at Rand (1963 to the 1980s), JOSS has more than lived up to its billing as "The Helpful Assistant." But more significant than its long service to the Rand staff has been its influence on interactive system design.

JOSS was the result of an experimental project at Rand that was meant to demonstrate, on a small scale, the value of time-sharing and easy access to computing power for the non programmer. Although few of today's time-sharing users may know its name, many are benefiting from JOSS concepts of user-oriented language and terminal. These concepts had their origin in the 1950s, when the electronic computer was transformed from a one-of-a-kind laboratory curiosity into a marketable commodity.

As new applications were promoted, the language of computer instruction sets rose from the level of machine code to that of assemblers and compilers. At the same time, professional programmers were replacing those pioneering scientists and engineers who had learned to cope with the complexities of the machine in order to use this remarkable tool. Formerly a researcher seated at the computer's console had had its immediate and undivided attention. Now such dedicated personal usage was generally an unaffordable luxury.

. . .
. . .
. . .

At the Rand Corporation, Cliff Shaw envisioned the possibilities for a particular type of user not likely to be satisfied by the prospective commercial or governmental systems. In 1955 Cliff expressed his concern for the Rand staff's "open-shopper," the non programmer physicist, engineer, economist, or mathematician with a small numerical problem. Sometimes such a problem required only the features of the desk calculator of that day (which was far more elementary than today's hand-held calculator). At other times the sophisticated capabilities of an electronic computer were needed Cliff was also aware of the researcher's occasional desire to intervene in the calculation process.

A 1959 memo from Rand engineer Willis Ware suggested that future information processors would require "a multiplicity of personal input-output stations, so that many people can interact with the machine at the same time." In November 1960, meetings were held at Rand to decide the future of the JOHNNIAC.  Attending were Paul Armer, head of the Computer Sciences Department, Willis Ware, associate department head, engineers Keith Uncapher and Tom Ellis, and research programmer Clff Shaw.

Cliff recommended that the JOHNNIAC be used full time to service the open shop by means of hard-copy stations, and he looked forward to the challenge of resolving communication and monitoring difficulties.  The JOHNNIAC, in its youth, had served well as a production machine. When its storage and speed limitations eventually caused a shift to more modern equipment, this aging home-grown computer became attractive for the JOSS experiment.

There have been many creative contributors to JOSS's evolution from experimental to operational system (see the Appendix for the JOSS "Cast of Characters"). Among them, Cliff Shaw set out to establish the JOSS precept of "easy to learn and easy to use as a principle to govern the entire system design. As Cliff later recalled, his intent was not to make JOHNNIAC machine language available, but instead to provide a computational service through a new, machine-independent language. The system would be designed specifically to show the value of on-line access to a computer via a language tailored to a special class of user (the non programmer) and a special type of application (small and numerical).

It was in March 1961 that JOSS was formally proposed as "an exploration into continuous and intimate contact between a human user and a computer." So began the first phase of the U.S. Air Force sponsored Information Processor Project, whose goal was to improve communication between human and machine An unsung hero of the project, recalls Chuck Baker who directed the second phase, was John Williams! the head of Rand's Mathematics Department. John according to Chuck, "backed JOSS administratively in his own quiet but powerful way. His influence was vital when it came to getting the funds for the work His encouragement and enthusiasm helped us all over some very rough spots, too. And he had a lot of influence in what JOSS I finally looked like." John Williams, along with Norm Shapiro, Art Smith, Oliver Gross, Bill Sibley, and others, provided valuable feed. back to designers of JOSS I and II.

2. From JOSS I to JOSS In
Participants with Cliff Shaw in project discussions were Tom Ellis, Ike Nehama, Al Newell, and Keith Uncapher. Included in their plans were 10 typewriter terminals in fixed locations. Tom Ellis and Mal Davis directed the construction of the required multiple typewriter communication system during 1961 and 1962. Meanwhile Cliff was designing the interdependent details of language, terminal, and conversational environment.

JOSS was inaugurated in May 1963 with an initial five terminals and a minimal system. One terminal was installed at the JOHNNIAC, and four were located in the offices of Rand staff selected to evaluate JOSS. There were two principal elements of the terminal, from the user's point of view. One was an IBM model 868 typewriter. The other was a small box that indicated the status of the temlinal's communication electronics and controlled their functions.

The first schedule of operation, 9 to 12 each weekday morning, was announced on June 17. It was hoped that this limited edition of the system could provide a controlled feedback that would help shape the complete version. But the new users taught others so quickly that the experimenters had to resort to after-the-fact questionnaires.

January 1964 marks the official birth of the full JOSS implementation on the JOHNNIAC. By July of that year, JOSS service was extended into the evening hours to accommodate Rand staffers' growing addiction. The final version of JOSS I went into use in January 1965. Jean Sammet, in her history of programming languages (1969), notes: "The most amazing aspect of the whole JOSS activity is that it provided such a useful tool to so many people at the Rand Corporation." "And continued to provide," might be added as we consider the progression to the 1980s version."  [Annals of the History of Computing,  Vol. 4  #1  January 1982]