Book Review — “Filterworld: How Algorithms Flattened Culture”

Yves here. This article describes how algorithms amplify certain cultural artifacts, like ditties and visual images, narrowing the range of expressing and lowering the odds of truly new works breaking out into the mass market. While this is clearly a troubling development, it is not as if this is the first time commercial interests have set out of influence broader tastes. We had power law returns in entertainment before algos played any role. Recall payola back in the day when radio air time played a big role in record sales. And as an example of pervasive blandness, Muzak also way predated algorithms.

But in other fields, technology changes have steepened the rewards curve. For instance, in publishers used to buy and make good money on so-called mid-list books, which they marketed to a fair but not intense degree and were profitable, often long-tail sellers. And some would break out and do very well. I am told the mid-list has collapsed. Publishers for the most part have stopped paying moderate advances that used to be the bread and butter for many authors. And please don’t buy the hype. Self-publishing involves a lot of extra work normally done by the publisher, from marketing to proofreading to indexing, and seldom pays.

By Elizabeth Svoboda, a science writer based in San Jose, California. Her most recent book for children is “The Life Heroic.” Originally published at Undark

About a decade< ago, the anthropologist Natasha Dow Schüll — who’d spent years studying casino slot players — popularized a cycle of repeated engagement she called the “ludic loop.” Each time people pull a slot lever and watch the fruits align, they get a hit of rewarding brain chemicals that entices them to pull the lever again and again. Though alluring, the ludic loop is also fundamentally hollow: People emerge from it anesthetized and depleted, having gained nothing in the end.

The ludic loop concept predated the reign of “the algorithm” — the stretches of computer code that dictate which new content rolls onto our social media feeds, playlists, and streaming services. Yet Schülll’s theory anticipated how millions of users would engage with content streams tailored to their desires — passively, almost thoughtlessly, ever seeking the next dopamine hit.

BOOK REVIEW“Filterworld: How Algorithms Flattened Culture,” by Kyle Chayka (Doubleday, 304 pages).

In “Filterworld: How Algorithms Flattened Culture,” Kyle Chayka, a writer for The New Yorker, argues that the programs and apps serving up this addictive content have bled art and culture dry. By promoting creators who churn out attention-friendly “most liked” material, and lulling us into hoovering it up, algorithms highlight “the least ambiguous, least disruptive, and perhaps least meaningful pieces of culture,” Chayka writes. The result, he contends, is as empty as the ludic loop itself: content that “embraces nothingness, that blankets and soothes rather than challenges or surprises, as powerful artwork is meant to do.”

Scientists have long known that our brains are geared toward meme-making. We replicate and pass along concepts and ideas we like — which explains why tribes developed distinctive beadwork styles, or why Impressionist painters all adopted similar line-blurring brushstrokes. But algorithms have intensified and globalized this meme-making process by relentlessly promoting what’s most engaging, a trend Chayka traces to scientists’ early ideas about how to filter the deluge of online content. “We need technology to help us wade through all the information to find the items we really want and need,” MIT Media Lab researchers wrote in 1995, “and to rid us of the things we do not want to be bothered with.”

Yet that proposal has reaped unforeseen consequences. There’s a reason cafes from London to Reykjavik to Beijing all feature subway tile walls, reclaimed wood, and Edison-bulb lighting: After Instagram algorithms served many business owners the same most-liked content, “one cafe owner’s personal taste would drift toward what the rest of them liked too,” Chayka writes, “eventually coalescing into a net average.”

It’s this predictable coalescing, this regression to a bland common denominator, that Chayka finds corrosive. While rubber-stamped cafe interiors are largely harmless, other algorithm-driven content has a more insidious effect: It hypnotizes viewers and courts likes but makes scarcely any lasting impression. “When it’s over,” Chayka writes, “the experience immediately leaves your mind like the bubbles effervescing in seltzer.” For Filterworld’s deep-pocketed feed managers, the ideal user experience is the frictionless plane, the better to ease the way to more content consumption.

Though most people sense the seltzer-bubble vapidity in what their feeds serve up, what’s harder to grasp is how algorithms shape the act of creation itself. Chayka deftly explains the complex incentives that guide artists in the algorithmic era — and how some of these forces act on them without their awareness. Because “Insta-poets” get the most online engagement with brief, simple verse, some now write such verse almost exclusively, while visual artists veer toward inoffensive, bright-colored sketches. (All the while, they remain in the dark about how social networks dictate what gets widely distributed, a state Rutgers computer scientist Shagun Jhaver calls “algorithmic anxiety.”)

By promoting what accrues likes rather than more challenging fare, algorithms are devaluing content that has the power to disturb us and upend our assumptions, Chayka writes. “We miss out on culture that is truly progressive and uncomfortable.”

To feed the social media beast is to recognize this pruning at work. Selfies and family snaps can rack up high like counts and views, but disturbing photos, like a swastika scratched on a food court wall, may get much less traction. (People don’t want to like them, after all.) And blatant misinformation, like claims that ivermectin cures Covid-19, can dominate feeds if it reconfirms viewers’ priors (say, that out-of-the-box remedies trump established ones).

The book’s final chapters focus on how to preserve art and culture’s power to challenge us in the algorithmic era. In the face of our relentless numbness to so much — racism, authoritarianism, genocide — Chayka’s appeal for constructive friction feels crucial, if a bit quixotic. Some cultural creations, in the tradition of Elie Wiesel’s “Night,” ought to be warnings: not all-caps screeds that fuel angry engagement, but bone-deep auguries that linger past the next ping or refresh.

Yet Chayka’s proposals for reviving this dynamic are incomplete. He recommends seeking out human curators beyond the internet, in places like art museums and movie theaters, who highlight boundary-breaking art and deepen our appreciation of it. These curators “make sure that what merits exposure gets it,” he writes, “and introduce us to what’s new, challenging us enough that we avoid homogeneity.”

While this advice is sound enough in itself, it glosses over the contributions of shrewd online curators who operate within algorithmic confines. Instagrammer Mimi the Music Blogger, for instance, has her ear to the ground in the rap community, allowing her to feature cutting-edge artists her followers may never have encountered. Since subversion sometimes does come nested in feeds, the way forward may involve finding and supporting creators that embrace it, rather than opting out of Filterworld.

At the same time, it’s hard not to wonder whether the algorithmic dominance Chayka so vividly describes is starting to ebb. There’s plenty of evidence that social media fatigue has mounted and that fewer posters now feed the kinds of content streams that draw eyeballs. Leaving Twitter’s (now X’s) chaos and Instagram’s airbrushed scenes behind, many of us are gravitating toward old-school online forums or semi-private spaces like Slack, where recommendations and cultural debates can thrive unfiltered.

Ultimately, though, “Filterworld” is about something bigger than social media or tailored feeds. It’s about our tendency to passively consume what’s spoon-fed to us — and our obligation to push back on that tendency, to seek out art that splits us open like Franz Kafka’s frozen axe.

“I’m operating on the belief,” writer Tracy K. Smith once said, “that poetry can restore me to the large original self I haven’t yet learned to fully recognize.” There is more than one route to that large original self, and some of those routes may be digital. But Chayka’s call to break out of hypnotic content loops — to reject the empty absorption of the slot jockey pulling the lever — resonates no matter the medium.

Print Friendly, PDF & Email


  1. Carolinian

    Sounds like the book itself is a “meme” given the long history of cultural complaints from the “booboisie” through the “dream factory” to the “vast wasteland.” Pauline Kael wrote an essay called Trash, Art and the Movies and you could say her entire career theme was about by the numbers entertainment versus “art.”

    By this view algorithms are merely doing what Nielsen ratings and box office receipts used to do–provide media creators with popularity feedback that often resulted in “lowest common denominator” fare that would garner the most eyeballs. And the role of critics like Kael was to themselves be a filter to steer us to the good stuff.

    Hopefully NC is doing that for us now. The New Yorker–Kael’s onetime home–has been receding in importance for some time now.

    1. anahuna

      A minor point: From what meme did Ivermectin emerge to trouble the writer’s mind as a prime example of “misinformation”?

      1. BrooklinBridge

        I don’t think your point is minor. Algorithms may indeed sand paper out some elements of public beliefs and reactions to events that are disturbing but valuable to recognize truth including it’s darker sides, but the interpretation or mapping of such algos to their nefarious or benign results remains highly subjective and prone to the subtle and not so subtle prejudices of the mappers.

        As I understand it, NC and particularly IM Doc has stripped away what are essentially financially motivated objections to Ivermectin. What is left is that Ivermectin is indeed effective at reducing the effects and severity of Covid 19 if taking it is started early in the infection stage. I don’t remember if this early use also catches and reduces incidence of the disease outright, but regardless, low severity is almost as good as. The fact that Ivermectin is so cheap is the basis of the problem and likely also the reason that any related algos would, by-in-large, rub it out as a viable treatment.

        More to the point of your observation being important, however, in my mind at least, is that money rather than algos exerts the greatest influence in the direction social and cultural movements take. Another way of saying that is the algos, like politics, follow the money rather than the other way around.

  2. Vassilikos

    Not sure if Chayka mentions this in the book, but many of the irl curators he advises us to seek out are now tailoring their selections to be “instagrammable” or “viral” – see the phenomenon of pop-up “museums” that serve more as adult playgrounds than as places to appreciate art. I find it fascinating (and disturbing) that these algorithms don’t just change what we see on our feeds, but can alter the physical realm as businesspeople are incentivized to conform their spaces to what’s most likely to be photographed, shared, and circulated.

  3. scott s.

    Sounds elitist. “Culture” should be disruptive, ambiguous, challenging, that “disturb us and upend our assumptions” I don’t see why that is a requirement.

    1. Morincotto

      I don’t see why culture generally should be like that, but I definitely can see why some of it being like that is not a bad thing at all.

      And if that is elitist it doesn’t bother me much in this particular context.

    2. digi_owl

      Ever since the 60s it seems like cultural academia hopes to capture that particular kind of lighting and bottle it for big media, so that they can produce Beatles-mania on command.

      It won’t happen though, as any such attempt will be a manufactured good and thus lack “soul”.

  4. esop

    “Nonsocial cultures will decay”.
    Pyotr Alexeyevich Kropotkin was a Russian anarchist and geographer known as a proponent of anarchist communism.

    1. gcw919

      Kropotkin is also noted for observing, after an extended period in Siberia, that “nature red in tooth and claw” is only part of the equation. In fact, cooperation (“Mutual Aid”) is an essential factor in evolution, and while it may be widely recognized, it still seems to get short shrift as a “meme” in our current cultural atmosphere. As an example, our so-called ‘entertainment’ is replete with examples of people, groups, countries, etc, trying to destroy each other. As the saying goes, “if it bleeds, it leads.”

  5. AG

    thx for this insight.
    Of course from a European “state culture” POV this can easily be countered by ramping up state subsidy of the arts.
    e.g. publishers in Austria and I assume Switzerland too are profitting a lot from what in essence is state support.

    German state broadcasting is nothing but state sponsored visual entertainment.
    Same goes for motion picture funding.

    The fact that the quality of the output is often beyond discussion is again one of political decision-makers.
    France is a good example. They have a film industry solely due to state intervention based on the explicit goal of the government to build a French film industry.

    Ditto South Korea.

    In Germany the dedication that France show towards movies is true for theatre and orchestras which makes for a rather vibrant and unusal richness and culture on German theatre stages.
    And, most important in this, guarantees a source of income.

    Regardless of algos a state can decide to channel massive funding into arts just like with everything else.
    Defense industry works the same way.

    If there are no MoDs on this planet, there would be no market for arms sales.
    Whether you have an MoD is a normative decision of the state.

    There is no difference between book publishers and plants for tank production.
    And tanks get sold without algos too. To put it boldly.

  6. digi_owl

    Once more Zappa’s lament comes to mind, about how the recording industry was better off when the labels were ran by cigar chomping old men willing to do small production runs of random acts that walked in their door.

  7. Mark Gisleson

    I just got done listening to a new album that included tributes to Delia Derbyshire and Laurie Anderson, and a related video referenced Judith Butler. The artist had previously helped with a charity album for Ukraine. It’s a nice album, unless you’ve listened to Anderson or Derbyshire and then it’s disappointing. Selzer bubbles.

    Yesterday, I finished up sorting through a couple dozen electronic albums by a very talented but not well known duo. It was never boring even when they dived into minimalism and drone music. I’m still not sure if it’s really music or just sound effects cleverly strung together but it was always interesting.

    The artist I first referenced may well have used AI and algoes to help, at least it sounds like it. It’s easy for academic creatives to mix and match what exists to create something “new” and interesting, but it’s not new and quickly becomes uninteresting but because they’ve learned how to jiggle the algo handle on the toilet that is popular culture, they’re doing OK. Society rewards them, in this case with grants and reviews so effusive they’re actually offensive.

    Society generally does not reward true creativity unless accompanied by notoriety but the scene is shifting. Thanks to the internet, people you’ve never heard of are collaborating with other people no one’s ever heard of. I’ve listened to dozens of collaborations where the musicians had never met face to face. That’s music that goes to the other places where the wild things are. Etc. We’re not running out of good music but we do need better search tools to help find it.

  8. Lefty Godot

    Sounds like Chris van Tulleken should do a sequel to his ultra-processed food book about ultra-processed culture. Because that is basically what social media and search engines have turned into.

Comments are closed.