← Retour au blog
tech 22 June 2026

Everything is Logarithms: A New Perspective on Units of Measurement

Logarithms are not just obscure mathematical tools. They offer a unique perspective on how we measure and understand the world. Let's explore how everything, from bits to nats, is a matter of logarithms.

Article inspired by the original source
Everything is logarithms ↗ alexkritchevsky.com

Introduction

Logarithms are often seen as complex mathematical concepts reserved for mathematicians and engineers. However, they play a crucial role in our understanding of the world, particularly in terms of measurement and representation of quantities. In this article, we will explore how a logarithm-centered perspective can offer a new way of thinking about units of measurement, from digital to physical.

Defining Logarithms

Before diving into the implications, let's quickly review what a logarithm is. A logarithm is a function that answers the question: "How many times must a number, called the base, be multiplied by itself to obtain another number?" Formally, for a base b logarithm, \\(y = \log_b(x) \\) means that \\(b^y = x\\).

Baseless Logarithms

The concept of baseless logarithms introduces a new way of thinking about logarithms as abstract objects. Instead of tying them to a specific base, we can consider them as units of measurement in themselves. For example, \\(\log N\\) can be viewed as an independent entity, and based logarithms are fractions of these entities, as explained by Alex Kritchevsky.

Logarithms in the Digital World: Bits and Nats

In the digital world, logarithms are ubiquitous. Bits, the fundamental unit of information, can be expressed in terms of binary logarithms. \\(\log_2(N)\\) indicates how many bits are needed to represent a number N. Similarly, "nats," a unit of measurement in the context of natural logarithms ( ), are used in thermodynamics and information theory.

A concrete example: consider data storage. If you have a database containing 1 million entries, the number of bits needed to access a specific entry is given by \\(\log_2(1,000,000)\\), which is about 20 bits.

Logarithms and Scale Changes

Logarithms also allow us to handle scale changes, which is crucial in many scientific fields. For example, the Richter scale, used to measure earthquake intensity, is logarithmic. This means that a magnitude 6 earthquake is 10 times more powerful than a magnitude 5 earthquake.

In finance, logarithms are used to calculate compound growth rates and annual returns. The natural logarithm ( ) is often used to simplify these calculations and represent exponential changes over time.

Applications in Machine Learning

In machine learning, logarithms are used to compute loss functions, such as cross-entropy, which measures the divergence between two probability distributions. Logarithms help stabilize numerical calculations and avoid overflow in optimization algorithms.

Conclusion

Logarithms are more than just a mathematical tool; they are a lens through which we can better understand the complex phenomena of the world. Whether measuring digital quantities or analyzing scientific data, embracing a logarithmic perspective can transform our way of seeing things.

Let's discuss your project in 15 minutes.

logarithms measurement digital machine learning scale changes
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call