Why does culture feel so monotonous? A walking tour reveals the math
Sign up now: Get ST's newsletters delivered to your inbox
Chain storefronts on a street in London. Data scientist Lauren Leek says predictive algorithms are flattening the everyday world, from pub menus to pop lyrics.
PHOTO: JOANNA YEE/NYTIMES
Scott Reyburn
- Lauren Leek says predictive algorithms are making culture look and feel the same.
- Her research on 640 London shopping streets and 4,000 Hot 100 songs shows chains and streaming push safe, repeated choices.
- Leek and Cory Doctorow warn that AI and platform data favour the average person and bigger profits. This can erase local variety and turn cities into “homogeneous no-places.”
AI generated
LONDON – On a recent afternoon, Lauren Leek stood on a typical London shopping street, in front of a Victorian-era pub and a predictable array of corporate-owned food, drink and retail outlets: Pret a Manger, Starbucks, Joe & the Juice, Subway.
“What we see in front of us used to be culturally diverse. Now it is culturally generic,” said Leek, a Dutch Australian data scientist who is an assistant professor at Leiden University in the Netherlands. “Most of what we see is the big chains we see everywhere,” she said. “It creates a sense of placelessness. We could be anywhere.”
Leek recently posted a Substack article, with the headline “Temperature Zero for Culture: Why Everything Is Starting to Look the Same,” that analysed the “flattening” effect that predictive algorithms have had on music, movies, hospitality and almost every aspect of today’s consumer culture. The post was shared and discussed by widely read technology commentators such as Cory Doctorow and John Naughton.
Taking 640 shopping streets in London as a test case, Leek used advanced data-analysis tools to examine how predictive algorithms have helped create what she calls a homogeneous “monoculture” in which, she writes in her Substack post, “we lose the ability to tell the difference between what people wanted and what the system made easy to want.”
Leek’s insight into everyday consumer behavior is, in some ways, reminiscent of Roland Barthes’ classic 1957 collection of essays, “Mythologies.” Back then, Barthes used the recently developed discipline of semiotics, the study of signs and their meanings, to turn things as disparate and as seemingly banal as soap powder, wrestling, margarine and automobiles into subjects of serious cultural consideration.
But unlike Barthes, who was steeped in the French philosophical tradition, Leek is a London School of Economics-educated data cruncher. She uses her own algorithms, digital mapping and text analysis to show how large language models are making consumer culture, wherever it happens to be, feel the same.
A recent excursion with Leek in south-west London to experience her analysis in action began in the Prince of Wales pub. The bar is owned by one of Britain’s biggest brewers and hospitality retailers, which runs 2,600 sites across the country. This British company is in turn owned by a Hong Kong-listed property development and investment conglomerate.
It was just after noon, and the old-fashioned wall decor and wooden furniture lent the pub a traditional feel. Multiple TV screens were showing the buildup to a Premier League soccer match. Generic mood music played in the background.
“Often these pub chains have playlists that artificial intelligence automatically adjusts according to the sound level,” Leek said. “If the pub isn’t very full, it will play chilled background music,” she explained. “When it gets louder, the music gets louder.”
Data scientist Lauren Leek used advanced data-analysis tools to examine how predictive algorithms have helped create what she calls a homogeneous “monoculture”.
PHOTO: JOANNA YEE/NYTIMES
Pop music is an area in which Leek’s research has revealed how sharply culture has “flattened” in recent years. Her Substack post includes a graph that tracks the verbal richness of lyrics from over 4,000 songs in Billboard magazine’s Hot 100 hits from 1950 to 2015. Since the early 1960s, Leek found, the range of distinct words has contracted more than a quarter, while the number of repeated lines has increased nearly a third.
“Streaming has created very concrete incentives – pay starts after 30 seconds,” Leek said. “A repeated hook in a song is the rational response.”
Leek studied the pub’s menu. It included staples such as chicken wings, burgers, and fish and chips, all of which could be ordered via a QR code. “Pub chains get the data from people who have previously ordered. They typically aggregate ordering data across their sites to optimise their menus. Everything has been analysed,” Leek said. “Local differences have been taken away. Quirkiness is taken out.”
Leek’s tour moved on to an upmarket cafe and bakery nearby. It is one of a chain of more than 150 in London majority-owned by an American private equity company that manages more than $200 billion of assets. A cappuccino costs £4.40.
As a steady stream of customers ordered coffees and pastries, Leek explained that chain stores use specialized site-selection algorithms to choose the optimum locations for their branches.
“They use public, open socio-economic data, like small-area census statistics. They have their own data, which neighborhood restaurants don’t have access to. And they have their own data analysis teams,” Leek said.
Patrons wait in line at Gail's, an upmarket cafe and bakery chain with more than 150 locations.
PHOTO: JOANNA YEE/NYTIMES
For Leek, the key to the flattening of consumer culture lies in the mathematics of algorithms and their methodology of next-token prediction. Language models like ChatGPT write one small chunk of text, called a “token,” at a time and predict and select the most likely next token at each step. The models have a “temperature” setting, a set of parameters that control how adventurous that pick is. At “temperature zero,” the model always chooses the most likely next token and gives the same answer every time.
“It’s predictable and safe,” Leek said. “When the model is uncertain, it still has to make a prediction, so it always falls back on the most expected answer.” That means there is no randomness, no surprises. Leek argues that algorithms are pushing culture toward a blandly predictable, mathematically determined temperature zero.
“The algorithm is making predictions for the average person, but the average person doesn’t exist,” she said. “Each person has some niche preferences, and the niches differ. A system that optimises for the average of everyone’s tastes produces something that nobody particularly dislikes but nobody particularly loves,” she added. “It’s built for a person who doesn’t exist.”
Leek is not the only observer warning that algorithmic optimisation is degrading the texture of daily life. Among those championing her work is Doctorow, a regular New York Times contributor, who has popularised the concept of “enshittification.” In his 2025 book “Enshittification: Why Everything Suddenly Got Worse and What to Do About It,” he defines this as the process whereby the performance of advanced digital platforms and devices gets worse while extracting more money from their users.
“A.I.’s abilities are pathologically overhyped by A.I. vendors and overestimated by AI users,” Doctorow said in an email, “but there is one thing A.I. is undeniably good at: automatically performing massive multivariate analysis to figure out who will pay more for a good or service, and who will accept less for performing those services.”
Nowhere is this more tangible, Doctorow added, than in the transformation of city streets into “homogeneous no-places.”
But Leek offered a slightly different diagnosis for what is happening to consumer culture. She prefers to call it “standardisation.”
“The individual coffee or restaurant often isn’t worse. The flat white may well be better than it was 20 years ago. What’s lost is variety,” Leek said. “Enshittification means things get worse. Standardisation means things get more alike,” she added.
“And with AI,” she added as she headed to the nearest Underground station, “that happens almost instantly, everywhere at once.” NYTIMES
This article originally appeared in The New York Times.
