Improved Bayes Risk Can Yield Reduced Social Welfare Under Competition

by   Meena Jagadeesan, et al.

As the scale of machine learning models increases, trends such as scaling laws anticipate consistent downstream improvements in predictive accuracy. However, these trends take the perspective of a single model-provider in isolation, while in reality providers often compete with each other for users. In this work, we demonstrate that competition can fundamentally alter the behavior of these scaling trends, even causing overall predictive accuracy across users to be non-monotonic or decreasing with scale. We define a model of competition for classification tasks, and use data representations as a lens for studying the impact of increases in scale. We find many settings where improving data representation quality (as measured by Bayes risk) decreases the overall predictive accuracy across users (i.e., social welfare) for a marketplace of competing model-providers. Our examples range from closed-form formulas in simple settings to simulations with pretrained representations on CIFAR-10. At a conceptual level, our work suggests that favorable scaling trends for individual model-providers need not translate to downstream improvements in social welfare in marketplaces with multiple model providers.


page 1

page 2

page 3

page 4


The impact of bundling licensed and unlicensed wireless service

Unlicensed spectrum has been viewed as a way to increase competition in ...

From Recommendation Systems to Facility Location Games

Recommendation systems are extremely popular tools for matching users an...

When to Limit Market Entry under Mandatory Purchase

We study a problem inspired by regulated health insurance markets, such ...

Inverse Scaling: When Bigger Isn't Better

Work on scaling laws has found that large language models (LMs) show pre...

Paid Prioritization with Content Competition

We study the effects of allowing paid prioritization arrangements in a m...

Optimizing Long-term Social Welfare in Recommender Systems: A Constrained Matching Approach

Most recommender systems (RS) research assumes that a user's utility can...

Data Justice in Digital Social Welfare: A Study of the Rythu Bharosa Scheme

While digital social protection systems have been claimed to bring effic...

Please sign up or login with your details

Forgot password? Click here to reset