This paper outlines the major themes running though the interview between Professor Mahoney and A.G. Fraser, head of the Computer System Research department at Bell Labs. In the interview, Fraser modestly relates his role within computing science research at Bell Labs and presents some of the factors which influenced the development of UNIX. He begins by outlining his early work in England and relates the European view of computing in the 1960s. Fraser then goes on to explain the role of Multics and the 1969 Consent Decree in shaping the perception of computer science at Bell Labs. He relates this to both the mentality of the programmers there and to the creation of UNIX . Through various anecdotes the reader is treated to an inside view of many of the changes in computing over a 20 year span and is introduced to numerous important (although perhaps underrated) figures in computer science.
Perhaps it would be prudent to first provide some back-ground knowledge on Fraser in order to set the stage for his arrival at Bell Labs. Fraser studied aeronautical engineering at Bristol, England, and from there he traveled to Cambridge where he received a Ph.D. in computer science. It was at Cambridge that Frasers future in computer science was established. Here he attended summer courses in Numerical Mathematics, and worked for the Ferranti corporation. Ferranti funded the development of a time sharing system at Cambridge, known as the ATLAS 2, for which Fraser helped build the operating system. He notes that in the early 60s in Cambridge, there existed no computer science department and computing was treated as a mathematically-oriented side discipline. As well, most of the computer research at the time focused on increasing machine efficiency. However, ATLAS limited budget and need for integrated system components began to re-orient research towards programming. As Fraser relates, this was also the case at Ferranti, where the initial sales to commercial establishments had instilled the importance of data representation and combining file systems with programming languages.
When Fraser arrived at Bell Labs in 1969, he remarks on the despondent nature of the people there due to the failure of Multics. This is significant, for the consequence of this event had a great impact on the computing mentality at Bell and subsequently influenced the development of UNIX. Multics had been a new time sharing system, devised in conjunction with GE and MIT, to function as the next generation computer service. Despite the large investment in Multics, including the purchase of a new GE-645 to run it, the inability of Multics to meet the expectations of customers resulted in Bell Labs pulling out of the project. There was a great loss of momentum in the computing research area due to this. Computer science activities became disassociated from the rest of Bell Labs and impetus shifted to other areas. This was characterized by the loss of computer science personnel to other departments or research organizations, and the absence of both space and funding for computer science projects. For example, Ed David refused to provide Ken Thompson with a PDP-10 on which he could run UNIX. Instead, the initial coding was done on an old PDP-7, which forced the Bell Labs programmers to be resourceful and creative - attributes which were seen to become a part of the UNIX system.
Another of the other main factors influencing Bell Labs at this time was the Consent Decree of 69. As a result of an anti-trust case brought against Bell Labs, the Consent Decree effectively kept the corporation out of computing. As Fraser stresses, the legality of minicomputers with packet switching attributes was constantly called into question, as was the case with protocol translation. Subsequently, the fact that UNIX had both of these characteristics made it a legal hot potato. Hence, Bell Labs put UNIX on the back burner, creating the perception that UNIX was not a legitimate product of the corporation. For example, when 1127 (the department headed by Sam Morgan in charge of UNIX) finally received a computer for systems research, it was placed in the attic - far removed from the rest of the corporation. This created something of an underground mentality amongst UNIX programmers and freed them from the time restrictions and commercial constraints that became commonplace after the demise of the Consent Decree.
This mindset of exploration and innovation was the lodestone for Bell Labs. With the loss of so many of the research staff over Multics, Bell Labs sought to rejuvenate its think-tank with experienced and well educated people. As we well know, the focus of this group was to think both theoretically for the future and practically, for the immediate commercial interests of the corporation. This freedom permitted much experimentation that wasn't in the direct interests of the company. In terms of UNIX, this meant creating a system for the programmers themselves, where access to information was equally important to information security. Further, the entrepreneurial mentality and a lack of time restrictions permitted UNIX to be continually polished and restructured. Doug McIlroy had a large role in this, being considered by Fraser to be a catalyst for the whole program. The importance McIlroy placed on the creation of an accurate and comprehensive manual resulted in programs written well and with an intellectual honesty reflective of works of literature, not simply lines of code. Moreover, projects such as Frasers design aid system refocused computer research on the integration of system components, marking the shift towards computers as universal tools.
Although very modest concerning his contributions to UNIX and computing science in general, Sandy Frasers documentation of the Bell Labs environment during the creation of UNIX provides valuable insight not only into UNIX itself, but the creators of UNIX. More specifically, he presents how the Consent Decree of 69, along with the failure of Multics, fostered the creativity and free will that refined the UNIX system and moved computing science into the age of systems programming.