Scrolling through Jonah Peretti’s chat with Brian Morrissey on The Rebooting Show—episode 152 if you missed it—I kept nodding (and occasionally muttering “exactly”) at his diagnosis of the algorithmic attention trap. Peretti argues that once platforms told deep-learning models to optimize for time-spent, the feed evolved into pure SNARF: Stakes, Novelty, Anger, Retention, Fear. The AI wins engagement; humans lose agency.
the-rebooting-show.captivate.fm
Here’s where I land: the problem isn’t that machines are curating; it’s what we ask them to maximize. If your north star is “keep ’em scrolling,” you’ll inevitably reward outrage and anxiety. But flip the brief—optimize for constructive participation, delight, and real-world uplift—and AI can amplify the best parts of human creativity instead of the worst.
That design choice sits at the heart of a social project my team is quietly engineering now. We’re building with the same cutting-edge models, but the KPI isn’t doom-scroll minutes; it’s positive actions sparked—proving that algorithms can serve people, not hijack them. More soon. Until then, the podcast is worth every minute: the-rebooting-show.captivate.fm/episode/trb-...
