Lynn Conway's Retrospective
Lynn Conway [Draft of 12-30-03.]
Copyright © 1999-2003, Lynn Conway.
All Rights Reserved.
 
 
 

 
PART IV: ON TO SUCCESS
 

The MIT '78 VLSI design course provided a means of testing the system design methods and the draft textbook - - it was quite an intense experience - - - students learned the basics of design in the first part of the course, and then did substantial chip design projects that were quickly implemented - - - it all went extremely well - - - validation! everything works! - - but then the pushback begins, from the establishment and from the other computer lab at PARC - - - we're accused of creating "unsound methods" - - So the big question now was: How do "unsound methods" become sound methods?
 
Then I got an idea: What if we conducted a dramatic, large-scale test of the methods in a bunch of universities by exploiting the Arpanet and an automated "VLSI implementation system" at PARC - - this happening became known as "MPC79", and was a great network adventure that further tested, validated and propagated the new Mead-Conway methods - - -
 
Suddenly, this was all a success! - - - with huge impact - - - followed by awards, start-ups, DARPA buying in - - - everything was different now - - - and the door opened up to many new opportunities - - - However, there are questions too - - - why have I been completely distracted by my work for so long - - - is this all there's going to be to my life now? - - - and what about the question of love?
 
 
It is easier to obtain forgiveness than to get permission -
- Admiral Grace Hopper
 
When the going gets weird, the weird turn pro -
- Hunter Thompson
 
 

 
 

 CONTENTS OF PART IV

 
7. M. I. T.
The '78 VLSI System Design Course
The Results
8. Back at PARC
Serious "pushback" from the establishment begins
The idea of MPC79
Running MPC79: The Network Adventure
Then suddenly, success!
The DARPA VLSI Program
The VLSI Startups
MPC Technology Transfer to Start the MOSIS Service
Mead and Conway: collaborators and antagonists
Memories of ACS-1 remain alive
The Question of Love
Reflections on PARC
Formation of the Knowledge Systems Area at PARC
Explorations in Knowledge Programming and Collaboration Technology
Time to Move On?
 
 

 PREFACE

  PART I

 PART II

 PART III

 PART IV

  PART V

 PART VI
 
 

 
7. M. I. T. [on leave from PARC, 8-78 to 1-79]
 
The '78 VLSI System Design Course
 
- - - the M.I.T. course was a proving ground for the new form of VLSI design course based on the M-C text - - a pivotal experience and result - - designing the course - - the course plan, involving quickly teaching the basics, and then having students do projects - - - working with Professor Jon Allen, who was the MIT faculty coordinator for my visit and for my new course - - and with Glen Miranker, an M.I.T. Ph. D. student who was the TA for the course - - to kick off "Course 6.978" - - -
 
Night Scene: The Dome at M.I.T.
photo by Liang-Wu Cai
 
 
- - - the housing arrangements for me at MIT were great - - - I moved into an apartment in the Eastgate building that was reserved for visiting faculty - - -Eastgate is a family housing building at 80 Wadsworth St. in Cambridge - - - my wonderful upper floor apartment on a southwest corner of the building had a spectacular view out over the Charles River, the Boston skyline and M.I.T. too - - - and was a great place for concentrating on all the hard work of that fall - - - and it was within easy walking distance of the main MIT buildings, so even in bad wintry weather I would have no trouble getting back and forth to classes and to the lab for the course - - -
 
- - - the MIT environment - - the talented students - - many attracted by the fantastic perceived opportunity to possible "design their own chip" (reminding me of how excited I was years before at being able to "create my own compiler" at Columbia) - - - challenges of teaching - - my shyness at presenting things in front of a group, and thus the need for over preparation - - lectures carefully hand-written almost down to the last work, and diagrams fully prepared in advance - - -
 

 Class Photo: Students and faculty observers, MIT'78 VLSI design course
   
 
- - - but all goes well - - my fears receded - - and I fell back into the intense, "wired", MIT style of mental life that I remembered all so well from the 50's - - - the old "killer instinct" for finishing things rose up in me and came to the fore - - - the excitement built as the course gets up a head of steam and the students begin their projects - - - some amazing projects begin to emerge - - - the M.I.T. '78 multi-project chip set - - using the Arpanet (the DoD/univesity precursor to the modern Internet) to coordinate the MPC process and communicate design files to/from Xerox PARC - - QTA implementation at Micro-Mask and HP - - access to HP fabrication facility coordinated through Merrill Brooksby - -
 
The Results
 
- - - the impressive project results- - packaged chips available about six weeks after the course was over - - - for example, Jim Cherry's transformational memory array - - and especially Guy Steele's LISP microprocessor - - - become competitive challenges to other architects and designers - - - if these are examples of what students might be able to do even with just one course under there belt - - - who knows what the ultimate possibilities might be - -
 

 Photo of the M.I.T. '78 Multi-project Chip Set
 
 
 
- - a side-effect of my "over preparation" for the course was the trail of extremely detailed lecture notes for the whole course (almost word for word) - - these would prove invaluable the following year as part of "The Instructors Guide" used to help propagate the course into many other universities - - -
 
- - - for lots more information about these experiences at MIT , see the "The MIT VLSI System Design Course", which provides a detailed outline of the course, and lots of info on the projects, mask and fab logistics, and results - - - and see also the RLE Circuits and Systems Group History site, which positions this course as pivotal in the history of the circuits and systems work in the Laboratory - - -
 
- - - during the latter part of the course, and while waiting for the fab to complete, I spent some time reacquainting myself with Cambridge and Boston - - it's such a wonderful area, so much to do there and so much history - - - I also retraced many of my steps from 18 to 23 years before - - bringing back many bittersweet memories - - - I couldn't help thinking how far I'd come - - back then I'd never dreamed that such a complete life would ever be possible for me - - -
 
- - - the drive back across the country, through the southwest, in February 1979 was really something - - - I took about a week; my first real break in years - - - I was on quite a high - - realizing what the success of this course might mean - - I also felt a new level of confidence and new capabilities for providing better team leadership back at PARC- -
 

 
8. BACK AT PARC [Feb '79]
 
- - - completing the Mead-Conway text - - drawing on all the details of the MIT experience to get rid of many remaining bugs in the details of the methods, the text, the prototyping methods, etc. - - - scaling up for a larger number of Universities to offer M-C VLSI design courses - - thinking about what would be needed out there - - preparation of Lynn's Instructor's Guide to VLSI Design - - Bob Hon's and Carlo Sequin's Guide to VLSI Implementation, which included critical cell library components designed by Dick Lyon, and the detailed CIF2.0 description by Lyon and Sproull - - -
 
Serious "pushback" from the establishment begins:
 
- - - growing "pushback" from the "electronics establishment" - - - - - between "non-optimal design rules" and "foundries", and claims that "you can teach VLSI design in half-a-semester", more and more establishment folks in the universities and industry who heard about our work got major allergic reactions to it - -
 
- - - also, by this time, Carver really enjoyed flaming about our "powerful, structured design methods" anytime he could get onto a podium (Carver thought that "structured" was a better term than "simplified", since it made them seem more complicated, and thus "less unsound") - - he also liked to remind people about "his theories of the true potential of MOS VLSI, of the ultimate limits and ultimate significance of the technology" - - - I think this started to irritate some traditionalists out there - - - heck, it even started to irritate me sometimes! - - -
 
- - and instead of publishing in regular journals, we were just self-publishing reams of stuff at PARC and just giving it away to anyone who seemed seriously interested in it - - to many folks, this was beyond the pale - - "no one is peer-reviewing this stuff" - - it may be totally bogus - - - the typical reaction from traditional quarters now began to be "Who are these crazy people anyways? Are they completely nuts?" - - especially this "Conway" person - - After all, what can she know? She doesn't even have a Ph.D! - -
 
- - - even at PARC, folks in the other computer lab (CSL) looked down their noses at our work, and felt that we were using "unsound methods" - - many folks seemed to react to us as if we were like Kurtz, in Apocalypse Now - - operating some kind of weird cult - -
 
- - - of course, by now I suppose there was some truth to that - - and it was partly a deliberate construct, as in had been in CSL itself - - the networked "VLSI community" had been patterned somewhat on CSL's dealer style of "open collaborative competition" - - and it had really taken off and become a sort of VLSI clan of its own out in the network - - you couldn't miss our clan artifacts at the various universities already involved - - our bible was "the book" - - there were giant chip layout plots everywhere - - we used the colors blue, red and green, which had special meanings - - the excitement of the rituals of design and QTA implementation - - then, there was all the special language that only we knew - - it was "us against the establishment" - -
 
- - then it got serious - - we began experiencing great pressure within PARC from leaders in CSL, especially Butler Lampson, who was constantly running down the VLSI work in efforts to hurt the project and shift funding from it to CSL - - Lampson was the dominant technical persona of CSL - - - Bob Taylor's "top gunfighter" - - - he loved to verbally confront and attack people - - a walking encyclopedia of all past theory - - he'd often shoot folks ideas down if they weren't totally perfect on the smallest theoretical details - -
 
- - - however, Butler often missed the big picture of what was being worked on - - a "detail guy", he didn't seem to think much about "context" - - - heck, he never even asked us what we thought we were doing - - without checking it out, how could he possibly have visualized the VLSI phenomenon that was starting to loom large outside of PARC - - all he saw was the tiny little piece of the clan within PARC - - and whatever that was, he didn't like it - - - so this became a very, very big problem for us within PARC - - -
 
- - - a factor was now at work that we only fully understood later - - PARC was now undergoing a rapid escalation of internal turmoil due to the failure of Xerox to capitalize on all the creative work done there on the new Alto, ethernet, laser printer, etc., computing environment - - and this led to all sorts of terrible political infighting among the PARC laboratories - - it was as if a sort of implosion was going on all around us, just as we found our backs against the wall too - - it was actually very frightening - -
 
- - - anyways, sometime during the spring of '79, I began to realize that gads, we've really got to prove that this stuff truly works before we're totally defamed out of existence - - - both inside and outside of PARC - - - and there wasn't much time - - the book would be coming out that next fall - - - it could fall flat on its face if there wasn't a critical mass of support - - we needed a better, larger-scale demonstration of the effectiveness of the methods and the instructional materials - -
 
- - then I got an idea - - - I tried to interest folks in the General Sciences Lab at PARC to think about using the new PARC semiconductor fab line to explore how to do fast-turnaround prototyping - - thinking that we might be able to support lots of Xerox designers with remote access to a network server at PARC, and then to QTA internal fab at PARC, for enabling QTA design and fab throughout Xerox - - thinking that this might be a "win-win" situation for GSL and the VLSI Systems Area, and that it might demonstrate useful accomplishments of our VLSI work from Xerox's viewpoint - -
 
- - - I made some sketches of the flow of infrastructural activity showing how to do this - - but got absolutely no interest in this at all from the Xerox GSL folks - - - the Lab manager was very polite all through the process, but slow-rolled any follow-up - - - GSL never offered any cooperation on our effort then or later - - nor did any other Xerox group or facility ever express any interest in that we were doing - - - what an amazing contrast to the eager collaboration we got going with HP - -
 
- - - it wasn't till much later that I really realized why Xerox had never shown any interest in applying our VLSI work, even within PARC - - - gradually it dawned on me that the CSL sniping had given our work a strange image with PARC's management (other than Bert Sutherland) - - it was probably considered to be rather "toy-like" and naive (until the smashing success of MPC79, and all the resulting attention in the outside world) - - I guess you can't be a prophet in your own land - - - anyways, Xerox missed out , and never got anything out of our VLSI work - - however, there was a wonderful silver lining to this, for my team: since our work wasn't taken very seriously, Xerox let us publish openly, collaborate openly without restriction with other companies like HP, and generally do whatever we wanted with our results - - pretty amazing, eh! - - -
 
The idea of MPC79:
 
- - - anyways, looking at the sketches and the proposal to GSL - - and thinking about how cool it would have been to ramp up a large activity within Xerox, I got a sudden brainstorm - - - why not organize a great network adventure, not within Xerox, but out in the Arpanet! - - and provide chip fab for a whole bunch of student projects and courses at many different universities in the fall of '79 - - - sort of like the MIT effort, but scaled way up - - - by building an Arpanet network server to handle the interactions with many users: e-mail announcements, file transfers, design file checking, chip area allocations, final design file merges, etc. - - an early form of "internet commerce" - - this might just work - - and we could bootleg it all the way - - - relying on our outside friends for maskmaking and silicon fabrication, and not needing much other support to do it - - -
 

 Map of the ARPANET, which linked many DoD and elite research university sites, at about the time of MPC79

 
 
- - I explored the possibility with Merrill Brooksby of HP for gaining HP support for doing the QTA fab again, as in the MIT '78 course, but on a larger scale - - Merrill quickly gave the green light - - (in return, we later collaborated with Merrill on generating an extensive VLSI video course, in HP facilities, that HP used internally in their design and CAD groups) - - - Merrill really saved us when we were on the edge - - - so we didn't need Xerox GSL after all - - - we could do it on "our own"! (with HP) - -
 
- - - that spring and summer I worked with Alan Bell and Martin Newell to plan, design and implement a new form of infrastructure and network server - - Alan came up with many novel concepts for automating various logistical interactions with users - - semi-automating the e-mail interactions, etc. - - it looks like it will be ready in time - - - we gather up our nerve to really announce it and commit to it - - - Bert gave us the final go ahead, plus he ok'ed funds for maskmaking, and he got ready to run interference for us - - - I realized that if we could pull this off successfully, just as the text was being introduced, there'd be no stopping "the movement" - -
 
- - - of course, I didn't want to think about the downside - - - I vividly recall one PARC researcher comment "God, you guys sure are high-rollers!" - - that was probably a good characterization of how we felt that summer, as we announced "MPC'79" - - we bet it all on one roll of the dice - - -
 
- - looking back on these years, I see now that I benefited tremendously from my earlier experiences in climbing, in my transition, and in generally doing many things that had to be done even though I was very frightened at the time - - those daring adventures in the past seemed to provide me with almost magical powers during the peak VLSI years - - I was able to just go do stuff that almost no one else would have had the nerve to do - -
 
- - - but something else had happened by then too - - it was now some years since my second pubertal passage, and I'd been in the new female social, emotional, physical and hormonal regime for quite a while now - - I began to sense that my mental powers had greatly expanded in important new dimensions - - it's hard to put a finger on it, but I was just ever so much more imaginative and creative than before transition - - especially noticing that I had vastly improved capabilities at visualizing and mentally simulating complex social interactions - -
 
- - in the process, as we moved from MIT78 into MPC79, I'd become the Erin Brokovitch of the VLSI movement - - I was kind of a "in-your-face" proletarian type of gal - - - the leader of the VLSI movement who didn't let anything stop her - - someone like that is always underestimated and made fun of by opponents - - but they can also be an inspiring leader to those who sense what's happening - - - I think my enthusiasm and doing it that way that helped cause the thing to snowball and succeed - -
 
- - - there was also an eerie aspect to now running much of the revolution through the Arpanet - - in that domain, you never met most of your collaborators, nor they you - - it's summed up well in a cool comment Bob Hon made one evening while working intensely at his Alto - - Bob was a CMU grad student and a summer intern at PARC, yet he was playing a critical role in MPC79 - - sitting there in his T-shirt on a warm summer evening, kind of talking to himself, he said: "if they only knew who was doing this " - - - I thought to myself "Wow Bob, you're right, and not just about you but about me too!", reflecting on my T-girl past - -
 
- - during this period Doug Fairbairn and Dick Lyon put together a short-but-intensive videotape course on VLSI design - - we used that course and personal coaching to help initiate a number of key folks - - in particular, a young Stanford professor by the name of Jim Clark took advantage of this - - we coached him on design as he worked with us at PARC that summer on a prototype design for a fast graphics processor - - he got it ready in time for MPC79 - -
 
- - - Carver also made an extensive effort that summer at instructing instructors of VLSI design, collaborating with Ted Kehl on a VLSI instructors workshop at University of Washington - - he was able to exploit Doug and Dick's videotape VLSI course, and all the instructors went away with a copy of my Guide for Instructors of VLSI Design, for their detailed course planning - - -
 
- - - but as Carver got wind of the MPC79 concept and our plans at PARC, he became very concerned - - - he didn't seem to sense the possibilities for huge scale-up and finally hitting "critical mass" - - but instead was concerned that the DARPA folks would "get control" of things and then they would "steal credit" for things (Carver was always concerned about gettting credit for things, and he was actually right about DARPA - they later did try to claim credit for the VLSI innovations and for innovating MOSIS too) - - - he thought that each university should make direct, local arrangements with its own "foundry" - - - this worked quite well in the case of Univ. of Washington that summer, given his intense personal involvement - - - but in my mind, it wouldn't scale up quickly enough - - -
 
- - - Carver and I had an intense disagreement over the local foundry vs MPC infrastructure issue - - - he was incredibly strongly against doing MPC79 - - - in the end, however, it's good that both methods were pushed independently - - - as both produced very useful results and experiences that could be built on by others - - - his methods were more easily adopted right away in the small valley firms, where there was local access to restructured "foundry"-type fab lines, but little experience yet with the world of servers and the internet - -
 
- - - however, my methods were more suited to the very large, but scattered university community, where local direct (tape/disc) access to foundries was seldom possible, but where there was access to the internet, and thus to the MPC service - - - (as we'll see, MPC79 had a powerful effect in propagating the Mead-Conway methods and courses in the universities, and in the end it led to a natonal infrastructure called the MOSIS service, which was patterned on my MPC79 innovations) - - -
 
- - -anyways, in parallel with all this MPC work, I was coordinating the completion of the Mead-Conway text - - copy-editing got underway during the spring of '79 - - with the goal of having the Addison-Wesley published text available for the VLSI courses in the fall of '79, concurrent with the operation of MPC79 - - as it turns out, A-W fell a bit short, and couldn't get the complete text done till mid/late fall in '79, so they issued a special 5-chapter softcover version, with a credit for the full text when available - - (those 5-chapter versions are "collectables" now) - - but finally, in the fall of '79, the new textbook was published - - See the Computer Museum's computer history entry for a short description of Introduction to VLSI Systems - -
 
 

 Introduction to VLSI Systems,
published in late fall, 1979
 
 
Within a few years, this seminal text was adopted for chip design courses at over 100 universities thoughout the world.
 
 
This text and the associated "MOSIS" prototyping infrastructure based on Lynn's MPC79 innovations (see below) were largely responsible for the huge wave of chip design innovations during the 80's and 90's, especially in Silicon Valley.
 
 
Running MPC79: The Network Adventure
 
- - then, as the fall semester went on, MPC'79 amazingly built up to a feverish level of activity - - - you could feel the involvement out there - - the emotion and the intensity out in the network community - - - to the students and faculty at the many universities, access to our site meant so very much - - even though from our viewpoint our site was just barely able to cope with the situation - - there were so many ways it could have come off the tracks - - - but it didn't - - the inventiveness and superhuman efforts put forth by Alan Bell, Martin Newell, Dick Lyon, Bob Hon and Barb Baird made it all work - -
 
 Flowchart of the overall MPC79 activity
 
 
- - - there was huge external participation - - ongoing coordination was conducted via carefully calculated e-mail logistical control messages - - based on the exact sequence of project events in the MIT'78 course - - on which the new courses were based - - the project area bidding and planning process worked perfectly, as did the final file submissions, project merges and maskmaking - - - with Alan Bell overseeing the complex network interactions - - - - the collaboration with Merrill Brooksby and also with Pat Castro at HP to interface the HP fabrication facility went smoothly - - and in the end, we got very fast-turnaround - - the wafers were ready in early January - - - we then quickly diced, packaged and wire-bonded the chips at PARC, and sent the packaged chips with wire-bonding maps and "implementation documentation" back to the designers - - - only one month after the "final merge" - - it had all worked, just as planned!
 
Alan Bell
using the MPC Implementation System
to merge the
MPC79 projects
 
December 1979
 
- - this new form of internet-based infrastructure and protocols for rapid prototyping of large numbers of VLSI chip designs was refined, and operated again in the spring of '80 as MPC580, with a new PARC team under Ted Strollo overseeing the process - - MPC'79 and MPC580 supported rapid prototyping for students taking "Mead-Conway" design courses at many, many universities - - and thus institutionalized a VLSI ritual of learning, thinking up a design concept, designing a prototype, and getting it implemented and tested - - all this in an open community that was sharing design ideas, design tools, etc. - -
 
 
 Photo of one of the many MPC79 multi-project chips. This one is from Stanford University and contains Jim Clark's first prototype for the Geometry Engine (lower-right corner) , the chip that later launched Silicon Graphics, Inc.
 
 
- - about this same time, many foreign universities also expressed interest in starting VLSI courses - - and I sent "care packages" out to them too - - courses began at many foreign universities, and this further enhanced the credibility and visibility of our work - - however, being outside the US, they didn't have access to the MPC service (or the later MOSIS service), so some countries began to set up their own similar infrastructure - - notably the CSIRO program in Australia established by Craig Mudge, a DEC computer architect and one of our early collaborators, who had returned to his homeland - - -
 
For more information about MPC'79 and MPC580, and how they validated the overall design methods, textbook, course design, instructor's guide, etc., see the paper entitled The MPC Adventures, a 1981 Xerox PARC Technical Report that reflects on these amazing times.
 
- - - seeing possibilities for wider, immediate impact in the electronics design community in Silicon Valley, Doug Fairbairn decided to start-up LAMBDA Magazine (later known as VLSI Design Magazine) - - - and this magazine played a critical role in the wider dissemination of the VLSI methodology - - - seeing that Xerox wasn't adopting or exploiting our VLSI work, Doug was also very disappointed that Xerox hadn't followed up the work he'd also done on the NoteTaker portable computer - - which was work he did in collaboration with Alan Kay on realizing the dream of "the Dynabook" - - - so it wasn't long before Doug jumped ship, and left Xerox to join the VLSI Technology startup - - Barb Baird got an excellent opportunity there and went along with Doug - - fortunately we were able to bring Terri Doughty (now Wanke) on board as our administrative assistant, and Terri provided outstanding support to the team over the next phase of our work - - -
 
Bert Sutherland
and Terri Doughty,
 
Xerox PARC
 
1980
 
Then suddenly, success!
 
- - - suddenly, with the success of MPC'79, the "pushbacks" stop - -
 
- - it was amazing - - during much of our VLSI work, Carver had only modest support from Intel and an ONR grant at Caltech, and later a bit more from the "Silicon Structures Project" funding at Caltech - - - while much of my work at PARC was work over and above the usual hours, i.e., most of the truly novel work was "bootlegged" on the "Alto design tools" story - - not intentionally covert, but just sort of "invisible" to most folks within PARC until it hit it big outside - -
 
- - but now, everyone wanted a piece of our action - - the past year, DARPA had been encouraging some preliminary probes of applications of our methods by researchers at a few universities - - but now, with the MPC79 success, Bob Kahn and Duane Adams at DARPA proposed that DARPA establish a major new VLSI Program to provide significant funding for VLSI systems research in the university community, largely to build upon the new Mead-Conway methodology - - -
 
All of a sudden, I began to feel more relaxed and rather joyful about this wonderful knowledge adventure we'd all been on - - it was going to "take hold" after all - -
 
 Lynn Conway
in 1981
 
 
Relaxed and happily enjoying the emerging success of the VLSI chip design revolution.
 
In the fall of 1981, Electronics magazine awarded Carver Mead and Lynn Conway their annual Award for Achievement. Electronics was very widely read and highly regarded in the industry at the time. The Electronics award article sketched the story of the Mead-Conway work, and described the major impact their work was having within just 2 years of emerging from Xerox PARC and Caltech. This article brought a lot of serious attention to our work, and led to an even more rapid spread of our results - - In addition, I was promoted by Xerox Corporation to Principle Scientist and Manager of the VLSI Systems Area and then to Research Fellow and Manager of the VLSI Systems Area. Research Fellow is a very high rank as a scientific researcher in major U.S. corporations, so this promotion was a great honor for me.
 
The DARPA VLSI Program
 
- - - Shortly after the success of MPC79, DARPA established it's VLSI research funding program, and Paul Losleben was assigned as DARPA VLSI program manager. Under his leadership this program took off spectacularly - - - many outstanding university folks joined in the effort - - the DARPA VLSI community took shape - - impact of the program meetings - - - the "VLSI clan artifacts and rituals" spread out into a hundred universities - - key computer architects became early adopters of our methods and the MPC prototyping service - - not only Jim Clark, but also Dave Patterson at Berkeley, and Forest Baskett, John Hennessy and others at Stanford - - - - and a whole range of new explorations took off in CAD, with early participation of John Ousterhout, Richard Newton, Dave Johannsen, and Randy Bryant - - -
 
 
Gaetano Borriello & Dick Lyon
 
Xerox PARC researchers in Lynn's
VLSI System Design Area
 
 
 
1981
 
- - - Carver's star now rose to great heights with select industry leaders, especially the leaders of Intel, with whom he'd long been well-connected - - our work undoubtedly had great impact there - - I greatly admired Noyce, Moore and Grove, and often wished that I'd been introduced to them, but somehow that never happened - - of course none of those folks had any idea about my computer architecture and design methods background at IBM, which were my carefully guarded secrets - - - they probably just thought I was Carver's "helper" or "assistant in writing the book" rather than the primary architect of the entire new design methodology - - - oh well - - -
 
- - - then too, there seemed to be an inherent difference in the easy acceptance of gals within the computing industry (IBM, Memorex, Xerox PARC) as opposed to their outsider status in the semiconductor industry (Intel, etc., etc.) at the time - - I never quite figured out why this should be, but it was just one of those facts of life - - could it be due to differences in the origins of computer work in mathematics and electronics design, as opposed to the lab environments and culture of device physics? - - who knows - - - but it's probably just as well that I didn't take that job at Fairchild! - - -
 
- - - reflecting on the rapid propagation into the larger university community - - we can now see how huge was the pent-up demand in that community for knowledge about chip design - - most universities just assumed that what we were teaching was "how it was done in industry" - - and they were astonished to find out how "easy it was" - - - in most cases not recognizing that we were propagating controversial, newly coalesced methods - -
 
- - so, the idea of "creating the text that looked like a text after 10 years of practice" had worked - - - the "unsound methods" became "sound methods" in a sudden paradigm shift - - - after all, what works, works! - - - and so, there would now be the needed thousands of VLSI system designers ready and available to explore this new territory of electronics - - -
 
The VLSI Startups
 
- - start-up companies form and take off - - Jim Clark's MPC79 project chip design prototype leads to his start-up of Silicon Graphics - - Patterson's RISC VLSI architecture work at Berkeley exploits the M-C methods - - as would John Hennessy's work that lead to MIPS, etc. - - - Mead and Conway received the Electronics magazine award, and that further helped to counter some remaining grumbling - - - and the text rapidly spread into use at over 120 universities - - in effect, a new common design culture and a new wave of EDA tools spread rapidly out into the industry - - thru new firms such as Mentor Graphics and Cadence - -
 
- - Following my promotions to Principle Scientist at PARC - - and then to Research Fellow in 1980, I was finally able to buy a home nearer work after all the years in California - - I'd climbed the real estate ladder from a condo in Los Gatos in '72, to a home in the hills of Belmont in '75 - - and now in 1981 I finally moved up to a beautiful English cottage style home at 2190 Webster Street in Palo Alto - - just a couple of miles from PARC! - - it seemed almost unbelievable - - - (I've often thought that housing needs and commuting difficulties in the valley often added that little extra bit of juice for many folks when they were down, and helped keep them plugging away hard at work) - -
 
MPC Technology Transfer to Start the MOSIS Service
 
- - after MPC79 and MPC580 we realized that with success in hand, and with Xerox not interested in exploiting the prototyping service - - it needed a home somewhere - - - Bert Sutherland explored this challenge with Bob Kahn at DARPA, and a series of exploratory meetings were held with the Information Sciences Institute (ISI) at USC, a DARPA software contractor - - - Bert and I interacted with Keith Uncapher, the Director of ISI, on the business and logistical aspects of the transfer, and Alan Bell, Ted Strollo, Martin Newell and I collaborated with Danny Cohen of ISI to conduct the detailed technology transfer of the system - - -
 
- - - the MPC software technology and interaction protocols were officially transferred from Xerox to USC-ISI in 1981, where it became the foundation for the "MOSIS" service. Danny Cohen extended Alan's remote user interface by development of a fully automated e-mail interaction interface for the system, perhaps the first of its kind, and he and Vance Tyree also developed interfaces to numerous foundry fabrication sources for the system. ISI has continuously operated and evolved MOSIS ever since, with support from DARPA and NSF, as a national infrastructure for rapid prototyping of VLSI chips by university and research organizations.
 
Note: The folks at DARPA in conjunction with the folks at ISI, in efforts to boost the Agency, later rewrote this history to claim that THEY had envisioned and innovated MOSIS from the outset. Of course that's in the same category of bogus claim as when DARPA took credit for the M16, which Stoner had pioneered, when all DARPA had done was conduct advanced field trials that proved the weapon's lethality.
 
In the end, by successfully simplifying and demystifying the previously extremely complex process of silicon chip design, the work of the Mead-Conway team was largely responsible for the rapid progress in VLSI chip design and design tools in the 80's.
 
There is also information on the Mead-Conway contributions to knowledge in the new book Funding a Revolution: Government Support for Computing Research, National Academy Press, 1999, including thoroughly-vetted information on Lynn Conway's innovation of the MOSIS service via the MPC prototype.
 
 Lynn Conway
in her office at Xerox PARC
 
photo by
Margaret Moulton
 
1983
 
 
 
Mead and Conway: collaborators and antagonists
 
- - Looking back on this period, it's clear that Carver and I had formed a very unique, incredibly stimulating and productive collaboration, right from the start - - - we each had just the needed complementary knowledge and experiences to become a truly great team, and make all this wild stuff happen - - - and somehow we goaded and stimulated each other to an ever higher pitch of activity as we went along - - and the excitement and intensity drew others into the team - - -
 
- - - Carver was an expert in device physics, LSI circuit design, and semiconductor processing methods - - he was widely known and respected out in many of the smaller, high-tech start-up firms - - and had many key connections in the companies we needed access to - - - he also had the vision and had made the key predictions regarding the huge density of circuitry that could eventually be fabricated in VLSI - - - and had a tremendous passion for seeing that scale-up become a reality and be productively exploited by designers - - as our work got up a head of steam, he was able to attract extremely bright Caltech students into the CAD work at Caltech/PARC and especially into the OM2 project, and made the OM2 effort a showcase of the emerging methods - - and he went on to convey the methods to many instructors via his "instructors' courses" - -
 
- - - I knew the world of digital system architecture and large-scale computer architecture - - - and had already had key experiences in the design of design methods (at ACS)- - - I also had an inherent understanding of how to "impedance match" the new knowledge to the working digital designer of the day - - - and I had developed a passion for somehow making it possible for other system designers, like me, to be able to quickly design digital systems and computers into silicon - - - I also had the good fortune to make key innovations, implement them and make them stick - - the scalable design rules and the concept of simplified design methods - - the idea of "the book" in terms of form, scope, and context - - - the idea of a standardized "project-oriented" one-semester VLSI course - - the idea of MPC79, leading to MOSIS - - -
 
- - - but these different, complementary backgrounds meant that we sometimes saw things rather differently - - and we worked in different, though related, spheres of collaborative activity - - - thus we were sometimes antagonists as much as collaborators - - wanting to take things in very different directions - - - all the mutual cross-checking and critiquing meant that our results were very well-vetted - - we needed each other to make it happen - - - but we also put each other under tons of pressure - - - fortunately, the "team" held together (barely) just long enough to make it all happen - - -
 
- - - interestingly, once success was in hand, we pretty quickly went our separate ways - - - Carver loved the limelight, gave many talks and interviews, got lots of awards and generally became a "pundit" - - - he especially became a favorite of George Gilder, the misogynist right-wing technology futurist and author who has ever since lionized Mead - - I met Gilder once, but he wasn't interested in really talking with a woman about her work - - - I doubt Carver informed him that I'd innovated most of the stuff - - (gads, imagine what Gilder will think when he learns my past) - - -
 
- - - I shied away from the limelight anyways, for various reasons including a still inherently shy personality concerned that too much public exposure might somehow accidentally reveal my past - - I simply went on to new research interests - - - Carver and I haven't seen each other but a few times in all the years since - - - although we are sort of on OK terms now, I guess - - -
 
- - - some people distant from our collaboration have likely wondered, and have probably gossiped a bit, about whether there was ever anything personal between Carver and me - - - but those who saw us working together at the time must certainly know better! - - - and anyways, I had always kept my personal life quite separate from my career, for a whole bunch of good reasons - - - (although, come to think of it, there were a couple of earlier exceptions along the line that were just too irresistible!) - - so, during my career, I tended to maintain a sort of assertive, in-your-face, tomboyish front at work, and a somewhat aloof personal stance, seldom revealing my earthy womanly feelings, saving those for a social life well away from work - -
 
- - - Note: Carver did not know about my transsexual past while we were collaborating on the VLSI work in the 1970's - - - however, he apparantly picked up on it during the late 80's sometime, either from another TS woman who thought he knew I was TS and who spilled the beans, or when my nomination for election to the NAE in '89 got me outed in the nominating committee - - - (anonymous letters came in to the NAE in '88 saying I was TS, probably in an effort to keep me out of the Academy)- - -
 
- - - reflect on key roles of the many people in the distributed "VLSI team" - - Bert Sutherland, Ivan Sutherland, Carver Mead, Doug Fairbairn, Jim Rowson, Dave Johannsen, Bob Sproull, Dick Lyon, Carlo Sequin, Bob Hon, Alan Bell, Martin Newell, Ted Strollo, Chuck Seitz, H. T. Kung, - - - Jon Allen, Glen Miranker, Ted Kehl, Rob Mathews, John Newkirk - - - Jeanie Treichel, Barbara Baird, Terri Doughty, Mary Hausladen - - - Martin Haeberli, Gaetano Borriello - - Bob Kahn, Paul Losleben, Duane Adams - - - Keith Uncapher, Danny Cohen - - Craig Mudge, Reiner Hartenstein - - Guy Steele, Jim Clark, Dave Patterson, John Hennessy, Forest Baskett - - John Ousterhout, Richard Newton, Clark Baker, Randy Bryant - - -
 
Memories of ACS-1 remain alive
 
- - all through the '77 to '81 period, I was itching to do something with the superscalar ideas and my dynamic instruction scheduling invention from the IBM-ACS days - - why hadn't IBM done anything with all that stuff? - - and no one else had done anything either - - but I obviously couldn't just go use those ideas myself in a major design, since that would be like stealing - - - and also I greatly feared accidentally surfacing various aspects of my personal life way back then - -
 
- - - but I so badly wanted those ideas to get out into real machines - - this was truly a great dilemma for me and it kept on buzzing in my head - - especially as it became clear that VLSI technology would eventually support ACS-scale machines on a single chip - - - so, what should I do, what could I do with the ACS knowledge? (remember: I'd saved all my original work from ACS) - - -
 
- - - the answer I came up with was simple: propagate the knowledge out into the emerging community of VLSI system architects - - - I began using the ACS-1 micro architecture as an "elegant supercomputer design example" that suggested the tremendously exciting things we might be able to do on VLSI chips within a few years - - I presented the ACS-1 micro architectural ideas right along with other examples, in many meetings with systems folks who visited PARC to learn about the VLSI work there - - but without attributing the ACS ideas other than to say they were from an "old IBM supercomputer project" - -
 
The Question of Love
 
It came up while I was attending a conference in Belgium, in '82 I think - - - I'd given a talk, and then gone out to dinner at a wonderful restaurant with my University hosts there - - - I was seated at the table next to a very interesting gentleman - - - we discussed my VLSI work at some length, including the sort of life that I led at PARC - - and then he asked the question that has stayed with me to this day - - he simply said, "but, what about love?" - - -
 
- - - there was sincere concern in his voice - - concern that all I was doing was working - - and I didn't have an answer for him - - only a silent nod, to show that I understood the question, and that it was a valid one - - he was very sweet - - of course he couldn't know that by then I'd kinda given up on finding love - - figuring that someone with my background couldn't realistically expect to find it - - - I hadn't dated any new men for several years now, relying on occasionally having intimacy and sexual encounters with several confirmed bachelors I'd gotten to know and could hang out with when I was lonely or frustrated - - but coming out of the blue like it did, the question really stuck with me - - it kept recurring in my mind from time to time - - - I realized that at least HE thought I should have found love - - - hmm, maybe I should go looking again! - - - I think this helped me eventually break free from my habitual preoccupation with work - - - and finally go out and take some chances on finding a mate again - - -
 
Reflections on PARC itself
 
- - some final reflections on the incredible research environment at Xerox PARC - - TBD - - - meantime see Michael Hilzik's 1999 book Dealer's of Lightning, the definitive history of the remarkable computer revolution spawned at PARC:
 
 
 
Formation of the Knowledge Systems Area at PARC:
 
- - in 1981 as VLSI success set in, I realized that this phase of my career was concluded - - it was time to move on - - - I opened myself up to new possibilities, new waves to catch - - I really looked around at what sort of exciting things were going on now - - I'd become increasingly curious re certain new movements in CS, especially artificial intelligence (AI) - - could there be possible applications to more supportive design tools, etc? - - and all the talk and work regarding knowledge representation languages intrigued me - - what was that all about? - - - exploited proximity to Stanford and early AI work at PARC to learn about this stuff - - then, as the VLSI work wound down - - and with Bert's encouragement and support, formed a new Research Area at PARC: The Knowledge Systems Area - - recruited Mark Stefik, one of Ed Feigenbaums's top Ph. D. students from Stanford, and also Danny Bobrow, from CSL, as that lab phased down its AI activities - - -
 
At the same time, my career at Xerox was going really well. I'd received promotions and salary increases, and had ridden the rising real estate market with my past homes, so that I could now finally afford a nice home in Palo Alto. Anyone who has lived on the Peninsula knows what the real estate ladder is like there, and how many people struggle in their careers just so that they eventually can live in the more expensive areas near the high-tech work locations. I bought a wonderful "english cottage" style home, with old brick facing, leaded-glass windows, and a cedar shake roof at 2190 Webster Street, in Palo Alto, and moved in with Sam and Punzel. This was a really wonderful home. It brought me a lot of quiet joy to be settled-in so nicely, even though I was alone with just my cats and didn't have anyone to share it with - - -
 
Explorations in Knowledge Programming and Collaboration Technology
 
- - - doing explorations in AI and early collaboration technology with Mark Stefik, Danny Bobrow, Sanjay Mittal - - we'd recruited Sanjay as a new Research Staff member, and Mary Hausladen as our new administrative assistant - - - - - AI impinged in interesting ways on my interests in cultural processes of knowledge evolution and diffusion, etc. - - it seemed strange: the AI folks doing knowledge engineering had reified knowledge as being extant, absolute, yet intangible - - rather than seeing it as tangible, artifact and behavior-based, carried by various clans, subject to diffusion and displacement processes - - I did some interesting work with Stefik to articulate and propagate this alternative viewpoint into the AI community - -
 
Sanjay Mittal, Mark Stefik and Danny Bobrow
 
 
Xerox PARC researchers in Lynn's
Knowledge Systems Area
 
 
1982
 
- - - the Knowledge Systems Area team prototyped improved tools for engineering knowledge - - the LOOPS knowledge programming system - - - and our team hosted some bizarre and party-like events known as "knowledge competitions" as part of conducting our research - -
 
- - - also, we began to question the "PARC solo-workstation paradigm" - - - the solo computer had its limitations - - - seeing a new space of possibilities and opportunities for "collaboration technology" we became one of the very first research groups to do original work in this new field - - - doing the early work that led to the later "Colab Project", and participating in interactions with the larger research community re knowledge engineering, expert systems, collaboration technology - - - the "Truckin" knowledge programming game - - - My rank as Research Fellow at PARC gave me considerable flexibility to help define research directions in this new area - - an area of research and technology that later on in the 1990's would begin to have considerable impact - - -
 
 
Time to Move On?
 
- - then, in late '82, Bob Cooper and Bob Kahn from DARPA visited PARC - - an opportunity was taking shape to participate in DARPA's new Strategic Computing Initiative - - - my research management background and contributions in AI, computer architecture and microelectronics design methods and infrastructure made me an ideal candidate for a key role in that program - - - and in the spring of '83 they asked me if I'd like to come to DARPA and lead the planning and organization of the program - - - now to called the "Strategic Computing" - - -
 
- - - I recalled the family stories about how my father had been one of the key leaders (Chief Engineer) of the U.S. synthetic rubber program in WW II - - - he had directed the critical "shootout" of pilot plants to select the group of processes for producing the various types of rubber during the early years of World War II - - - thus he had played a strategic role in critical defense technology back then - - - suddenly there was an opportunity for me to play a key role in a major Department of Defense initiative at this point in the cold war - - and this struck a chord in me - - - perhaps I was trying to prove something to a father, even though he was long dead - - perhaps trying to outdo a father who had so despised my female characteristics - - -
 
- - - something else was happening too - - - with success, I'd grown increasingly restless - - - maybe this was the time for me to start to break out of a lot of old habits, and move on in life - - - I began to wonder what I'd do longer term - - - Strategic Computing was an exciting new program at DARPA and I'd have a chance to learn a lot and to shape the new program in interesting ways - - - this also seemed like an opportunity to take a few years away from PARC and the Bay Area and explore new directions in my personal life too - - -
 
 

 
 

 PREFACE

  PART I

 PART II

 PART III

 PART IV

  PART V

 PART VI