Previous Entry Share Next Entry
multics
jwg

The Project MAC 50th anniversary and Multics Reunion

(A long post with information about a very important period of my life and career)

On May 28 and 28th I attended a 2-day conference at MIT to celebrate the 50th anniversary of Project Mac - the second afternoon of which was a Multics Reunion.

Project MAC (Mathematics and Computation, later backronymed to Multiple Access Computer, Machine Aided Cognitions, or Man and Computer) was an interdepartmental project at MIT with lots of ARPA funding. It has transmogrified into MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) and sits in the Stata building - a Frank Gehry masterpiece.

One of the key projects in Project MAC was Multics (Multiplexed Computer and Information Service). It started as a joint project with MIT, Bell Labs, and General Electric and eventually became a Honeywell product. It was a many-user multiple access "time-sharing" system with an emphasis on security, 24-7 operation, ease of use, segmentation and virtual memory, and had a number of very new design ideas for the time. Many of the innovations have since appeared in other products as people who worked on Multics joined/started other companies and/or were users of it.

In 1967 I left Honeywell where I had spent the beginning years of my career as a Fortran Compiler writer and joined GE to work on this project. My last assignment in Honeywell had been in a Systems Planning organization looking at new Operating System possibilities and new ways to write software and was very frustrated at the difficulty getting them to think ahead. They sent me to an Industrial Liaison Symposium at MIT which was about Multics. Several months later I was working for GE on Multics .

GE was the hardware vendor who planned on making a commercial product out of this. In 1965 at the Fall Joint Computer Conference there was a set of papers presented describing this system. By 1966 the project had started design and development. Bell Labs dropped out in 1969 (the people working on Multics at Bell labs went on to invent Unix (castrated Multics - a joke name).

The GE people and the MIT people were working together, but not very well. My boss at GE, Charlie Clingen, got promoted to be the lab manager for our organization (Cambridge Information Systems Laboratory - CISL) and I got promoted to be manger of the development unit. We started working on getting people in the two organizations to work together. Several years later I became overall (lower level) project manager of the joint project and continued to work- on merging the people in the two organizations (on separate floors in the same building in Tech Square). Sometimes project members did not know which organization their peers worked for. In 1970 Honeywell bought the GE Computer business and committed to making Multics a product. We hired some of the MIT people and went on to make 12 major releases to a number of prestigious customers. It was a great model of a University Research/Industry joint project with effective technology transfer.

After the product was capped in 1985 and CISL closed in 1986 - wotj product maintenance and enhancement handled by our counterpart group in Phoenix and a University if Calgary spinoff company that had done some of the develop,emit in the later years. Some of my team moved to the local Honeywell office and worked with other people on a project, called Opus, to create a new OS for the existing mini-computer system. Opus was cancelled in 1988 - mostly because the growing popularity of PCs was killing the mini-computer business. Some of us also had an attempt for a spin-off to produce secure systems but that didn't pan out and besides Bull - the new owner of the Honeywell computer business wouldn't give up the technology rights - it took a long time for them to agree to release the source code.

The first day and a half of the conference had a number of presentations from Project Mac veterans, some current researchers, and university people from Cornell and Georgia Tech on new educational/research ideas. Some of the presentations described the origins of Project MAC, particularly those projects which are big deals today - either in industry or in the research arena. It was very interesting to attend and it was good to get back in touch with what is going in this industry and research lab that I worked in for 30+ years.

The Multics reunion was on the second afternoon and we had a presentation from the key MIT professor on this project and it's predecessor: Corby (Fernando Corbató) and several other presentations. It ended with a panel session that I chaired; the panelists were Corby, Jerry Saltzer - MIT professors with key roles in the project (and that is an understatement), Bob Freiburghouse - a compiler genius who was a colleague of mine at Multics and previously at Honeywell), and Peter Neumann who was the Bell Labs leader. The Panel session had short talks by each of us and then lots of Q&A which continued after the session was officially over.

There were lots of opportunities to reminisce with many people who I worked with in the past and even before that. I got to talk to Doug MacIlroy - one of the Bell Labs guys who produced a compiler for Multics when the contractor GE had chosen failed, and who was my M-11 (Calculus) instructor in my freshman year at MIT (1956)! I talked to Joel Moses, one of the MIT professors at Project MAC who had nothing to do with Multics but he and I were students at Columbia in 1961 when we both took an introductory computer programming course - which to me was the catalyst that launched my career.

It is interesting to note that this project was conceived in 1964, first ran as a service at MIT in 1969, became a product in 1974, ran in mission-critical applications at a number of big customers - with a total of about 70 systems (5-10 million dollars per system) and whose last system was shutdown in 2000. It was used as a Software Factory in Bull and Honeywell for other products. There is a huge amount of information about the system on the Multicians Web site including the one formal paper I wrote. The source code is on an MIT web site and there are several people working on emulators. And lots of the ideas were adapted by other companies for their products.

The people who worked on it were a great set of people and many of us remain in touch in various ways these days. One of those people is my husband.

  • 1
Nice write-up, JWG! :-) Spencer said he had a good time at the conference too.

Edited at 2014-06-05 09:42 pm (UTC)

And Honeywell, later H...Bull, is where I met you both. :-)

  • 1
?

Log in

No account? Create an account