User:Russell Noftsker

Most printed references to me contain errors. For anyone who cares, the following biographical sketch will provide a first hand source. Numerous references to the beginnings of the MIT AI Lab, the development of the Lisp Machine (one of the first workstations), and Symbolics, Inc. have been the subject of discussion in case studies[], thesis’s and books (|Hackers by Steven Levy& The Brain Makers by Harvey Newquist). Hard as those authors may have worked to get everything right, details were inevitably scrambled if only in minor ways. Hopefully this article offers answers to those who wondered. Some of the Lisp Machine development details were told to me and may require editing by those with first hand knowledge of that project.

Russell Noftsker

April 2007 ___________________________________________________________

Russell Noftsker
I was born February 1942 and attended public school in Carlsbad, New Mexico, entered New Mexico State University (Las Cruces) after my Junior year of high school and graduated with a BSEE (Scientific Option) in June 1964. During my undergraduate years I participated as a co-op student employee of the on-campus Physical Sciences Lab in the Office of Naval Research funded, Transit satellite tracking research project. This project proved the science and initiated development of the technology which grew into today’s GPS position location and navigation system.

In the spring of 1965 while employed in Lexington, MA, I was introduced to Prof. Marvin Minsky at the MIT Project MAC, AI Group in Project MAC's offices by one of Marvin's undergraduate students, Robert L. South, a childhood friend of mine. My undergraduate thesis on AI had referenced work reported in a book edited by Professor Minsky.

Marvin was on sabbatical at that time as a guest of Stanford Prof. John McCarthy's Stanford Artificial Intelligence Laboratory but happened to be back at MIT the weekend we met. Marvin asked me to tackle an engineering project for him. I did so in my spare time. After a few weeks he asked me (by phone) to join his AI Group in order to help organize and expand it. I accepted. Dick Mills, manager of Project MAC under Director Robert Fano, put me on the Project MAC payroll as a research staff member.

When I joined the AI Group, Dan Edwards previous manager of the group, Daniel Bobrow and Warren Teitleman had or were just graduating and leaving so by the end of the summer of 1965, Marvin, his secretary Faith Dilworth and I were the only staff or faculty members of his AI Group although a number of undergrads came back that fall.

I helped Marvin prepare and submit his first research funding proposal to ARPA's Information Processing Technology Office (IPTO) managed by Dr. Ivan Sutherland. Late in 1966, we received a funding commitment of approximately $700,000. This enabled us to build up a staff, expand inside Project MAC space, and take on a significant number of undergraduate and graduate students to conduct research on computer vision and manipulation. I mainly performed management and engineering tasks.

After a few years there was a general ARPA cutback imposed on Project MAC but not on the AI Group. Prof. J. C. R. Licklider, Project MAC’s Director at the time, felt it would only be fair for the AI Group to share in this cutback thereby reducing the pain on Project MAC. Marvin and I with the support of IPTO felt that this was unwarranted. When Lick appeared too hesitant about leaving our budget alone, Marvin and I pushed for and received authorization from the MIT administration to split off from Project MAC and form the AI Lab. This is how I became a co-founder with Prof. Marvin Minsky of the MIT Artificial Intelligence Laboratory.

A few years later in 1973 Prof. Marvin Minsky, Director, and Prof. Seymour Papert, by then Co-Director of the AI Lab, decided to pass the baton to one of Marvin’s PhD graduates, Patrick Winston. Since I had managed the AI Lab under Marvin’s direction, I offered the new Director his choice of my staying on as his manager, going back to engineering or leaving. He asked me to turn management over to his designated replacement and to stay on as a research staff engineer until I found something else interesting to do inside or outside the Lab.

I did as he requested but because demands on my time were no longer so pressing, after a short hand-over period, I took my accumulated 72 days of vacation time. During that time I went to Southern California with MIT Prof. Edward Fredkin as he was leaving on a sabbatical to Cal Tech for the academic year. During the course of various engineering consulting jobs in Southern California I started a company, Pertron Controls, with Kirk Mathews and Tom McMahon to design and manufacture the first production computer control for resistance welding machines.

Pertron was such a success that I decided not to return to MIT. Also, raising funding was faster and less painful in business than in academia.

After my departure from the AI Lab, Richard Greenblatt and Jack Holloway, two of the undergraduate students who had become senior research staff members while I was there, were inspired by a lecture delivered by Peter Deutsch at Berkeley advocating the building of large address space computer architectures optimized for the execution of specific higher level languages. Greenblatt and Tom Knight decided to explore this concept targeting the AI research programming language Lisp which had been authored by John McCarthy when he was at MIT. Greenblatt managed the project and obtained a small grant for support from IBM.

Tom Knight designed and built the microcoded Lisp execution engine which they called the CONS machine. Holloway connected it to the AI Lab’s main computer, a time shared DEC PDP-10, using an early base-band collision detection network, the Chaos Net, which he had developed and with which he and others later wired the MIT campus.

Richard Greenblatt wrote the CONS microcode to enable Lisp execution as a back-end processor attached to the DEC PDP-10. It provided significant improvements in execution speed over Lisp programs run on the ITS time sharing system which had initially been written by Greenblatt for the PDP-6 then later moved to the PDP-10. Greenblatt’s Lisp machine project attracted the efforts of a number of talented computer science and other interested students.

Among those contributors was Dave Moon who designed and built a disk controller and wrote a virtual memory manager for the processor which as it evolved was re-named the CADR. Tom Knight and/or Jack Holloway and Dave Moon designed, built and wrote the software to control a bit mapped display attached directly to the CADR. Another student, Howard Cannon, inspired by a demonstration of Smalltalk at Xerox PARC, wrote his “Flavors” multiple inheritance object oriented extension to the CADR Lisp. With Flavors (after it was turned into a production tool by Dave Moon) he wrote the world's first object oriented windows system. Around this time the CADR became more widely know as the MIT Lisp machine.

In his efforts to boost the number of the now stand-alone Lisp Machines for AI researchers, Greenblatt and his associates had contacted various potential manufacturers and had started in-lab production of Lisp Machines. As requests from other AI research facilities came in, the number they were to produce in-house rose to more than two dozen. After evidently having been turned down by IBM, DEC and Xerox, and contemplating how they could otherwise get Lisp machines produced, I was asked about the feasibility of setting up a manufacturing company.

In February 1979, all the key parties involved in the MIT Lisp Machine development met with me in the Project MAC conference room. Knowing how capital intensive a computer system manufacturing operation could be, I offered to try to set up and raise financing for an organization that, starting with the MIT Lisp Machine, would sell, design and build what at the time would have been the first commercial software development workstation. These computers would initially exploit tagged architectures designed to efficiently execute object oriented software and the Lisp language, and would incorporate hardware assisted garbage collection.

Unfortunately, at that February 1979 meeting, Greenblatt insisted on having control of and the biggest stake in any spin-off venture to produce Lisp Machines. He intended for the founders to construct the Lisp Machines, defer their compensation until full collection for deliveries, and to purchase material out of 50% deposits required with each order. I proposed a more conventional venture capital funded organization employing standard pay-as-you-go employment and billing-on-delivery practices.

None of the other interested parties were willing to accept Greenblatt’s demands and he wasn't willing to compromise so we agreed to give him a year at the end of which time if he hadn't made progress in launching a spin-off, we would do it our way. By the summer of 1979 I had become acquainted with Bob Adams, one of the entrepreneur founders of the successful venture capital funded minicomputer company Scientific Data Systems. Bob was confident that his contacts would enable him to raise VC funding.

In December 1979, because Greenblatt did not appear to be making headway, Jack Holloway and I met in LA to sketch out a business plan. Interestingly, Holloway said we needed a stand-alone full-custom single-chip 10 MIPS workstation to sell for $5,000 by the summer of 1984 and that we had to strike a deal with one of the commodity microcomputer chip manufacturers that would persuade them to integrate our tagged architecture requirements into commodity chips before the end of the 1980 decade. This was our only chance to get onto the Moore’s Law rocket. Unfortunately three of our VC’s didn't buy into that vision.

On April 9, 1980, just after the lapse of our stand-in-place agreement with Greenblatt, Bob Adams as President and I as Secretary incorporated Symbolics, Inc. in Delaware with the help of Andy Egendorf (see below) our legal counsel, then immediately started raising money. .

By early June 1981 we had raised and borrowed just under $500K from founders, friends, interested parties and a bank, had licensed the Lisp Machine technology from MIT, re-engineered the MIT Lisp Machine and begun its production in Woodland Hills, CA, had opened an R&D Lab in Cambridge, MA, had worked out employment arrangements with all 19 active founders and had introduced our company to numerous venture capital investors.

In an agreement with two of the VC’s who declined to invest with Bob Adams in charge, I took over the presidency of Symbolics in the spring of 1981. Symbolics shipped its first four LM-2 workstations to Fairchild Semiconductor AI Lab in early June 1981. A few days later on June 24th three Venture Capital firms invested $1.5 million in Symbolics. The three VC’s were Memorial Drive Trust represented by Dan Alexander and managed by Jean de Valpine (see below), the dean of practicing VC’s at the time; American Research and Development represented by Gene Pettinelli, the US’s first VC firm started by Harvard Prof. General Doriot; and Patricof Associates represented by John Baker, brother of our co-founder Henry.

Symbolics grew to annual sales of over $100 million by 1986, raised $90 million in equity financing and reached close to 1,000 employees before Moore’s Law passed us by. Commodity chips passed our products’ computing power before we could make the switch to commodity architectures. I retired early in 1988.

In April 1992, with help and investments from Jean E. de Valpine and Landon Clay, I helped Dr. Richard Petti save the symbolic math solver Macsyma software product from the collapse of Symbolics. Macsyma had started as the PhD thesis work of Joel Moses and Bill Martin in the MIT AI Lab. It was written in Lisp and may still be the most extensive application program written in that language. Moses used DEC PDP-10 computers under ARPA and NSF support. One of the graduate students who contributed to that project, Richard Fateman of Harvard, wanted a copy of Macsyma to take with him when he graduated and joined the Berkeley Math faculty.

Fateman persuaded Bell Labs to license Unix to him along with a re-distribution license in exchange for porting Unix from the DEC 16 bit machines which Bell Labs was using at the time to the DEC VAX, one of the first powerful 32 bit computers. Fateman and his Berkeley group succeeded in porting Unix to the VAX, writing a version of Lisp for this VAX Unix, and porting Macsyma to his VAX Lisp. With this done, Fateman began distributing Macsyma in the famous Berkeley tape distributions which included VAX Unix. It was through this path that the academic world got its hands on a shared re-distributable Unix and ultimately ported it to numerous other platforms

Andrew Egendorf
Andrew Egendorf (born 1945 in Philadelphia, Pennsylvania) was instrumental in establishing "Nader's Raiders" (1968), in creating Harvard University's JD/MBA (law/business) Joint Degree Program (1968), and in the founding of the first dot-com company, Symbolics, Inc. (1980).

Andy Egendorf was born in 1945 near Rittenhouse Square in Philadelphia, Pennsylvania, in the same house as actor Kevin Bacon, thereby conceivably garnering him a Bacon number of 1.

Egendorf showed an early aptitude for invention, science and math, and in Junior High School, as a science project, designed and launched a weather balloon – purchased from Edmund Scientific - which brought him to the attention of the Science Faculty at Cheltenham High School and Norm Edmund of Edmund Scientific Co. Egendorf subsequently was asked by Edmund to make crystal models for the introduction by Edmund in its 1960 catalog of the scientific modeling tool D-Stix.

As a High School freshman, Egendorf presented a paper at the Franklin Institute in Philadelphia on the results of a magnetism experiment he performed with equipment he designed and built, following which he was invited to attend a National Science Foundation sponsored Mathematics Institute at Temple University and as a Junior in High School was offered the opportunity by the Penn Mutual Life Insurance company of attending an actuarial training course.[ ]

Penn Mutual previously had given Egendorf the opportunity to program its Remington Rand 409-2, the world's first mass-produced computer,the first one of which was shipped to a customer in November 1951. As one of a handful of surviving people who actually programmed the machine, Egendorf helped fact-check information on the 409 line of computers for the recent book on the history of computing[ ]. Egendorf now is actively involved with the Rowayton Historical Society, Rowayton Connecticut, in its efforts to accurately present the history of the earliest commercially-available computers, particularly since most of the history has been lost, and these machines have been eclipsed in the popular press and the popular mind by the Remington Rand UNIVAC.

After graduation from Cheltenham High School in 1963, Egendorf continued his education at the Massachusetts Institute of Technology (MIT) on a Merit Scholarship, where he earned a Bachelor's Degree in Earth & Planetary Sciences in 1967. He then entered Harvard Business School.

In January, 1968, Egendorf and a friend wrote to famed consumer advocate Ralph Nader offering to work with him that summer if he would consider working with students. Egendorf convinced Nader that there was a huge reserve of available "student power", and as a result Nader established the student task forces which William Greider later dubbed "Naider's Raiders". . Egendorf's group studied the Federal Trade Commission and in March 1969 testified before Congress. Nader suggested to Egendorf that he obtain a law degree as a better means of effecting social change than obtaining a business degree.

Not wishing to abandon his MBA degree, Egendorf fought for the establishment of a dual-degree program at Harvard, and in 1968 Egendorf was selected as the first entrant for Harvard's newly-created joint Law-Business (JD-MBA) program, from which he graduated in 1971 with honors. As a programming project for Professor Ralph Zani while at Harvard Business School, in March 1970 Egendorf wrote the first electronic spreadsheet program for the use of his classmates on the school's teletype-input-output, time-shared IBM computer.

In 1967, while at MIT, Egendorf joined AI Lab Prof. Seymour Papert’s Summer (computer) Vision Project where he met Russell Noftsker, the co-founder and General Manager of the MIT Artificial Intelligence Laboratory, and in 1980 with Noftsker and a group of 20 other people, mostly from the AI Lab, founded Symbolics, Inc., the company which manufactured and sold one of the earliest workstations and the first computer built specifically to run AI software.

Egendorf, Noftsker and Robert Adams incorporated Symbolics, Inc. in Delaware on April 9, 1980. AI enables computers to emulate many aspects of human cognitive processing including reasoning and decision making, and has become an integral part of computer technology today. On March 15, 1985, Symbolics was assigned "Symbolics.com" as the first dotcom and first registered domain on the Internet.

In 1977, Egendorf joined the law firm of Widett, Slater & Goldman, a major general-practice firm in downtown Boston. In 1983 he left his position as a senior partner to become General Counsel of Symbolics. Symbolics became a public company in 1984 and traded on the NASDAQ under the symbol SMBX.

In the early 90s, Egendorf became interested in patent law as applied to the emerging field of electronic commerce, and subsequently registered with the Patent Office in 2001 and was granted numerous patents worldwide dealing with e-commerce. He currently is President and CEO of Tradecraft Corporation, an intellectual property holding company based in the Boston area.

Egendorf's wife is Linda Egendorf, an internationally-recognized sculptor. Art Calendar magazine selected "Aftermath", a work by Linda Egendorf, for the cover of its October, 2001 issue, which was "Dedicated to those who lost their lives on 9/11", and in 2005 she was the only woman sculptor accepted from the United States into the International Sculpture Biennale in Toyamura, Japan.

Jean de Valpine
Jean England de Valpine (born 1921 in St. Louis, Missouri) became the chief executive officer of Memorial Drive Trust in 1960 and immediately began providing new venture financing. In 1973 he helped organize the National Venture Capital Association which he served for many years as a founding director. de Valpine played a key role in changing Venture Capital into an industry. His own venture capital activities helped launched the cable TV and Artificial Intelligence industries.

Jean E. de Valpine was a 1943 field artillery ROTC and engineering science graduate of Harvard College, and of the Tank Destroyer School at Camp Hood (later Fort Hood), Texas. He served as forward observer and battery commander in the 25th Infantry Division in the Philippines and Japan, where, - as commanding officer of Battery A, 89th FA - he was assigned occupational military police responsibility for approximately one third of the city of Nagoya. He graduated from Harvard Law School in 1949.

He practiced law in Boston for ten years during which he formed the antitrust committee of the Boston Bar Association. In 1960, at the request of Royal Little, Chairman of Textron, lnc., and a trustee of Memorial Drive Trust (MDT), de Valpine undertook the management of MDT, the profit sharing trust and sole stockholder of Arthur D. Little, Inc. de Valpine immediately began investing part of the Trust assets in real estate development and early stage ventures, an investment category then called "risk capital," later to become known as "venture capital".

Starting with early cable television operators, this asset portfolio grew over the years to include an array of armaments, electronics, cryogenic technology, computer, communications and software companies. Jean de Valpine served as a director of many of these companies as they evolved from small startups to mature public enterprises. Inter alia, he was chairman of Realty Income Trust, the first publicly traded REIT, and in 1964-66 served as Assistant Attorney General of Massachusetts.

From 1969 until 1977 he was chairman of Space Research Corporation, an armaments firm engaged under the direction of its president and chief technical officer, Dr. Gerald Bull in the design of artillery systems and extended range ammunitions, based on gun technology research and development conducted mainly at the company's aeroballistic laboratory and 10,000 acre range straddling the Quebec-Vermont border and also at its Barbados range and its Antigua field artillery test site. Principal projects included gun launched satellites and “Supergun” concepts tracing back to World War I German research data relating to the Paris Gun.

He was active in financing the development of the cable television industry between 1965 and 1990, including American Television and Communications, later merged into Time, Inc., and Continental Cablevision, which ultimately became part of Comcast's systems.

In the 1980's he supported the emergence of Al technology based enterprises out of the MIT Artificial Intelligence Laboratory and other research institutions. The most notable of these enterprises was Symbolics, Inc., the first dotcom. Later he became associated with Russell Noftsker, Richard Petti, Landon Clay (benefactor of the Clay Mathematics Institute) and others in Macsyma, Inc., an attempt to migrate MIT's pioneering computer algebra system, authored by MIT Prof. Joel Moses and his colleagues on DEC timesharing computers, ported to DEC minicomputer Unix platforms (along with Unix) by Berkeley Prof. Richard Fateman, first commercialized by Symbolics, Inc. on Symbolics workstations and DEC VAXes, and finally ported to the rapidly evolving Wintel PC standard under MS Windows by Macsyma, Inc.

In 1973, together with a number of other nationally prominent venture capitalists, he worked on the organization of the National Venture Capital Associationwhich he served for many years as a founding director.

Since 1994 he has served from time to time as director of several asset management firms and as chairman of MDT Funds, a mutual fund family now known as Federated MDT Funds. He is currently a director of Tradecraft Corporation. Netcraft Corporation, a subsidiary of Tradecraft Corporation, is a patent licensing firm managed by Andrew Egendorf.