Decision Making at Scale: Algorithms, Mechanisms, and Platforms

on YouTube ↗37m

" summary

An algorithm called triadic consensus repeatedly samples random triads, asks each triad to pick its one-median, replaces the three sampled participants with three copies of that median, and runs until one winner remains. Under the assumption that triads can truthfully identify their median and the space is a median graphs metric, the process runs in about n log n steps and yields a 1 + O(sqrt((log n)/n)) approximation to the global median. Any locally consistent pairwise interactions fail to approximate the median.

AI-written summary · about

" chapters (15)

1. 0:00Start
2. 0:32Decision Making at Scale
3. 2:45High Level Design Goals
4. 3:58Talk outline
5. 7:02What is the ideal way to regulate off-road traffic?
6. 7:58The l-median (aka the geometric median or the Fermat-Weber point)
7. 9:48Going beyond voting
8. 10:58Triadic Interaction
9. 17:22A binary tree
10. 21:18A square grid
11. 26:48Is this reasonable?
12. 28:02A l-median magic box
13. 28:55NASH EQUILIBRIUM?
14. 30:03A two person reduction?
15. 31:44FUTURE