Google's AdWords Market: How Theory Influenced Practice
summary
Proposes a lightweight online bid scaling rule: multiply each bid by F(alpha)=1−e^{-(1−alpha)} where alpha is fraction of that advertiser’s budget spent, then award the impression to the highest scaled bid. Proves under small-bid assumptions this obtains a 1 - 1/e competitive ratio (~63%) versus greedy’s 50%, runs in O(1) per impression, handles heterogeneous budgets, and spawned real-world adoption and later stochastic refinements (Devanur & Hayes).
AI-written summary · about
chapters (13)
- 0:00Introduction to Google
- 0:28The Google business model
- 1:42Targeted advertising
- 3:43AdWords and auction mechanism
- 8:07Budget constraint problems
- 11:03Real-time algorithmic needs
- 12:43Budgeted matching algorithm
- 16:05Performance and impact
- 19:11Theoretical foundations
- 21:12The Balance algorithm
- 24:21LP based proof
- 32:09Broader economic impact
- 33:16Q&A and final thoughts
