Why algorithm-driven recommendations stopped working — and what to use instead
Algorithms learned to optimize for engagement, not enjoyment. Here's what changed, and how to opt out without disappearing.
The shift
For most of the 2010s, "the algorithm" was a quiet helper. Netflix surfaced things you'd like. Spotify's Discover Weekly felt eerily personal. Amazon's "people also bought" was useful, if a little obvious.
Somewhere around 2020, the deal changed. Engagement metrics ate everything. The algorithm's job stopped being "find what you'll love" and became "find what you'll click on." Those are different problems.
Why this happened
Three forces compounded:
- Ad models reward attention. Time-on-app beats satisfaction-per-minute.
- Catalog inflation. Every platform has 100x the content it did in 2015. Surfacing the right thing is harder.
- Engagement is measurable. Joy isn't. You optimize what you can count.
The result is a kind of recommendation fatigue. We trust the suggestions less. We default to what's trending. We end up watching the same five shows everyone else is watching.
The alternative isn't no recommendations
It's human recommendations. The friend who knows your taste and tells you "you'll love this" has always outperformed any algorithm. The internet just made it harder to capture that signal at scale.
That's what Kyndrid is for. Your circle. Their vouches. Nothing in between.
If you're tired of being optimized at, try it.
Start your circle on Kyndrid.
Free on iOS. Recommendations from the people you actually trust.
— The Kyndrid Team