Skip to content
All articles
thermodynamicsphysicsinformation theory

What is Entropy?

Entropy measures disorder and irreversibility — from steam engines to information theory to the arrow of time.

8 min read · 2026-06-06

Boltzmann defined entropy as S = k_B ln Ω, where Ω counts microstates. Clausius gave the thermodynamic definition dS = δQ/T. Shannon later reused the same mathematics for information.

The second law

Entropy of an isolated system tends to increase. Hot coffee cools. Perfume spreads. Eggs don't unscramble. Time has a direction because probability favors disorder.

Entropy in the digital age

Shannon entropy H = −Σ p log p measures information surprise. It sets limits on compression, guides machine learning loss functions, and connects thermodynamics to data science.

Explore interactively

Open equation page