Why can't I customize my Instagram algorithm?
If you log onto ChatGPT these days and go into settings, you’ll find this interface that asks you a couple of questions about yourself—how you’d like to be referred to, what you do for work, and anything else you’d like GPT to know about you to better customize its responses.
For instance, for me, I told it that “I generally code in R, so when I ask coding questions, give me answers in R,” and “I prefer a slightly more academic tone, since I’m a sociologist.”
But it dawned on me earlier today when I went in to check in on these settings that, like, why isn’t all of social media this way? Like, why can’t I tell Instagram or Twitter specifically that I’d like to mostly just see updates about Ohio State recruiting, La Liga highlights, and new innovations in computational sociology? Or, even better, the negative—I’d like Twitter to never show me content about the next idiotic thing Trump did or quasi-fascist claim from Elon.
Given recent advances in LLMs, it feels somewhat remarkable that my algorithms must be predicated on me at my worst moments—doom-scrolling, shitposting Andrew Yang, tired as hell—rather than me at my best moments, reflecting on what kind of content would be best for me given my personal and professional goals.
Instagram just patted itself on the back for introducing the ability to “wipe your algorithm clean,” but why not allow users to choose what they want to see? Why the fuck else were humans given consciousness and the ability to reason if not for situations like this—when human nature and the good life are in clear conflict?
Sure, some people are extremely good at “curating” their algorithm through intentional likes and marking “show less content like this” for posts that seem irrelevant. My partner Laura is one of these people. She also is a terrorist on my algorithms, often sneaking onto my phone only to like tons of content from a children’s YouTube show featuring a small green girl named Mona…But for most of us, social media is a blurry landscape where our most animal selves are native.
The answer to why this kind of function doesn’t exist is probably twofold. On the one hand, Meta or Twitter probably just doesn’t have the computational capacity or infrastructure to do so. On the other hand, these are publicly traded companies, and introducing a feature like this—something that would no doubt benefit the world greatly—would reduce shareholder value. This is, unfortunately, the classic bind of market capitalism and its greatest species failure as an economic system.
Despite this, I imagine that the first company to introduce this kind of elevated control over social media content will profit greatly. Perhaps it is OpenAI that is best situated to do so, given its existing infrastructural capacities. Who knows?
All I know is that I’d rather these algorithms be based off the best forms of myself rather than my worst.