Frances Allen, a computer scientist and researcher who helped create the fundamental ideas that allow practically anyone to build fast, efficient and useful software for computers, smartphones and websites, died Tuesday, her 88th birthday, in Schenectady, New York.
Her death, in a nursing home, was confirmed by her great-nephew Ryan McKee, who said the cause was Alzheimer’s disease.
In the mid-1960s, after developing software for an early supercomputer at the National Security Agency, Allen returned to her work at IBM, then the world’s leading computer company. At an IBM lab in the Hudson River Valley town of Yorktown Heights, just north of New York City, she and her fellow researchers spent the next four decades refining a key component of modern computing: the “compiler,” the software technology that takes in programs written by humans and turns them into something that computers can understand.
For Allen, the aim was to do this as efficiently as possible so programmers could build software in simple and intuitive ways and then have it run quickly and smoothly when deployed on real-world machines.
Together with researcher John Cocke, she published a series of landmark papers in the late 1960s and ’70s describing this delicate balance between ease of creation and speed of execution. These ideas helped drive the evolution of computer programming — all the way to the present day, when even relative novices can easily build fast and efficient software apps for a world of computers, smartphones and other devices.
In 2006, on the strength of this work, Allen became the first woman to win the A.M. Turing Award, often called the Nobel Prize of computing.
“She was crucial in providing easier and easier ways for humans to tell computers what to do,” said professor Michelle Mills Strout, who teaches computer science at the University of Arizona and specializes in compiler technology.
Frances Elizabeth Allen was born Aug. 4, 1932, in Peru, New York, near Lake Champlain and about 30 miles from the Canadian border. Her parents, John and Ruth (Downs) Allen, owned a dairy farm, and she, the oldest of six children, grew up there without running water or electricity. Electricity did not arrive until the early 1940s, and even then it ran only to the barn, not to the family home.
Allen attended a one-room school less than 1 mile away and did her part on the farm, from milking cows to helping with the field work. After graduating with honors from the local high school, she studied at the New York State College for Teachers (now the University at Albany, part of the State University of New York).
Returning to Peru, she worked briefly as a teacher. Her sister, Catherine, was among her students.
Allen earned a master’s degree in mathematics at the University of Michigan in 1957 and took a job with IBM at an office in Poughkeepsie, New York, as a way to pay off her college debt. Although she had planned to return to teaching, she stayed with the company for the next 45 years.
At first Allen taught incoming employees how to use a new programming language called Fortran. Previously, engineers had programmed computers using a language of ones and zeros, which was understood by the computer hardware. With Fortran, one of the first high-level programming languages, they could build software in more intuitive ways, without mastering the arcane operations of the computer hardware.
It was a concept that Allen would push to new heights.
Allen joined the top-secret effort to build a supercomputer at the National Security Agency in the early 1960s. (In doing a background check on her, government officials had descended on her hometown to talk to local farmers about her.)
The NSA machine, called Stretch-Harvest, was intended to analyze communications intercepted by listening posts operated by U.S. spies around the globe. Allen helped build the machine’s programming language and compiler.
In those early years of computer design, compilers were terribly inefficient. Programmers could build software without learning the minutiae of the hardware, but when the compiler converted their programs into ones and zeros, they were far too slow and took up far too much space.
As part of an IBM research project created in the late 1960s, Allen worked to change this dynamic. In the beginning, she and her colleagues built more efficient compilers for the massive mainframe computers of the day. In later years, they applied similar ideas to “parallel computing,” a newer technique that spread digital tasks across multiple computers.
The result, several decades on, was modern computer programming. Programmers can now build smartphone apps, like Facebook, that respond to the touch without delay, delivered from vast computer data centers spanning tens of thousands of computers.
Allen’s work plays into “pretty much every software system anyone uses: every app, every website, every video game or communication system, every government or bank computer, every onboard computer in a car or aircraft,” said Graydon Hoare, creator of a programming language called Rust.
“Without good compilers,” he added, “the whole world of software would be much slower, costlier, more error-prone, less capable.”
Allen’s marriage to Jacob Schwartz, a computer science professor at New York University and one of her collaborators on compiler research, ended in divorce. She is survived by two brothers, Phillip and James Allen, and her sister, Catherine McKee.
In a field long dominated by men, Allen was a force for change. In the 1970s and ’80s, thanks largely to her own efforts, women accounted for half the experimental compiler group inside IBM.
“One the many things Fran did was attract women to her field,” said Jeanne Ferrante, who worked alongside Allen for more than a decade. “She looked out for the people who were underrepresented.”
In 1989 she became the first female IBM fellow, a rare honor bestowed on the company’s leading engineers, scientists and programmers. But when she received her award at an IBM retreat in Southern California, the company identified her as a man (“In recognition and appreciation of his outstanding technical contributions … ”).
The award — including the mistake — remained on her office wall until she retired in 2002.
“She broke the glass ceiling,” said Mark Wegman, another IBM fellow, who worked with Allen for decades. “At the time, no one even thought someone like her could achieve what she achieved.”