The Importance of Personal Taste in the Age of Algorithms

As the consumption of culture becomes increasingly controlled by algorithmic recommendations, the meaning of taste has fundamentally changed.

Who is Fran Lebowitz in the year 2021? I ask the question not in some practical, biographical way (She is a humorist, past coworker of Andy Warhol, and as of January, the star of Netflix’s new docuseries Pretend It’s a City that she made with Martin Scorcese), but conceptually. Here is a woman who is, above all else, revered for having strong opinions. She published her last book 30 years ago. She openly reviles working, writing. And yet, when she mouths off about anything, many of us listen. “The only thing I remember about Phantom of the Opera is it was unbelievably horrible,” she spats to an overjoyed Scorsese before waxing salty poetic on E-Cigarettes, the internet, and so much more on the new show. Throughout it all, she never fails to express her opinion strongly. No disclaimers. No wavering considerations. The woman has a taste of her own. The universal appreciation of Lebowitz speaks to our growing desire to hear strong-minded convictions; in an age of information overload and cultural delirium, a clear-eyed expression of taste is a rare and valued thing.

Taste is an ambiguous concept. More than a mere list of likes and dislikes, it is overarching, a developed intuition that guides one's preferences across mediums, forms and contexts. John Waters, director of works of camp cannon such as Hairspray and Pink Flamingos, said the following in 2011: “Everyone is a curator of their own life: what they have around them, what they read, what they watch. Everyone has taste and it’s how you define yourself against the world.” For Waters, taste is at once a curation, a list of favorite objects, works and ideas, but it also amounts to how you define yourself in relation to others. It becomes a guiding self-understanding, a personalized ideology that helps you make sense of an overwhelming world.

Today though, curation is largely fed to us via various algorithms. Spotify observes our listening patterns and then makes suggestions based upon it in the form of weekly and daily personalized mixes. Netflix’s recommendations are derived from our past viewing habits, spotlighting whatever is similar to what we just binged. Tik Tok’s “For You Page” is a direct response to what we like, watch, share and comment on, feeding us more and more of what the algorithm expects us to enjoy. What then, is taste in this context? Is your personal taste developed over time through your relationship with algorithmic recommendations? Is an algorithm even an accurate representation of your overall taste, its constant surveillance an objective metric of aesthetic preference? Or does the presence of an impersonal code somehow distort the very essence of taste? If so, can taste only be formed outside the cold calculus of platforms in smoky basements with passionate friends haggling over top ten lists and hot takes? 

It does feel that there is a missing link in this new curatorial, algorithmic ecosystem. Algorithmic recommendations have essentially replaced the role of critics who once were the guardians of what’s defined as new and interesting. To zero in on music, there is a tangible loss to the partial death of music media. In my early days of music discovery, I scoured music blogs big and small in hopes of finding work that moved me. Reading Pitchfork, Pigeons and Planes, the now defunct Potholes in My Blog, I read reviews and recommendations, listened, and then incorporated the work that resonated with me into my personal music diet. But this process (reading, listening, thinking, then consuming) has far more steps, and requires much more discernment, than receiving recommendations via Spotify playlists and reacting to what’s presented. Liz Pelly, today’s most important critic of streaming culture, writes, “Streaming services like Spotify create passive environments where listeners stream what they like, and more of what they like, and more of what they like—ad nauseam. Meanwhile, we know this much: listening patterns are studied by playlist programmers, who privilege songs with high completion rates and delete those with high skip rates.” Over time, algorithm’s know us better and better, giving us more and more of what we like, and thus limit the potential for our preferences to change. 

When talking about culture with friends, we often talk in the parlance of story, of context. “Have you heard this new SZA record? She made it with the Neptunes. Look at the dance people are doing with it on Tik Tok.” “Have you seen Promising Young Woman? The end was crazy! I don’t know how I feel about it, but . . . Carrey Mulligan! And that Charli XCX remix!!!” That stream of  factoids and stories signal a sincere engagement with the work, a nuanced reception rooted in more than: “this sounds like X so you’ll like it.” But as the algorithmic begins to take precedence over the journalistic, I worry that those roundabout, messy ways of sharing and thinking about culture will go away and the language of algorithms will soon win. In this world, songs are components for mood-based playlists. TV shows and films are ways to pass the time when your mind can't turn off to sleep. Books, that thing you have in your zoom background. And if you read it, it’s to join the rapidly devolving social media discourse. I realize I am presenting this as an apocalyptic hypothetical, but really: this world has largely arrived. During the pandemic, when there’s somehow far too much and too little time, works of culture have lost much of their spiritual, resonant quality and taken on a mere time-filling use.

Holly Herndon, artist and frequent critic of the platform economy, summed up the problem of algorithm recommendation and pay-per-play streaming models brilliantly in an interview with Pelly:

“For me, the most important music which has shaped me as a person and as a musician myself is not necessarily music that I listen to on repeat. It’s music that contains an idea that I really just needed access to . . . For example, Penderecki’s Threnody to the Victims of Hiroshima is not easy listening. It is not something that I am going to put on repeat. Whereas something like Pure Moods, 90s chill music, is something I can have on in the background.” 

The distinction highlighted by Herndon is core to the problem of algorithmic recommendation. When frequency of play and a surface metric of “likes'' become the only useful data set for an algorithm, what are we to do with the bewildering, the different, the challenging, whose very nature means they are less likely to be replayed and “enjoyed” in a traditionally fun sense? The problem is that grappling with such work is key to developing one’s own taste; the intrigue, or revulsion, or curious amusement inspired by something novel tells us far more about our own preferences than encountering something similar to what we already know. Algorithm’s can't see the nuances of our relationship to challenging pieces, and thus don’t recommend them. We are, unwittingly, boxed into a stasis of our own making.

I think we love Lebowitz because Lebowitz feels strongly. Her taste was developed over time through engagement and conversation, not the feeding tube of platform-based algorithms. When we hear her, we hear the words of someone who grappled. This is not to say that the past days of elite media were so undeniably amazing, but in her conviction, we get a glimmer of a different world of culture, of criticism, of taste. So yes, Spotify: I do love contemporary pop with cinematic textures and beats inspired by house and techno. Yes, if the singer sounds adrogynous, that’s a plus for me. But please, free from this niche dungeon of my own making! 

Actually, I think I’ll find the keys myself, and go in active search.