Dashboard Mechanisms for Online Marketplaces
summary
Proposes publishing per-agent bidding dashboards that estimate win probability for every bid and then inverting observed bids to recover values, allowing a platform that must use winner-pays-bid rules to implement the engineers’ value-space allocation; constructs three dashboards (last-stage inferred values, last-winning-stage, and a rebalancing dashboard that tracks payment residuals) and analyzes convergence: static convergence in two rounds, residuals bounded by ±V, and approximate strategic equivalence to the sequential truthful mechanism in dynamic settings. Uses bid inversion and a single-call implementation compatible with live-only algorithms; payments converge and the rebalancing prevents runaway imbalances.
AI-written summary · about
chapters (11)
- 0:00Intro
- 1:51Dashboard Mechanisms
- 2:40Online Marketplaces
- 3:39Algorithm Design vs Mechanism Design
- 6:10Winners Pay Bid Format Mechanisms
- 9:01How to Choose the Algorithm
- 18:02Outline
- 20:28Truthful Mechanism
- 28:58Dashboard Construction Analysis
- 30:48Dynamic Agents
- 34:13Dashboards
related talks
25mBudget Pacing in Repeated Auctions: Regret and Efficiency without Convergence
44mAdverse Selection and Auction Design for Internet Display Advertising
44mOnline Advertisements and Online Algorithms
44mClearing Matching Markets Efficiently: Informative signals and Match Recommendations
29mClearing Matching Markets Efficiently: Informative Signals and Match Recommendations
59mMechanism Design with Limited Commitment
10mSynchronized Priority Auctions
29mKeynote Playing games in the clouds