In a paper published in 1948, Claude Shannon proposed a definition of 'information' so far from the everyday sense of the word that several generations of subsequent readers have continued to be misled by the homonymy. Shannon's information was not a measure of meaning, content, or significance; it was a measure of the average reduction in uncertainty produced by a message drawn from a known distribution. The word he chose for the quantity — borrowed from a passing suggestion of John von Neumann — was 'entropy,' which carried over its own confusions from thermodynamics, where it has a related but distinct sense. The choice of terminology has cost the subsequent literature more than it has gained.
The technical content of the theory, however, has been transformative. Shannon showed that any source of messages can be characterised by a single number, the rate of its entropy production, which sets a fundamental limit on how compactly the messages can be encoded without loss. He showed further that any communication channel has a capacity, also expressible as a single number, beyond which reliable transmission is impossible regardless of how cleverly one encodes the messages. The two theorems together provided, for the first time, a theoretical foundation for the engineering of communication systems whose design had, before 1948, been a matter of accumulated craft.
The transformation has come with a persistent temptation. Because Shannon's quantity is called 'information' and is mathematically tractable, it has been imported into discussions of meaning, semantics, consciousness, and biological signalling in which the original definition does not, on inspection, do the work being asked of it. A genetic sequence carries Shannon information by virtue of the statistical structure of the symbols in it; whether it carries information in the broader sense of meaning something about the organism's environment is a separate question, on which Shannon's theory is silent. To use the same word for both is a habit of speech that no amount of qualifying footnotes has been able to dispel.
The theory survives the misuse, but the misuse continues. What the theory provides — a precise account of certain limits on encoding and transmission — is genuine and has not been superseded. What it does not provide — an account of why some messages mean what they do or matter to anyone — is left untouched by the formal apparatus. To remember the distinction is to keep both the theory and its limits in view, and to resist the suggestion that calling something 'information' amounts to having understood it.
The primary purpose of the passage is to: