Fortran developer John Backus dies
Another founder of modern computing passes away
Obituary John W Backus, team leader of the original FORTRAN development team at IBM, died on 17 March at the respectable age of 82.
I never met John Backus, but he was responsible for changing my life – and probably that of many other scientific programmers of my generation.
I was at the Research School of Chemistry at the Australian National University (failing a PhD due to the rival attractions of amateur opera) when I got interested in the programs various researchers used to generate their results – and whether these were ever verified as the context they were used in changed. I remember one spurious piece of research in a very respectable journal that turned out to depend on determining the intersection of two almost parallel lines – obviously, systematic errors made the published research results (complete with standard deviations) worthless.
So I went off to the physics school and got myself taught FORTRAN (I still spell FORTRAN in CAPS, FORmula TRANslation; or, originally: "The IBM Mathematical FORmula TRANslating System"). It was a very good way of learning programming style. Our programs were thrown back at us if they had more than the minimum GOTOs possible and we were expected to code in a style which meant that we could still work out what the programs did when we came back to them after a month or so (a possibility that was a total revelation to some of the academic programmers I knew). And that, after disillusionment with research chemistry set in and I got bored with a brief period as a hifi salesman, is how I ended up in IT.
FORTRAN was simply the right language for mathematical programming at the time (I have real issues with its use for general business-oriented programs, but let's not go there). And, roll forward a decade or so, and I remember being shown the first release of Windows NT. Were there any non-Microsoft compilers for this platform? Just one – FORTRAN – or, at this stage, probably Fortran (I think it went mixed case with Fortran 90). Today, FORTRAN is an ISO standard and the latest Fortran 2003 supports object orientation and generic programming.
Not a bad legacy for Mr Backus, who was given a National Medal of Science in 1975, the Turing Award from the Association for Computing Machinery in 1977, and the 1993 Charles Stark Draper Prize from the National Academy of Engineering.
Whatever you think of FORTRAN today, his great contribution was to demonstrate that computers could run efficiently without being low-level coded in Assembler – apparently, he found coding in assembler tedious (as an erstwhile assembler database programmer, I can only agree).
FORTRAN was the first high-level computer language when its first compiler was delivered in 1957.
Backus' name is also remembered in Backus-Naur form (BNF), the notation used to define language syntax formally.
He was also the principal author of the Algol 60 Revised Report (coincidently, probably my choice of Fortran replacement, and the main language used on my computer science course).
And, his later area of interest was the field of functional programming (his seminal lecture, "Can Programming Be Liberated from the von Neumann Style? A Functional Style and Its Algebra of Programs" is here). This focuses on describing the problem a computer is expected to solve rather than giving the computer step-by-step instructions. There's a guide to the impressive range of Backus' work here.
Backus had an interesting scholastic and academic career, broken up by a spell in the army, and culminating in a master's degree in Mathematics at Columbia University in New York. He wandered into IBM, was hired on the spot as a programmer (he joined in 1950), and remained with IBM until he retired in 1991. He truly was one of the founders of modern computing. ®