Dynamics of Learning under User Choice: Overspecialization and Peer-Model Probing

TL;DR

This paper investigates how multiple platforms learning from shared users can fall into an overspecialization trap, and proposes a peer-model probing algorithm to improve global performance in such multi-agent learning scenarios.

Contribution

It introduces the overspecialization trap phenomenon and proposes a peer-model probing algorithm to mitigate it, with theoretical analysis and empirical validation.

Findings

Overspecialization can lead to poor global performance despite low local loss.

Probing peer models helps learners access data from less-preferred users.

The proposed algorithm converges under certain conditions, improving overall risk.

Abstract

In many economically relevant contexts where machine learning is deployed, multiple platforms obtain data from the same pool of users, each of whom selects the platform that best serves them. Prior work in this setting focuses exclusively on the "local" losses of learners on the distribution of data that they observe. We find that there exist instances where learners who use existing algorithms almost surely converge to models with arbitrarily poor global performance, even when models with low full-population loss exist. This happens through a feedback-induced mechanism, which we call the overspecialization trap: as learners optimize for users who already prefer them, they become less attractive to users outside this base, which further restricts the data they observe. Inspired by the recent use of knowledge distillation in modern ML, we propose an algorithm that allows learners to "probe" the predictions of peer models, enabling them to learn about users who do not select them. Our analysis characterizes when probing succeeds: this procedure converges almost surely to a stationary point with bounded full-population risk when probing sources are sufficiently informative, e.g., a known market leader or a majority of peers with good global performance. We verify our findings with semi-synthetic experiments on the MovieLens, Census, and Amazon Sentiment datasets.