The Illusion of Control: Why Your "Not Interested" Button Barely Works

We’ve all been there: you stumble across a dance trend or a cooking hack you have absolutely no interest in. You dutifully tap the "Not Interested" button, expecting the app to learn your boundaries and curate your feed accordingly. Yet, a few days later, a nearly identical video pops up on your screen. Are these algorithms simply ignoring us?

A recent study by computer scientists at Northwestern University set out to answer exactly that by conducting an "algorithm audit" on TikTok’s famous For You Page (FYP). Their findings reveal a fascinating dynamic: users possess far less agency over their personalized feeds than the platform's interface leads them to believe.

The disconnect lies in how recommendation engines weigh different types of data. While hitting "dislike" or "show me less" is an explicit signal of your preference, modern algorithms are famously driven by implicit signals—most notably, watch time. If you linger on a bizarre video out of sheer confusion before finally swiping away, the system registers your prolonged attention. To the algorithm, your eyes matter much more than your clicks.

The Northwestern researchers discovered that negative feedback isn't entirely useless, but its effect is fleeting. When users tell the app they don't want to see a specific type of content, the algorithm temporarily adjusts. However, it quickly "relapses," slowly reintroducing the unwanted content unless the user aggressively and consistently repeats their negative feedback.

As co-author Piotr Sapiezynski pointed out, this raises a compelling question about platform design: why offer a user-control feature if its impact is so easily overridden by the system's default behavior? Sapiezynski's team uses these audits to understand how platforms fail and how they might trap individuals in unwanted content loops.

This phenomenon highlights a broader issue in modern social media: the illusion of control. Platforms provide buttons that make us feel like the active curators of our own digital spaces, but the underlying mechanics are relentlessly optimizing for engagement. Understanding this temporary "relapse" effect doesn't just explain a frustrating quirk of our favorite apps; it reminds us that in the attention economy, our unconscious scrolling habits speak much louder than our stated preferences. Digital literacy today means recognizing this invisible tug-of-war.

Key Points

• Recommendation engines prioritize implicit signals like watch time over explicit "not interested" clicks.
• Negative feedback only temporarily suppresses unwanted content before the algorithm relapses.
• Users must repeatedly and consistently reject content to permanently alter their feed.
• Features like "show less" often provide a false sense of agency to social media users.

Why It Matters

Understanding how algorithms weigh our attention against our stated preferences is crucial for maintaining digital autonomy and breaking out of unwanted content loops.

Sources:

• TikTok users don't have as much agency over their FYPs as they think — Ars Technica AI