From Social Graphs to For-You Feeds: How Entertainment Ate Communication

Social platforms were built to connect people; then their economics shifted to entertain them. What’s notable here is not just the TikTokization of every app, but the technical inversion behind it: recommender systems optimized for watch time displaced the social graph as the primary routing layer. Under the hood, objective functions like dwell time and session length drive ranking models, which in turn reward high-entropy video over low-friction conversation. The result: feeds recentered on creators and clips, while “communication” retreated into private group chats and encrypted DMs-harder to monetize, less indexable, and largely invisible to discovery.

Worth noting, even messaging products now ship broadcast primitives-Channels, Stories, Reels-because the ad stack, creator tooling, and CDNs are tuned for one-to-many video. The infra tells the story: on-device editors, auto-captioning, ABR streaming, and aggressive codec pipelines make production and consumption trivial; contact import and open APIs, once growth engines for new comms startups, are throttled or closed. The bigger picture is a distribution realignment: public attention flows through algorithmic feeds that favor entertainment, while genuine conversation goes dark, fragmented across small networks with weak virality. For builders, this means new communication plays must either anchor in utility (workflow, identity, interoperability) or accept the gravity of entertainment-first discovery.