Skip to content
← All lenses✦ Lens · Information theory

Signal and noise

A way of reasoning built from the published work of Claude Shannon. It is a framework, not the person.

InstinctSignal, noise and uncertainty

Which distinctions survive transmission, and which are lost?

Life
Petoskey, Michigan 1916 – Medford, Massachusetts 2001
Horizon
1937–1961
In Feynman
/lens shannon
See membership
✦ Their times

The times that shaped it.

Shannon studied electrical engineering and mathematics at Michigan, then at MIT showed that relay circuits could be designed with Boolean logic. At Bell Labs from 1941 he worked on fire control and cryptography during the war, and met Alan Turing when Turing visited in 1943. The telephone network was his setting: engineers needed to know how much could be sent down a noisy line, and how reliably. He became a professor at MIT in 1956.

✦ The ideas

What it reasons with.

Each idea is tied to the work where it appears and the year that work was published.

  • 1938 · A Symbolic Analysis of Relay and Switching Circuits

    Logic in hardware

    Boolean algebra can describe and simplify switching circuits, so logical reasoning can be built out of switches. It became a basis of digital circuit design.

  • 1948 · A Mathematical Theory of Communication

    Information as resolved uncertainty

    Entropy measures how uncertain a source of messages is, and information is the reduction of that uncertainty. Meaning is set aside on purpose: the theory concerns how reliably symbols can be reproduced, not what they say.

  • 1948 · A Mathematical Theory of Communication

    Capacity and noise

    Every noisy channel has a capacity. Below it, suitable coding can make errors as rare as you like; above it, no code can.

  • 1951 · Prediction and Entropy of Printed English

    Redundancy

    English is highly predictable from context, which is why readers can correct errors and fill gaps. Redundancy is what protects a message against noise.

  • 1956 · The Bandwagon

    The limits of the theory

    He warned that information theory was being applied too freely to psychology, economics and biology, and that its results hold only where its assumptions do.

✦ The questions

How it tests an idea.

Its characteristic question first, then the questions that follow from its ideas.

  1. Which distinctions survive transmission, and which are lost?
  2. What is the signal here, and what is noise that only looks like signal?
  3. Is more being claimed than the channel, or the data, could actually carry?
  4. Is a measure of information being mistaken for a measure of meaning or truth?
The horizon

What it could know.

Born 1916Horizon 1937–1961Died 2001

Inside

It reasons only with ideas documented between 1937 and 1961. It closes in 1961; Claude Shannon lived until 2001.

Anything later reaches it only through the evidence a review retrieves, labelled as supplied.

Beyond

His major research had largely ended by 1961. Later coding theory, the internet and modern machine learning lie outside the lens.

The cutoff is deliberate: a lens that can’t see today’s consensus can’t simply repeat it. This card hasn’t yet been reviewed by a scholar of Claude Shannon’s work.

✦ Blind spot

Where it fails.

Accurate transmission establishes neither meaning nor truth. The theory sets meaning aside by design, and Shannon himself warned against stretching it to fields where its assumptions fail.

✦ Sources

Where it comes from.

Works by Claude Shannon, dated to first publication, and references about them.

Links checked 31 October 2026. The notes on this page paraphrase. Nothing here quotes Claude Shannon, and the lens never speaks as them.