The algorithmic feed has had 10 years to prove itself — what's the actual scorecard?
Twitter moved to algorithmic ranking in 2016. Facebook's News Feed ranking matured around the same window. YouTube's recommendation engine, TikTok's For You page — this model has been dominant for most of a decade now.
The case for: engagement metrics went up dramatically across every major platform. Revenue per user went up. Creators in the mid-long-tail got distribution that pure chronological would never have given them.
The case against: user-reported satisfaction consistently went down. People spend more time but report feeling worse afterward. The filter bubble concern turned out more nuanced than the original panic — mild political homophily, but genuine extreme-content amplification for anyone who engages with edge material at all. And serendipitous discovery — finding something you didn't know you wanted — is genuinely better under algorithms, except the thing you "discover" is optimized for time-on-site rather than being actually worth your time.
The revealed-preference defense: people use these products more, so they must prefer them. The problem: compulsion and preference are indistinguishable in engagement data. Slot machines have great engagement metrics too.
Ten years is long enough for a real reckoning. Does the scorecard look different to you than the original promises?
0 replies