What if recommendation systems were prediction markets?
Open market predicting what users want or should see. Is this effectively an ad market all the way down? What are the incentives for users?
Recommendation systems are often used as a way to verticalize and retain usage through first-party data. How does this work practically when the system is open?
Are LLMs isolated recommendation systems? A recommendation system optimizes what's next to maximize expected value to the user. What is the intention of an LLM - to predict?
The design space between recommendations, predictions, and markets is underexplored.