[THVV] For many Multicians, the six FJCC Multics papers were the foundation of our knowledge about Multics.
The papers were first presented on Nevember 30, 1965 in a special session at the 1965 Fall Joint Computer Conference, held in Las Vegas, NV, at the Stardust Hotel. ("Joint" meant that the conference was sponsored by the American Federation of Information Processing Societies (AFIPS) which had members from ACM, IEEE and IRE.)
General Electric obtained permission to reprint the papers, and produced a soft-cover booklet titled A New Remote-Accessed Man-Machine System, which bore the legend "distributed as a courtesy by General Electric." GE salesmen gave these away to prospective customers. Here are links to the papers in HTML format.
- "Introduction and Overview of the Multics System," F. J. Corbató and V. A. Vyssotsky
- "System Design of a Computer for Time-Sharing Applications," E. L. Glaser, J. F. Couleur, and G. A. Oliver
- "Structure of the Multics Supervisor," V. A. Vyssotsky, F. J. Corbató, and R. M. Graham
- "A General Purpose File System for Secondary Storage," R. C. Daley and P. G. Neumann
- "Communications and Input/Output Switching in a Multiplex Computing System." J. F. Ossanna, L. Mikus, and S. D. Dunten
- "Some Thoughts About the Social Implications of Accessible Computing," E. E. David, Jr. and R. M. Fano
These papers have been widely referenced in other operating systems papers.
I sent mail to the authors I have addresses for, and asked them
- when was it decided that conference papers were the right way to present the system?
- who decided on how many papers to write, and who would write them?
- were there other papers considered, and why weren't they done?
- what was the writing process like? was there a lot of deadline pressure?
- was there a review process, and were there big changes made to any paper as a result?
- how was the FJCC chosen, and how was the special session arranged?
- how many people from the Multics team went to Las Vegas? Who presented each paper?
- what was the reaction of people at the conference to the papers?
- did the papers help enlist support or recruit staff?
Here are the replies.
Peter Neumann (06 may 2013) The key to the entire process was the Memorial Day 1965 week-long meeting at the AT&T training center in Hopewell NJ, which laid out the partitioning of the Multics development and the writing tasks. The FJCC was in our sights from then on. The group settled on the papers as they emerged, even though Fano and Ed David were not present in Hopewell. Their paper was simply postulated as highly desirable, and indeed it was written. The deadlines were imposed by the FJCC, although I believe there may have been some general agreement from the FJCC program chair that we could have a session of our own for those papers -- from the outset. Corby might remember the details on that, although it might have been Fano who negotiated the deal. As I recall, Vyssotsky, Ossanna, Ed David and I were all at the Stardust in Las Vegas. I think we had some questions about our potential lunacy for doing something that today would be thought of as as "clean-slate architecture". It certainly helped in getting new people on board, because they could read the papers and get a very good idea of what we were up to.
My favorite feedback along the lines of potential lunacy came from Fred Brooks at the Spring Joint 1967 in Atlantic City. He said to me, "Peter, You are an old-time information theoretician. On information theoretic grounds along, you should realize that Multics is IMPOSSIBLE."
He had heard that we could interrupt on any character, and interpreted it as we would interrupt on every typed character...
F. J. Corbató (10 may 2013) Peter's very good memory of the events certainly rings true to me, although I think I remember Fano insisting that we had to publish our intentions being a reason that we felt we had to create the session of papers. A key figure in the creation of Multics, of course, was the late Ted Glaser who worked intimately with John Couleur to modify the GE 635 hardware so that the software vision was possible. In hindsight, without the papers as a published goal, Multics would certainly have failed, since when we began to miss our self-declared target dates, there was serious pressure to abandon the effort. In hindsight, the papers served as inspiration for those actively involved, and certainly were key to recruiting newcomers. Your website, multicians.org, makes clear the tremendous scale of adoption by customers of GE and Honeywell during the system's peak.