John Warner Backus

My Work for/with John

Gene Amdahl's ``baby'': The Menlo Park Advanced Computing Systems laboratory was effectively dissolved during the summer of 1969. My manager Ed Sussenguth was kind to suggest IBM Research in San Jose that I be transferred there. So by July, I commuted 40 miles south (by car, a new Saab 99E) instead of 4 miles west (on bicycle, a Remington)! IBM Research San Jose was then located in building 25, a rancho-style, ``patio-filled'', one floor ``spread''. I immediately liked the place, its Library was significant. It took some time before management decided on my job assignments -- meanwhile I continued some (more-or-less) technical work on the design of data communication protocols and, later, on language oriented computer architectures.

I shared office with an Indian. I have forgotten his name. He wondered who the cowboy jeans clad `IBM Fellow' across from our office was. He had noted I spoke with him whenever he -- rather seldomly, sort of every other week -- came in for a day. Who was he? I told him: the inventor of Fortran, a co-designer of Algol 60, John W. Backus. After some time it dawned on my Indian roommate that Backus was indeed a ``very famous'', and hence in his eyes he, John, also had to be a very important, influential figure. Then he told his story: My Indian colleague got his PhD from University of British Columbia. He had read, back in Bombay, while doing his MSc, some papers by a prominent Canadian professor. Wrote him. Asked whether he could study under him. After having read some of my Indian room mates' papers, Oh Yes! Then he had to get his Bombay professors' accept. Made an appointment with his secretary to meet him. Had in fact never seen him. Although his professor, teaching assistants had been his only ``interface''. Three months later the day and hour of the appointment came up. He sat there. All day. In his front office. Waiting to be called in. At the end of the day the (male) secretary told him: Sorry, he hasn't come back from New Delhi! Another appointment, weeks later, also ended in a ``no show''. In the end the Canadian professor said: just come. In this way he met the Anglo-Saxon/North-American academic ideals: easy going, no formalities. Needless to say: his Indian professor was virtually an unknown quantity anyway. And in this way he expressed his great surprise at Backus' ``uniform''. My Indian colleague was always in a dark suit, white shirt, necktie, ``the works''.

My conversations with John Backus continued.

John had had no-one to discuss his search for solutions to vaguely, truly scientifically posed problems, for more than 10 years. After his seven to eight years on various Fortran language and compiler development projects for IBM he was made an IBM Fellow and left on his own: to do whatever pleased him at a very agreeable salary. But, as I had been told at ACS in Menlo Park, Backus was working all alone. Was last seen, driving across the Bay Bridge, for seminars at UC Berkeley in 1968. Well, I saw his name on the wall, to the left of his door. On a day he came in I entered the always open door. Told him who I was, what I had studied, and ``faked'' greetings from Peter Naur. (I wrote Naur about this soon after and got his late permission!) Over his next visits he told me, in a very engaged way, about his studies. I then commented: it looks ``similar'' to the $\lambda$-Calculus. What was that? So I told him. The next time he said: No, his calculus had no variables. So then I told him about Curry's Combinatory Logic. That was more like it. Meanwhile our group, as well as John, had been moved to some barracks, ``containers'', while we were all waiting for the new, very large and very ``luxurious'' building 28 was being built. One day John took me into my boss: Asked that I work for him. He really hadn't inquired with me. My boss then interviewed me ``privately''. Told me his impression of Backus -- something that clearly, and shockingly illustrated that our managers did not understand what real science, what real research is. Then he asked me: Dines, where do you want to be five years from now? A typical management question, and certainly a typical IBM-career one. So I pondered, unexpectedly, for a while, maybe 20 seconds or more, to respond: Not in that chair there, and I pointed at his. I got the permission.

Privately Kari and I invited Barbara and John down to our house in San Jose, for dinners or lunches, 2-3 times -- serving some local rather good vines. This lead to John mildly accusing me of having brought him back to drinking ! But more often we went up to San Francisco, to St. Germain Ave. Sometimes around noon, Saturdays, sometimes in the evening, typically Fridays, we went to SF. First had a glass of vine at their home, afterwards going out to a local restaurant for delicious food. So what could we possibly talk about: Not politics. There we basically agreed, although with more-or-less gentlemanly ways of expression. John's WC bowl had a picture of Richard Nixon glued to its more-or-less horizontal part. Barbara was one of these social sciences writers, too few of them, who really studied women's causes. At the time we first met Barbara was studying/writing on Ohio womens rights to carry ``own'' names into their marriage, as it was in the 19th century to their loosing that right in the 20th century. Her writings, it seemed to me, to provide the necessary, solid fodder for the more visible womens liberationists. I told them that my grandmother had been in prison for being a suffragette, and had later started a political party for which an uncle and an aunt of mine had been in Parliament. And that my mother had also ``stood'' for Parliament.

Working for John was both fascinating and difficult. He was different in everything. He came down only once every other week from his fine modern, wooden California architecture house on the slope of Twin Peaks in San Francisco: St. Germain Eve. He then had a Volvo 180! My job was to implement his language(s). Initially it was called RED/1, for ``the first reduction language''. I did so by simply mapping he ``rewrite system like rules'' into pairs of trees some of whose leaves were variables and stood for any subtree of a given of four - five different syntactic categories. Then I `automated' the construction of a finite state automaton which would traverse the source tree until it found an embedded tree matching a left hand side tree of one of the rules. If one such was found the algorithm copied and deleted subtrees onto a new source and the traversal was restarted at some place! If none was found the job was done! Often Backus would call, sometimes in the week ends, to either complain that my system, whose implementation he then knew nothing about, was probably wrong -- and he then explained. I would always, I am sure always, counter that he probably meant to change one of his semantic rules. So he did and would call me later the same or next day. Often he worked more intensely in the week ends so I would bicycle over to IBM, some 8 miles east on Blossom Hill Road and recompile the system with his new rule(s). One day, finally, he asked me how I had implemented the system. So I told him. In fact I showed him an almost complete report that I had sometimes mentioned to him that I was working on. Finally he understood and exploded. I had not been helpful etc., etc. Perhaps he was right. Perhaps he never really wanted to listen in those days and earlier. In any case, after 18 months the ``explosion'' led to him wanting me off! I remember walking around the Homestead, rather despondent. What now? I had really not been forming any career goals yet. It was 1971 in September, and I was getting on 34 years of age. So our first ``departure'' was less than harmonious. In April 1974, as Peter Lucas and I checked into our hotel in Palo Alto on a first visit back to the SF Bay Area after I had started in Vienna, there were telephone messages for us both: they `commanded' us both for dinner at Backus' coming Friday, i.e. next day! At that dinner John, somewhat edited, conveyed the above to Peter -- who had my story heard years before. But now it had another, extended ending: When John's first submitted paper was conditionally rejected on the grounds that it was nothing but a simple tree transformation system as I had several times explained him, then he started understanding that he really didn't want those meta-variables. And it was this way he really discovered FP and FFP. Then he sort of ``regretted'' that he hadn't listened. He had actually noted in his Design Book (a diary of a sort) my remarks on this topic on several occasions: Listen here John: you meta-variables are just variables, but you do not want variables. Here is what Curry does. But his mind was then elsewhere! By inviting us he wanted to redress and excuse. John was always a very kind and considerate man.

After that I kept in touch with John and Barbara throughout the many years. My relation with Backus probably ended this February (8th, 1997) when I believe I visited him for the last time in San Francisco.

An Obituary: John Warner Backus, 3.Dec. 1924-17.March 2007

I was asked by the editors of the Springer Formal Aspects of Computer Programming to compose an obituary for John.

John W. Backus, the main creator of Fortran, co-creator of Algol 60 and FP (a variable-free variant of functional programming), and the B ``behind'' BNF, died on March 17, 2007, 82 years old, at Ashland, Oregon, near one of his two daughters. Born, December 3, 1924, in Philadelphia, to well-to-do stockbroker parents, John Backus became, from 1954, one of the most influential people in computing.

John Backus in many senses was beyond `pedagogical reach' and thanks for that !

John Backus dropped in and out of schools, studied chemistry for one year (Univ. of Virginia), joined the army in 1943 but was soon rerouted to a pre-engineering programme at Univ. of Pittsburgh, then studied medicine at Haverford College and worked at the Atlantic City hospital. Fell ill with a brain tumour, was operated, but the inserted plate did not fit so John Backus invented a better one. Then quit the medical field and moved to New York not knowing what he wanted -- except that he wanted a hi-fi set -- liking music -- could net get a good set so he built one himself after having learned radio basics at a technician's school. There he helped a teacher doing some mathematical calculation, liked it, and then went to Columbia Univ. from where he visited the IBM World HQ (and computer center) in New York. That lead to his first, and only employment. From September 1950 and till his retirement in 1991 John Backus worked all his life at IBM, more than 40 years -- surely, and in no small measure due to John Backus's work, the most condensed period of scientific and technological achievements of any technology.

Within four short years John Backus was engaged in (i) the programming of the IBM SSEC (`Selective Sequence Electronic Calculator') for, amongst many things, the calculation of lunar orbit tables later used in the US Apollo project; (ii) for the IBM 701 (the ``Defense Calculator'' used during the Korean War), with Harlan Herrick, inventing (and implementing an interpreter for) the `Speedcoding' language simulating floating point numbers on an integer machine; and (iii) designing, later, the first hardware floating point unit for the IBM 704 and persuaded IBM to let the IBM 704 provide for indexing. All this lead up to John Backus proposing, in December 1953, and persuading IBM to start, contrary to the advice of John von Neumann, the ``automatic programming'' project that lead to the November 1954 report `The IBM Mathematical FORmulaTRANslating system FORTRAN' for which the FortranI compiler was available early 1957 -- the result of more than three years of work by John Backus, Robert Nelson, Harlan Herrick, Lois Haibt, Roy Nutt, Irving Ziller, Sheldon Best, David Sayre, Richard Goldberg and Peter Sheridan.

After the Fortran years followed years where John Backus was engaged in the committee work for the IAL (`International Algebraic Language'), a joint effort of GAMM (`Gsellschaft für Angewandte Mathematik') and ACM - started at Zürich May 27 to June 1, 1958. The work lead to Algol 60. Backus' involvement with Algol 60 ended sometime in 1963. Up till this time John Backus had worked ``out of'' New York. Two years as visiting professor at UC Berkeley (Calif., 1963-1965) saw Backus move from the US East to the US West Coast. He spent the rest of his IBM career at IBM Research in San Jose, Calif. In the sixties Backus, now as one of the first IBM Fellows, pottered around studying the four color problem, participating actively in the anti-war and the anti-Star Wars movement, etc.

From around 1969 (maybe even earlier) John Backus got interested in what he first called closed applicative languages. For a short while the author of this obituary worked for John Backus, implementing the first versions of these, the reduction languages which later evolved into `FP' and `FFP'. During John Backus' last years at IBM, mid 1970s to 1991, with a group of people, including Alexander S. Aitken, Peter G. Lucas, John H. Williams and Edward L. Wimmers. FP evolved into FL: Function level programming.

A unique rainbow spanned the professional life of John Backus: from programming the very low level SSEC to FL there seems, to me, to be a clear, direct line: FL ``arches'' back to fundamentals of the SSEC.

John Backus was honoured several times: 1967: The IEEE W. Wallace McDowel Award; 1975: The [US] President's National Medal of Science; 1977: The ACM Turing Award; 1993: The Charles Stark Draper (US$ 375,000) Prize - to mention a few.

When John Backus retired, he retired completely from the field of computing. For a little more than ten years he stayed on in his spectacular San Francisco home before he, after his second wife's passing away, after 35 years of marriage, moved to Ashland, Oregon in 2004. Here he spent a few years, close to one of his daughters, Paula, and a grand child, Ally, and also, more often, met his other daughter Karen of New York. In his later years he studied the writings of Krishnamurti and Eva Pierrakos and practiced meditation. When he experienced loss of other than very short term memory he ended his own life.

John Backus is quoted as having said: Most scientists are scientists because they are afraid of life. It is most wonderful to be creative in science because you can do it, and - without clashing with people and suffering the pain of relationships - make your way into the world. It is a wonderful `out' - it is sort of an aseptic world where you can use the very exciting faculties you have and not encounter any pain, The pain in solving a [scientific] problem is `small potatoes' compared with the pain in living. ... It's strange that by looking into yourself you really get an appreciation of the mystery of the universe.

John Backus was a peaceful man, almost ``ruthlessly'' direct; interested, in a very kind way, in the personal lives of his collaborators; shunned ``Corporation Speak'' and let those who spoke so, in no uncertain terms. He did not care about other people's opinion about him, John Backus.

His indefatigable search for understanding, his ways of expressing how he worked, his consistency in the matters we, his collaborators, knew him -- all appears to attest to a man who must have thought about all these things and much much more ever since his childhood.

He not only left a clear mark on our profession, but he also left an indelible mark on all those he touched.