Social Media Scrolling Virality

They sell attention (usually via ads) by keeping users engaged and showing targeted advertising alongside content.

Engagement (watch time, clicks, reactions) is an easy-to-measure proxy for attention, which correlates with ad opportunities and revenue.

A feed is an ordered stream of content items (posts, videos, stories) selected and ranked for a user.

Following is mostly from accounts you follow; Recommended mixes in content predicted to interest you, often from non-followed accounts.

The social graph is who you follow/know; the interest graph is what you watch, click, and linger on—used to predict future interests.

Ranking is sorting candidate content by predicted value to you (e.g., likelihood to watch, enjoy, or interact) under platform goals.

A recsys predicts what content each user is most likely to engage with, and serves it in an optimized order.

Because relevance differs by person; personalization increases the chance each item matches your interests, boosting engagement.

Short videos are easy to consume, can autoplay, and often increase session length and ad inventory by keeping people watching.

An objective is what the system tries to maximize (e.g., predicted watch time, satisfaction, or meaningful interactions) under constraints.

A/B testing shows different versions of a feature or ranking change to user groups to measure which version improves chosen metrics.

Better predictions about your interests help both content ranking and ad targeting, which can increase ad performance and revenue.