There was a time when discovering culture required a certain amount of effort. In recent times, a new band might come from a friend’s recommendation, a record shop, a radio programme, or a magazine review.
A restaurant could be found while walking through an unfamiliar neighbourhood. A novel might come from a bookseller, library display or a conversation with another reader. Today, discovery often begins before a person decides to search.
Open TikTok, Instagram, YouTube or Spotify and a stream of recommendations is already waiting. Restaurants appear between comedy clips. Songs arrive in automatically generated playlists. Books become familiar after repeated appearances in short videos. Fashion styles move from obscure online communities into shopping baskets within weeks.
This is algorithmic culture, a system in which software increasingly helps determine what people encounter, repeat, discuss and eventually buy.
The change is not necessarily a story of manipulation replacing free choice. Recommendation systems can introduce audiences to musicians, writers, filmmakers and communities that traditional cultural institutions might never have promoted.
But they have also changed the balance of cultural power.
The important question is no longer simply who creates culture. It is increasingly who decides what gets shown.
Discovery no longer always begins with a search
Recommendation systems solve a genuine problem. There is more music, television, video, writing, food content and entertainment available than any individual could realistically explore. Digital platforms therefore try to predict what a user might want next.
The prediction is based on behaviour. What did the user watch? What did they skip? What did they replay? Which creators do they follow? How long did they remain on a video? What have people with similar habits enjoyed?
The result can feel remarkably convenient. A listener no longer needs to understand a music genre well enough to search for it. A teenager can encounter an independent musician from another country without a radio station or record label first deciding that the artist deserves international attention.
That ability to cross traditional cultural boundaries is one of the strongest arguments in favour of recommendation systems.
Research published in Scientific Reports in 2025, based on listening patterns from about 50,000 Deezer users, found that algorithmic recommendations could introduce more novelty than users reached through some forms of self-directed listening. The picture, however, was complicated. Although algorithms exposed listeners to new artists, those recommendations could remain relatively close in style to what the listener already preferred.
That finding captures one of the central contradictions of algorithmic culture.
A recommendation can be new without being truly unexpected.
The internet can broaden taste while narrowing surprise
The familiar criticism of recommendation systems is the “filter bubble”, the idea that people are continuously shown more of what they already like until their cultural world becomes smaller.
The evidence is not quite that simple.
The Deezer study found that algorithms can increase novelty in the short term. Someone may encounter artists they have never heard before. Yet over longer periods, those artists may still occupy similar musical territory.
A person who listens to alternative folk music might be introduced to dozens of previously unknown folk musicians. That is real discovery. But they may still rarely be directed toward jazz, classical music, Ghanaian highlife or Indonesian pop.
The researchers described a pattern in which recommendations could break users out of one kind of bubble while reinforcing its broader boundaries.
A separate 2026 study of Italian music listeners found another problem: many people used recommendation systems routinely while having little understanding of how those systems actually shaped what they heard. Researchers described a sense of distance between listeners and the mechanisms selecting music for them.
The issue, then, is not simply whether recommendations are diverse.
It is whether people understand when a recommendation is influencing them.
BookTok shows what happens when recommendation becomes culture
Few industries demonstrate the power of social discovery as clearly as publishing.
BookTok began as a loose community of TikTok users discussing books. It has since become a significant force in publishing, capable of reviving older novels, promoting debut authors and influencing which genres publishers pursue.
According to an analysis by NielsenIQ BookData and Media Control cited by TikTok, more than 50 million books associated with BookTok recommendations were sold across major European markets in 2025, generating about €800 million in revenue. More than a third of people aged 16 to 39 in the surveyed markets were said to discover books through the community.
The phenomenon has become formal enough that BookTok bestseller charts now combine social engagement with actual sales data.
This is culturally significant because the path from recommendation to commercial success has shortened.
Readers once depended heavily on newspaper reviews, literary prizes, booksellers and publishers to identify important books. Those institutions still matter, but a highly emotional 30-second video can now create demand on a scale once reserved for major advertising campaigns.
That can democratise publishing.
Readers can elevate genres previously dismissed by traditional literary culture. Romance, fantasy and hybrid categories such as “romantasy” have benefited from communities of readers speaking directly to one another.
But success also creates imitation.
Publishers can see which themes, covers and genres perform online. Once a particular aesthetic or storytelling pattern becomes commercially successful, similar books quickly follow.
The recommendation system does not merely respond to culture. It begins influencing what gets produced.
Restaurants are now designed for the feed as well as the diner
The same pattern is visible in food.
A restaurant once depended heavily on location, reviews, word of mouth and repeat customers. Now a single video can produce hundreds of customers almost immediately.
In London, restaurants and food businesses have experienced enormous queues after going viral on TikTok and Instagram. One Caesar-wrap business told The Guardian that it had expected to sell around 50 wraps a day but quickly reached about 400 after documenting its launch online. Another viral food business reported customers camping overnight for a jacket potato.
For struggling hospitality businesses, this kind of attention can be transformative.
But owners have also described the pressure that follows.
Customers often arrive wanting exactly the item they saw online. Seasonal menus can become difficult to explain. Small businesses can be overwhelmed by demand they were never designed to handle.
One café owner told The Guardian that social-media virality can remove cultural context and narrow curiosity, as customers arrive expecting food to resemble a particular image or short video.
This raises a larger question about algorithmic culture.
Are people discovering restaurants because those restaurants suit their tastes, or are their tastes being formed by repeated exposure to the same highly shareable dishes?
The distinction is increasingly difficult to make.
Virality creates its own evidence of importance
Algorithms do not simply identify popular culture. They can help produce popularity.
A video begins receiving attention. More people watch it. The platform interprets that engagement as evidence that the content deserves wider distribution. More users see it, creating additional engagement.
The resulting popularity then becomes culturally meaningful.
A long queue outside a restaurant can attract more customers because the queue itself suggests that something important is happening. A song repeatedly heard in short videos becomes familiar before the listener consciously decides whether they like it.
Popularity becomes both the result and the cause of further visibility.
This process can be enormously beneficial to creators who once lacked access to traditional cultural industries.
A filmmaker does not necessarily need a studio. A musician does not always need radio. A writer can develop an audience before receiving mainstream critical attention.
YouTube argues that this has not destroyed mainstream culture but changed how it forms.
Research commissioned by YouTube and NRG in 2026 concluded that culture is more fragmented and personalised, but shared cultural moments still emerge. Among surveyed users aged 14 to 29, 63% said sharing interests with others made them feel part of something larger, while 69% of respondents reported watching content from communities they considered niche.
That research comes from a platform with a direct interest in creator-led culture, but the broader point is difficult to dismiss.
The mainstream has not vanished.
It increasingly begins in niches.
The algorithm can also reward sameness
The difficulty comes when creators begin making culture for the recommendation system rather than for the audience alone.
Food influencers learn which camera angles perform best. Musicians understand that the opening seconds of a song may matter more because they need to catch attention quickly. Publishers study which covers remain visible on a phone screen.
Creators have always adapted to distribution.
Radio encouraged particular song lengths. Television shaped episode structures. Newspapers affected how writers structured stories.
The algorithm is another distributor, but with an important difference.
Its rules are often unclear.
Creators know that certain things appear to work, but they do not necessarily know why.
This uncertainty encourages imitation.
If short, dramatic videos succeed, more creators make short, dramatic videos. If exaggerated food reactions receive more views, more people adopt them. If a particular interior design becomes popular, thousands of versions can suddenly appear across feeds.
The result is an odd mixture of enormous choice and visual similarity.
There may be more culture available than ever before, yet large parts of the internet can begin to look remarkably alike.
AI is adding another layer to cultural discovery
The growth of generative artificial intelligence makes the question more complicated.
Until recently, recommendation systems mostly decided which human-created material people saw.
They are increasingly being joined by systems capable of helping create the material itself.
AI music company Suno launched new models this week through partnerships with Warner Music Group and BMG. The system allows users to generate new music inspired by licensed work from participating artists. Spotify is also developing AI-powered tools that could allow users to create and share remixes of licensed tracks.
This changes the relationship between creation and recommendation.
Imagine a platform that knows exactly what kind of music a listener enjoys and can also generate an endless supply of music designed around those preferences.
The question moves beyond which song should be recommended next.
It becomes whether the next song needs to exist before the listener asks for it.
That possibility could create extraordinary new forms of creativity. It could also produce culture increasingly tailored to individual preference, reducing the role of surprise, disagreement and shared experience.
The case for algorithms remains strong
It would be easy to romanticise the culture that existed before recommendation feeds.
Traditional cultural discovery was never neutral.
Radio executives chose playlists. Publishers decided which manuscripts reached bookstores. Newspaper critics could build or destroy reputations. Film studios controlled distribution. Record labels determined which musicians received promotion.
Those systems excluded enormous numbers of creators.
Algorithms have weakened some of those barriers.
An unknown artist can reach an international audience. A regional cuisine can suddenly become visible far beyond its home. Readers can build communities around genres critics have overlooked.
Instagram chief Adam Mosseri has also argued that users may not actually prefer a world without algorithmic ranking. Responding to Australia’s proposed reforms, he said chronological feeds could overwhelm users with less relevant material, including large volumes of posts from brands they chose to follow.
That argument deserves consideration.
The problem may not be recommendation itself.
It may be recommendation without sufficient transparency or choice.
Governments are beginning to ask who should control the feed
That debate has moved from cultural criticism into public policy.
Australia proposed legislation this week that would require social-media platforms to give users a choice between personalised algorithmic recommendations and feeds limited to friends and creators they actively follow.
The European Union has already moved in a similar direction.
Under the Digital Services Act, very large platforms must provide users with an option for recommendations that are not based on personal profiling.
These policies suggest an emerging principle: recommendation systems may be useful, but users should know when they are operating and should have meaningful alternatives.
That could become one of the defining cultural debates of the next decade.
Discovery has not disappeared, but it has changed direction
The internet has not eliminated cultural discovery.
In some ways, it has created more of it.
People can encounter music from countries they have never visited, discover writers ignored by traditional publishing, learn about unfamiliar cuisines and join communities built around highly specific interests.
But discovery increasingly arrives without being requested.
That is the defining feature of algorithmic culture.
The feed observes, predicts and delivers.
The danger is not that people have stopped making choices. They continue to skip songs, close videos, reject recommendations and seek things out independently.
The change is that those choices are now made inside an environment that has already decided what options deserve to appear.
Culture was never discovered in a vacuum. Critics, broadcasters, publishers, friends and institutions have always influenced taste.
What is new is the scale, speed and personalisation of the system.
A human editor once selected one front page for millions of readers. An algorithm can create millions of different front pages, one for each person.
That can make culture richer and more accessible.
It can also make the forces shaping taste almost invisible.
The future of algorithmic culture may therefore depend less on eliminating recommendation systems than on restoring some friction to discovery.
That could mean deliberately searching for unfamiliar music, walking into a bookshop without checking online rankings, entering a restaurant without watching a review first, or choosing a film because another person, rather than a platform, suggested it.
The internet has become extraordinarily good at predicting what people might like.
The more important cultural question is whether people still leave enough room to encounter what they did not know they were looking for.
Don’t forget to follow us on Facebook | Instagram | Twitter | LinkedIn to get the latest updates from Cape Town Tribune
