John Backus

1924-

American Computer scientist

During the 1950s, John Backus headed the pioneering group at IBM that developed FORTRAN, the first widely used computer programming language. He was also instrumental in the development of ALGOL, which, though it did not gain wide commercial usage, had a significant influence on three languages that did: Pascal, C, and Ada. Winner of the National Medal of Science, the Turing Award, and the Charles Stark Draper Prize, Backus has continued to work on programming languages even after his retirement from IBM in 1991.

Backus was born in Philadelphia, Pennsylvania, in 1924, and grew up in Wilmington, Delaware. When he enrolled in the University of Virginia in 1942, he planned to major in chemical engineering, but these plans were stymied when, after one semester, he was expelled for skipping classes. At the time, America was embroiled in World War II, and Backus, divested of his academic commitments, was promptly drafted into the army in early 1943.

He first served in an anti-aircraft unit, but in September 1943 was sent to study engineering at the University of Pittsburgh as part of a specialized army training program. After completing his army engineering education in March 1944, Backus spent six months undergoing premedical training at a hospital in Atlantic City, New Jersey. This was followed by another six months at Flower and Fifth Avenue Medical School in New York City.

In May 1946 Backus left the army, and his interest turned from chemistry and medicine to a field that would prove a bridge into the realms of computer science and mathematics. Still living in New York City, he enrolled in the Radio Television Institute, a training school for radio and television repairman. While taking the course, he became so intrigued with mathematics that he began study at Columbia University, which granted him a B.A. in mathematics in 1949. A year later he earned his M.A. in mathematics from Columbia.

After graduating, Backus went to work for IBM, a company staking territory in the newly emerging field of computer science. Backus knew little about computers (few people did in those days) but he soon found himself on the cutting edge. In 1952 he led a group of researchers who produced the Speedcoding system for the IBM 701 computer, and a year later wrote what would prove to be a historic memo. In it he outlined for his boss, Cuthbert Hurd, the need for a general-purpose, high-level computer programming language. This was the origin of FORTRAN, an acronym for formula translator.

FORTRAN was the prototype for modern compilers—computer programs that translate high-level language statements into a form that computer hardware can read. Before FORTRAN, programmers had been forced to endure the tedium of logging in rows of zeroes and ones, the binary language of computers. FORTRAN allowed programmers much greater freedom and creativity, and enabled wide-ranging advancements—not least because, prior to its development, three-quarters of the cost of running a computer came from debugging and programming. IBM published FORTRAN I in 1954, and in the following year Backus, along with R. A. Nelson and I. Ziller, began working out the bugs in this first version of the language.

At the same time, two schools of thought about computer languages emerged: American researchers believed that programmers should develop languages as needed by users; a group of European computer scientists, led by F. Bauer of the University of Munich, Germany, held that this proliferation would create confusion. As a result, Backus, Bauer, and others met in Zurich, Switzerland, in the spring of 1958, and produced a report calling for an International Algebraic Language (ALGOL). Soon after the Zurich meeting, Backus went to work with the team developing ALGOL at IBM's Watson Research Center in Yorktown Heights, New York. The resulting language did not enjoy widespread commercial usage, but did have a great impact due to its influence on Pascal, C, and Ada.

During the decades since the development of ALGOL, Backus has worked tirelessly to refine computer languages, and in 1978 wrote a paper in which he called for their restructuring. He became an IBM Fellow in the early 1980s, which gave him an opportunity to step up the pace of his research. In 1991 Backus retired from IBM. He continued, however, to work on developing what he called a "functional" language—one that borrowed from already defined languages, eliminating the programmer's need to "reinvent the wheel" by spelling out each instruction in detail.

JUDSON KNIGHT