[Media] Art Curation in the Automated Age

On the curse of recommendation algorithms

Lisa vibing to vinyl

Ever notice the preoccupation with curation lately? Curation has been appearing as a selling point for online services in the past decade, mostly in the world of media (streaming, music) but delivery services have also begun to embrace the trend.

It speaks volumes that a recommendation being delivered by an “actual person” is a way for services to stand out from their contenders. At this point in the web’s evolution, the role of the ‘algorithm’ has overwhelmed discourse about social media and entertainment, with the majority of online content being siphoned by unreliable automatons before being served to the public. Despite the resurgence of curation as a feature, the most popular platforms remain indebted to automated recommendations, most notably YouTube, Netflix, Spotify, and almost any scroll-based social media platform. 

As part of the growing levels of tech illiteracy that accompanies a world of touch-based, user-friendly interfaces, few people know what the ‘algorithm’ actually does. The most complex and widely used platforms don’t rely on a single algorithm to serve recommendations; in truth, they’ve evolved to predict behaviors rather than work solely off of an equation. For example, Netflix tracks surprisingly minute user behaviors (including when and where you pause the video, how many searches it takes to find something you like, when you stop watching, etc.) and inputs the data into a machine learning model that constructs rows of recommendations based on prior user behavior. 

Source: Netflix Tech Blog

One on hand, this “content-based, behavioral” approach to recommending media sounds intuitive. Its complexity means that there’s less room for a single algorithm’s shortcomings to corrupt the results. The diversification allows for recommendations that, though flawed, feel varied enough as not to feel monotonous. That element of it, knowing when to vary content enough to keep a consumer hooked, is a boon for social media platforms like TikTok. Not because TikTok earnestly wants to recommend media to interested parties out of the goodness of their little hearts, but because TikTok makes billions of dollars in revenue from serving ads to hooked users. 

Those cruel intentions that lurk beneath large tech companies’ decision to utilize user data may be why curation has gained a niche yet sizable following. Devotees to curation-based services value the idea of “earnest” recommendations. These are product or media suggestions given under the veil of an expert’s endorsement, not artificially chosen slabs of data. In other words, curation is the organic alternative to algorithms. 

Part of this has to do with what curation actually is. Typically used in reference to museum curators, curation implies a certain contextual, emotional, and intellectual interpretation of art that computers have yet to fully replicate.

It’s my theory that the demand for curation services is partially a desire for less digitally mediated experiences overall.

In “Why Record Stores Mattered” by Amanda Petrusich, she describes a “collective nostalgia” for brick-and-mortar record stores, a condition “...in which we mourn the freshly antiquated for reasons that are unclear but still enormously potent”. Her description perfectly captures the loss felt by those living in a post-COVID society. The over-screenification of everyday life feels so typical that many are left grieving for a pre-pandemic reality for indescribable reasons. The result is a throng of consumers left grappling for anything that invokes unmediated sensations, namely media that’s been curated by “real humans”. 

---

Of course, these conclusions are affected by my own biases.

I don’t pay for Spotify despite using it often because its model of music presentation doesn’t resonate with me (I prefer the relatively ancient RYM). I’m often repulsed by algorithmically dictated media engagement; call it hyper-individualism or tech-fatigue, but it’s not my preferred avenue of discovery.

I find the recommendations given by Netflix underwhelming since I react cynically to surface-level associations between films. Though shallow associations can easily be made between similar films (Scandinavian crime dramas! Pride Month thrillers! etc. etc.), I’ve found that my most lasting experiences with film evade simple classification. I gravitate towards the ultra-specific, oft-cheeky sea of Letterboxd lists in this regard (the cottage-industry of Letterboxd lists deserves its own investigation, so I won’t dive too deep into it here).

I’ll admit—these decisions are mostly based on intuition and personal preference. What’s really lost when algorithms overtake human-curated media, anyway? 

For one, the integrity of the recommendations themselves. From the Journal of Public Policy and Marketing:

…stemming from a belief that algorithms hold greater domain expertise, consumers surrender to algorithm-generated recommendations even when the recommendations are inferior. [...] consumers frequently depend too much on algorithm-generated recommendations, posing potential harms to their own well-being and leading them to play a role in propagating systemic biases that can influence other users.

In short: consumers generally tolerate things that aren’t optimized for them, all because a recommendation system insists its best for them. I chalk this up to a combination of consumer apathy and what they study calls “algorithm overdependence”. Given the sedating effects of passive media like television and social media, this isn’t surprising. The more widespread detriment is that this behavior is watering down recommendation results for other users, thus making suggestions inferior for the platform overall.

That issue raises the question of if creative media lends itself to algorithmic optimization in the first place. Obviously Netflix knows to recommend movies based on surface-level aspects that can be quantified, like release year and genre, but what about more ephemeral traits? What about the subcultural impact, the aesthetic sensibility, or the styles of acting?

Even platforms with extremely successful recommendations, like Spotify, can’t generalize their approach to every consumer. Some people simply create connections between media in different ways than the majority of the population. I was struck recently by Jeremy D. Larson’s essay about the uncanniness of listening to music through Spotify. His essay codified and categorized a phenomenon I mull over constantly: the distinction between Passive and Intentional media consumption. I want to focus on his definition of a “Passive listener” of music:

Most of the world falls into the Passive category, absorbing music like inhaling oxygen: without much thought at all. For them, there is either music playing, or maybe it’s not music playing, who can be sure? There is perhaps little to no interrogation into why any sound is floating down from the speakers at the grocery store; it simply exists at the same megahertz as the shopping cart and the fluorescent lights and the cereal selection.

Spotify’s recommendations are trained on a large part of the population which necessarily averages out to reflect the ‘normal’ consumer, but this excludes consumers with abnormal systems of media association or connection. The listeners for whom “music is life” aren’t properly served by recommendation algorithms by their nature because when a platform is built to generate money, there’s no benefit in catering to the attentive (i.e. nerdish) minority. 

---

It’s no coincidence that curated media platforms market themselves towards the “Intentional” class. A platform like this that I’m most familiar with as a film lover is the Criterion Channel. 

Criterion Channel’s model benefits from the personable, hand-picked approach it takes to film recommendations. To those not familiar with Criterion’s streaming model (cancel your Netflix subscription and click here), every month a “collection” based around a theme, actor, or fim movement is dropped. These collections are accompanied by additional bonus content, usually from the original Criterion release discs themselves.

If this sounds drastically different from a service like Netflix, good. Criterion differentiates its model and derives its success from advertising itself as intended for a different “kind” of consumer. I understand that the complex relationships between films aren’t accurately represented by Netflix’s algorithmic model, so I opt for experts to recommend me content instead. 

Criterion Channel: Beyond Blaxploitation Introduction

Polished, erudite, and ‘certified’ film scholars are often the ones in the additional content describing and analyzing new collections, a far cry from the depersonalized approach that Netflix takes. There are introductions for each collection, connecting the films to each other through verbal, on-screen discussions between real people.

Keen marketers are aware that consumers who think of themselves as part of the Intentional class want a service that affirms their identity. It’s no stretch to that Criterion leans into its reputation because it appeals to and reinforces the consumer sense of self, and the result is what’s been called the “Cult of Criterion”. More cult members, more people buying $40 Blu-Rays.

---

Curation is becoming a novelty. Its rarity differentiates it from the sea of seemingly endless content that an algorithm can output, so marketing strategists have seized the opportunity to bolster their brand’s identity, whether it’s in service of books or dating apps. 

That’s not to say all curation-based platforms are cynical. Similarly, I know first hand how intensely personal algorithms can feel, so I don’t want to paint them with a broad brush either. I’m ultimately pro-technological advancement, and I forsee a stealthy potential for automation to make a splash in the curatorial market. But in the attention economy of today, it’s only smart to be mindful and critical about just why you feel so drawn in by a media platform, and whether that platform has your best interests in mind.

---

📼 A/N: Largely inspired by my favorite writers, I’ve revamped my Substack. The hiatus was due to a combination of schoolwork, graduation, and a lack of time management skills.

This newsletter is ultimately my attempt to engage meaningfully and emotionally with media analysis/criticism, which, after months of not thinking much of it, seems to be the only thing bearing on my mind. So, I decided to get writing again and serve up some more thoughts and opinons!

And I need some way to put this new media studies degree to use.