Scheduling with Time-Evolving Uncertainty for Content Review Prioritization in Social Media
Talkat Bridging Prediction and Intervention Problems in Social Systems (Simons Institute, 2026) · 2026-01-12..01-18·
summary
Introduces a new queueing model where post view counts resolve over time and develops an asymptotically optimal scheduling algorithm (an opportunity‑adjusted remaining cost method, OaRC) that prices capacity via a dual gamma and reduces the problem to ski‑rental–style decoupled jobs. Simulations on real YouTube/Meta data show 3–8% reduction in prevalence and 7–25% reviewer‑hour savings at fixed capacity; an internal A/B test at Meta yields promising early results.
AI-written summary · about
chapters (9)
- 0:00Social media moderation intro
- 1:15Harmful content urgency
- 8:50AI-human review pipeline
- 14:35Scheduling problem setup
- 26:38Interactive scheduling quiz
- 34:55Modeling post virality
- 40:39Algorithm and results
- 46:24Practical impact
- 48:24Q&A
related talks
16mHALP: Heuristic Aided Learned Preference Eviction Policy for YouTube Content Delivery...
27mLearning and Optimization of Static and Dynamic Appointments
16mIntelligent Resource Scheduling for Co-located Latency-critical Services: A Multi-Model
40mA Bayesian Approach to Online Resource Allocation
29mLearning Outcomes in Queueing Systems
31mAvoiding Cascading Failures for Time-of-Use Pricing
41mDynamic Ranking of Search and Recommendation Results in High-Traffic Systems
28mMultitenancy and Fairness at Scale with Kueue: A Case Study