Premise
- Recommendation engines rank and ration attention like planners with dashboards.
- The hierarchy of “needs” they enforce is learned from past data and business goals, not from plural public consent.
Systems view
- Data -> model -> feed -> behavior -> new data; bias compounds when feedback loops are closed and opaque.
- Gateways (app stores, payment rails) add another layer of centralized allocation.
Social and political stakes
- Voice distribution: who gets reach, who gets throttled, which communities are misread.
- Governance risk: policy by tweak, without transparency or appeal, can mimic decree.
Questions to explore
- What forms of user agency (ranked choice feeds, interoperability, audits) preserve plural signals?
- How to prevent a soft-planned attention economy from collapsing into monoculture?
Expansion notes
- Include examples: newsfeed ranking, creator monetization, app store policy shifts.
- Keep the systems thinking visible but accessible to beginners.