John Tukey Suggested the Word "Bit" to Claude Shannon and Printed the Earliest Known "Software" in 1958

Claude Shannon’s 1948 paper that started information theory needed a name for its unit. Base-2 logarithms gave you binary digits, which is a mouthful.

So on the paper’s first page, Shannon credited a colleague. The units “may be called binary digits, or more briefly bits, a word suggested by J. W. Tukey.”

“Bit” is the clean half of the story. Shannon’s sentence goes on: “A device with two stable positions, such as a relay or a flip-flop circuit, can store one bit of information.” The word was Tukey’s suggestion, and Shannon’s paper is where it went into print.

“Software” is messier. Tukey used it in “The Teaching of Concrete Mathematics,” in the American Mathematical Monthly in January 1958. “Today the “software” comprising the carefully planned interpretive routines, compilers, and other aspects of automative programming are at least as important to the modern electronic calculator as its “hardware” of tubes, transistors, wires, tapes and the like,” he wrote.

The quotation marks around “software” matter. Tukey wrote it like a word his readers might not know yet, which fits someone using a term as easily as someone inventing it.

Shapiro found the passage. “I published an article in 2000 in the IEEE Annals of the History of Computing, antedating the OED’s first use of the word “software” by finding it used by statistician John W. Tukey in 1958,” he wrote on the American Dialect Society’s mailing list in 2014.

Shapiro still credits Tukey with both words in the loose sense. “Tukey was a legitimate master word-coiner, also coining the computing term “bit,”” he wrote in 2021. In 2024 he added that “Paul Niquette’s vociferous claim that he introduced “software” in 1953 is wholly unsubstantiated.”

Words like these get credited to whoever lands in the record people search. The Times obituary did more for Tukey’s “software” credit than his article did. Expect the same with today’s AI terms: the first widely quoted paper or post will get the credit, whoever said it first in a meeting.

Digitized archives make each of these claims easier to move. Shapiro’s antedating came from searching old journals, and the more of them get scanned, the more likely an older “software” in the computer sense turns up.

Shannon’s paper doesn’t say when or how Tukey made his suggestion. And the OED still dates the computer-program sense of “software” to 1969. “I do not understand why 1969 is the first use, when their entry for “application software” has a 1961 citation,” Shapiro wrote in 2024.