Father of Fortran programming language dies

His lifelong mission after joining IBM Corp. in 1950 as a programmer was to work on ways to simplify computer programming. The vendor encouraged him to set up a research project and hire a team of developers. In 1957, the fruits of their labors appeared with the debut of the Fortran programming language.

"Fortran changed how people wrote programs on machines and also changed the way compilers were built," said Fran Allen, fellow emerita at IBM, in an interview on Tuesday. "It was a gigantic step at the time."

Allen's first job on joining IBM Research in 1957 was to teach Fortran to the company's scientists. "They weren't very happy about having to learn a new language," she said, explaining that the scientists were used to hand-coding their programs or working closer to the computer's machine code.

Fortran gave scientists and engineers a higher-level, precise language based on familiar algebra formulas in which to write programs. Not only did the new language simplify sharing code, but also made it much easier for scientists to express the solutions to the problems they'd programmed the computer to solve.

Most importantly, the performance of problems coded in Fortran running on computers was "exceedingly good," better than that of hand-coded programs, Allen said. "John knew that for Fortran to be accepted it had to perform at least as well as hand-coded programs," she added. "That was his stated goal -- to do that every time and he achieved it."

Fortran was adopted very quickly and became the programming language of choice for the engineering and scientific communities for many years and is still in use today. The influence of the compiler that Backus and his team built as part of Fortran back in 1957 can still be seen today in the way compilers are designed, Allen said.

Post-Fortran, Backus worked with Danish computer scientist Peter Naur to develop a notation to describe the structure of programming languages, dubbed the Backus-Naur form. The notation acts in the same way as a grammar for a language like English. Instead of formalizing the definition of verbs, nouns and adjectives, the Backus-Naur form formalizes the notation of variables and operators.

For his Fortran work, Backus won a number of awards including a National Medal of Science in 1975 and the 1977 Turing Award from the Association for Computing Machinery.

Backus remained with IBM until he retired from the company in 1991.

Backus was a bit of a maverick, wearing jeans to work and to events at a time when most IBM staff dressed more formally, Allen said. "He was a very precise person, but not a fuss," she added. "He was always seeking to understand things better and was always interested in new ideas. He was a person who was very pleasant to be with. You always had the feeling he was very interested in things around him."

According to Allen, Backus liked to relate how in the four years he and his team were working on Fortran, he'd always tell his project manager that the project was six months from completion. "He actively proposed that failures should be rewarded as well as success," Allen said, adding that Backus very much believed in risk taking to make advances in computing.

Backus died at his home in Ashland, Oregon, on Saturday. According to published reports, his family didn't know the cause of his death other than age.