Get all your news in one place.
100’s of premium titles.
One app.
Start reading
The Guardian - AU
The Guardian - AU
World
Nick Evershed

The Crunch: how much water and energy ChatGPT uses – and the singability of Taylor Swift songs

Composite image showing a UK police officer, Taylor Swift, and a bottle of water
What do British police, water and Taylor Swift have in common? They're all in the Crunch newsletter, of course! Composite: Alamy

Hello and welcome to another edition of The Crunch.

In this week’s newsletter we have charts on how much water and energy is consumed every time ChatGPT is used, some very nice maps of election search trends, the demographics of people arrested for riots in the UK, the growing divide between men and women in employment and university attendance, a round-up of climate emergency charts, and a visual analysis of Taylor Swift’s most recent album.

But first … what’s with all the smoke shops?

In Australia, tobacco and vape shops have been popping up all over the place. In our most recent entry for our One Big Chart series, Nick used business registry data to chart the rise of smoke shops, and see whether the government’s crackdown on vape imports is making any difference:

You can read more here.

And if you’re in the mood for illustrated charts, Josh drew one of the estimated lifetime greenhouse gases from three recently approved New South Wales coalmine expansions.

Four charts from the fortnight

***

1. What the data tells us about the 2024 UK rioters

The Guardian has scrutinised data that provides insights into the lives of those accused of taking part in the unrest.

The findings provide the first comprehensive picture of the alleged rioters – and challenge some of the assumptions made about them.

This chart shows that people arrested in connection with riots and disorder were more likely to come from very deprived areas which had higher rates of unemployment, and the majority of people charged were local to the disturbances.

Read more here.

***

2. ChatGPT uses a lot of water and energy

The datacentres powering the new wave of generative AI tools like ChatGPT require a lot of energy and water, and this excellent graphic from the Washington Post shows exactly how much ($):

While the usage depends on local conditions – like where water is scarcer or electricity is cheaper – it’s great to see the environmental impact put into the context of a human-level, everyday task such as generating an email, and then scaled up.

***

3. What are Americans searching for this election period?

Waves of Interest is a collaboration between Google Trends and Truth & Beauty, a data viz and design company.

The data tracks state-based search interest in specific election topics, such as tax or cost of living, and compares the change between election years. The beautiful maps are a notable part of this project:

These contour maps are wonderful, especially when animated smoothly between time points. Read and see more here.

***

4. Young women are starting to leave men behind

In John Burn-Murdoch’s latest column for the Financial Times ($), he highlights an interesting trend in developed countries – while girls and young women have been pulling ahead of boys and young men in education for several decades, there’s now also a gap opening up between women and men on employment:

The trend has some clear positive and negative implications, as Burn-Murdoch says: “If this were simply a case of women making strides, it would be something to celebrate – and that side of the story certainly is – but a substantial minority of young men are actively moving backwards, with growing numbers increasingly disengaged from society.”

Spotlight on … the climate emergency 🚨

Off the Charts

First, the runner-up for this week’s Off the Charts was this FAT BEAR WEEK graphic from Axios (everyone loves fat bear week).

The winner, however, was this intricate chart about Taylor Swift’s most recent album The Tortured Poet’s Department.

It did take me a while to wrap my head around this chart, which can be best described as a visual analysis of the album, song-by-song. Coloured bars show lyric categories, white lines measure “singability”.

I’m not quite sure if I personally can get much out of it, having never listened to the album, but I appreciate the ingenuity on display here, and I really like some of the features like the lyrics scrolling around the outside the record.

Sign up

If you would like to receive The Crunch to your email inbox every fortnight, sign up here.

Sign up to read this article
Read news from 100’s of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.