On-Line Systems with Long-Range Goals
summary
Presents an empirical-plus-theory package: on Stack Overflow the civic duty badge (300 votes) and Electorate badge (600 question votes) show a measurable ramp-up in the rewarded actions right before the win, and a Coursera randomized experiment finds higher forum activity when badge progress or bylines are made salient. Models combine a state‑space dynamic program and a directed acyclic graph with a present bias multiplier to explain procrastination, task abandonment, and choice reduction; proves exponential worst‑case cost ratios require a k‑fan minor and shows the node‑deletion “make traversable” problem is NP‑hard.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 0:44The curve issue
- 2:01Guiding framework
- 3:53Stack Overflow
- 6:58First Model
- 21:00Badges
- 23:00Multistep plans
- 25:00Geometric discounting
- 30:00Choice reduction
- 31:53Statespaces
- 33:28Story
- 34:12Task abandonment
- 35:21Deadlines and classes
- 37:41Graphs
- 39:11Graph Miners
- 41:32Choice Reduction Problem
- 43:16Latent Story
- 43:56Deceptive Practices
- 47:14Commitment Devices
related talks
20mBoomerang: Rebounding the Consequences of Reputation Feedback on Crowdsourcing Platforms
30mFrom Service To Platform: A Ranking System in Go
57mLearning to Love your Non-functionals
50mDiversity in Matching Markets
52mDashboard Mechanisms for Online Marketplaces
19mRabbithole Quests and The Future of Work
24mIntroduction to Quadratic Voting
01h 08mUnderstanding and Shaping Urban Mobility Using Dollars and Data I
