Evolutionary Prediction Games
This addresses fairness and stability issues in machine learning systems for users affected by algorithmic feedback loops, presenting a novel theoretical framework with incremental insights into real-world constraints.
The paper tackles the problem of disparities in prediction quality when algorithms serve multiple users, leading to feedback loops that shape both the model and user population. It introduces evolutionary prediction games, showing that under realistic constraints like finite data, stable coexistence between user groups becomes possible, unlike in idealized settings where competitive exclusion occurs.
When a prediction algorithm serves a collection of users, disparities in prediction quality are likely to emerge. If users respond to accurate predictions by increasing engagement, inviting friends, or adopting trends, repeated learning creates a feedback loop that shapes both the model and the population of its users. In this work, we introduce evolutionary prediction games, a framework grounded in evolutionary game theory which models such feedback loops as natural-selection processes among groups of users. Our theoretical analysis reveals a gap between idealized and real-world learning settings: In idealized settings with unlimited data and computational power, repeated learning creates competition and promotes competitive exclusion across a broad class of behavioral dynamics. However, under realistic constraints such as finite data, limited compute, or risk of overfitting, we show that stable coexistence and mutualistic symbiosis between groups becomes possible. We analyze these possibilities in terms of their stability and feasibility, present mechanisms that can sustain their existence, and empirically demonstrate our findings.