layman/terms
Inside an app

Algorithm

al · guh · rith · um

In shortA precise set of steps for solving a problem or making a decision, like sorting a list or choosing what to show you next.

“The algorithm decided.” It sounds mysterious. But an algorithm is just a set of steps, like a recipe.

A bread recipe says: mix flour, water, yeast and salt. Knead for ten minutes. Leave to rise for an hour. Bake at 220 degrees for thirty minutes. Follow the steps exactly and you get the same loaf every time.

An algorithm is a recipe for a computer. It’s a precise set of steps for solving a problem or making a decision.

Algorithms you meet every day

Simple and complicated

Some are a few steps, like “sort these names A to Z”. Others, especially those built with machine learning, involve millions of steps that even their makers can’t easily explain.

What it isn’t

It isn’t a mind. An algorithm doesn’t want anything. It follows the steps it was given, which is why the choices behind those steps matter so much.

Where the comparison breaks down

A recipe’s steps are written by a cook. Many modern algorithms are partly “learned” from data instead, so they can pick up patterns nobody intended, including unfair ones.

How the words fit together

An algorithm is the method. Code is how it’s written down. Machine learning is a way of building algorithms from examples.

Good to know

When people say social media's “algorithm”, they mean the steps that decide which posts you see first. It's usually tuned to keep you scrolling.

When you'll hear it

“The algorithm pushed it.” The steps that decide what people see showed this post to lots of them.

“It's a sorting algorithm.” A set of steps for putting things in order.

“The algorithm is biased.” Its steps, or the data it learned from, treat some people or things unfairly.

Say it in a meeting

“Could the way our system decides what each customer sees first be unfair to anyone?”