How to Train Your Algorithm
Our changing relationship with the math that rules our lives
Last week, I dipped my toe into sneaker drops.
I downloaded Nike’s SNKRS app, clicked “Notify” on a pair I liked, and asked my sneakerhead colleagues how I could optimize getting that drop. The first thing they told me was to engage with content on the app so the algorithm knows how much I care.
Algorithms form the basis of how marketers find and target their consumers on every platform that can be digitized. Especially for social platforms, the algorithm is part of its secret sauce, and marketers spend time learning what factors can help boost their content and get maximum attention.
Got your number
If you’ve spent time on TikTok or Instagram, you know the drill. The app feeds you content or ads based on what you’ve engaged with before. Whether it’s candle ads on Instagram, or cat videos on TikTok, many have become aware that the algorithm is responding to some element of their behavior, like engaging with certain content or clicking on certain ads.
Now people are learning a trick or two to adjust the content they receive when it gets calibrated away from their interests. Ad lady and Twitter joy @Aishaannhakim recently asked how to change her TikTok algorithm, and received all kinds of wisdom on how to engage with content differently to get different results.
Our relationship with algorithms is a funny one. Ten years ago, we were barely getting used to the idea that the internet would turn our behavioral data into a steady stream of targeted content for us (and dollars for the companies who got it right). But today we’ve settled into a funny relationship, one where we don’t know quite what the algorithm is up to, but we know it’s up to something. We complain when the algorithm doesn’t quite know us well enough, or when it knows us far too well. But we complain very little these days about the fact of the algorithm at all.
What are secrets among friends?
Algorithms are still a black box, trade secrets more closely guarded than KFC’s 11 herbs and spices. TikTok’s algorithm is one of its core differentiators. The ability to serve up content that keeps users engaged for hours has made it an incredibly valuable company. Recently, the New York Times published “How TikTok Reads Your Mind,” dissecting its addictive algorithm from an internal company document. To keep users watching and coming back, TikTok considers four main objectives: user value, long-term user value, creator value, and platform value. The document offers a rough equation for how videos are scored, in which a prediction driven by machine learning and actual user behavior are summed up for each of three bits of data: likes, comments and playtime, as well as an indication that the video has been played:
Plike X Vlike + Pcomment X Vcomment + Eplaytime X Vplaytime + Pplay X Vplay
Their algorithm is so effective that you get other headlines like this, by a woman who learns to embrace her bisexuality:
My feed has become curated so specifically to my tastes that it has alerted me to parts of myself I hadn’t fully embraced yet: It’s a medley of absurd animal videos, spooky happenings that are more than likely hoaxes… and queer femmes.
…Spending time on TikTok has opened my eyes to the glorious breadth of identity, presentation, and sexuality in the world. I don’t mean that I was unaware of it before, per se, but the narratives I see on TikTok are uniquely joyful.
The algorithms “know” much more about us than we do about them. Whether it’s life-affirming content or all-too-accurate ads, we’ve accepted that our lives are entwined with complex calculations beyond what most of us can understand.
In other algorithm news
To take a look under the hood of a recommendation engine, see Spotify’s Every Noise at Once, which shows the thousands of music categories that they’ve correlated to serve you up music you like.
I’ve explicitly left out the darker side of algorithms that drive users to more and more extreme content; there’s already plenty out there. But if you’d like another hit of dystopia, the Wall Street Journal recently ran a TikTok experiment, creating different personae to see what corners of TikTok’s content they would be exposed to. The results are discomfiting.
And Instagram, after years filling our feed with algorithmically-recommended posts, is reviving the chronological feed option in 2022.
I’m Anita Schillhorn van Veen, Director of Strategy at Mckinney in Los Angeles, and this is my little writing experiment. Pls share, like, and subscribe.