Claude Shannon

Overview

Claude Shannon was an American mathematician and electrical engineer who worked on the mathematical side of communication. He was born on 30 April 1916 in Petoskey, Michigan, and died on 24 February 2001.^[1] His career was divided mainly between two institutions: the Bell Telephone Laboratories site at Murray Hill, New Jersey, and the Massachusetts Institute of Technology, where he finished his studies and later returned as a professor.^[1][2]

Education and early research

Shannon took bachelor's degrees in electrical engineering and in mathematics from the University of Michigan in 1936.^[1] He then went to MIT, where from 1936 to 1940 he was a research assistant in the Department of Electrical Engineering.^[1] In 1940 MIT awarded him an S.M. degree in electrical engineering and a Ph.D. in mathematics.^[1][2] The summer of 1937 he spent at Bell Labs — an early stay at the institution where he would afterwards work for nearly two decades.^[1]

A mathematical theory of communication

His best known work is a paper rather than a book or a machine: "A Mathematical Theory of Communication", published in two parts, in the July and October issues of the Bell System Technical Journal, volume 27, in 1948.^[2]

The problem the paper addressed is easy to state and hard to answer. Sending a message — a telegram, a telephone call, a picture — means choosing one possibility out of many at the sending end and getting that choice recognised at the receiving end, across a channel that may be noisy, costly or slow. Before the paper these were engineering questions settled case by case, with rules of thumb about how much signal a particular line needed. Shannon's move was to strip away the particular apparatus and ask what is true of any communication system at all: what a message is, mathematically speaking; what it means for a message to arrive reliably; and where the limits of any channel lie.^[2]

Framing the subject that way changes what can be said about it. Once messages are treated as mathematical objects and channels as having definite limits, claims that were previously matters of experience can be posed as propositions to be proved or refuted, and a design can be compared with a known best case instead of only with rival designs. That is the sense in which the paper mattered: it did not build a device, but it gave communication a shared vocabulary and a set of questions sharp enough to be answered. The field that grew out of it came to be called information theory, and its terms and problems were later taken up far beyond telephone engineering — in computing, in the design of storage and transmission systems, and in any discipline that has to talk precisely about how much a signal can carry.

It is worth being clear about what the paper is not. It is not a handbook of circuits or a catalogue of equipment; it is an argument about the general shape of the problem, written for readers who designed real systems. Its two-part appearance in a journal read mainly by telephone engineers is a reminder of where the questions came from.^[2]

Career

Shannon worked at Bell Labs in Murray Hill, New Jersey, until 1956.^[1][2] In 1956 he returned to MIT, at first as a visiting professor, and a year later he became professor of communications science.^[1] He served on the MIT faculty until 1978.^[2] In 1966 he received the IEEE Medal of Honor.^[1]

His path — a decade and a half in an industrial laboratory, then a professorship at the university where he had been a student — was a common one for researchers of his generation, and it shaped the kind of work he did: problems that arose in operating telephone systems, formulated with the generality of a mathematician.

Influence

Shannon died on 24 February 2001.^[1] The questions he set out in 1948 remain the starting point for anyone who needs to reason about how information is represented, transmitted and received, and the vocabulary of that paper is still in ordinary use among engineers.^[2] His name is attached less to a product or a company than to a way of stating the problem.