Research & Papers

Chaos in Autobidding Auctions

The algorithms running your ads are inherently unpredictable, study finds.

Deep Dive

New research proves that simple autobidding systems in online ad auctions can exhibit formally chaotic behavior, making their long-term outcomes inherently unpredictable. The study shows these dynamics can simulate classic chaotic systems like Chua's circuit and connect to chaotic population models. This strengthens previous findings that only showed quasiperiodicity, revealing that ostensibly simple second-price auctions are far more complex and unstable than previously assumed.

Why It Matters

This means billions in digital ad spending could be subject to unpredictable, chaotic market fluctuations.