Modeling Content Creator Incentives on Algorithm-Curated Platforms

  title={Modeling Content Creator Incentives on Algorithm-Curated Platforms},
  author={Jiri Hron and Karl Krauth and Michael I. Jordan and Niki Kilbertus and Sarah Dean},
Content creators compete for user attention. Their reach crucially depends on algorithmic choices made by developers on online platforms. To maximize exposure, many creators adapt strategically, as evidenced by examples like the sprawling search engine optimization industry. This begets competition for the finite user attention pool. We formalize these dynamics in what we call an exposure game , a model of incentives induced by algorithms including modern factorization and (deep) two-tower… 

Supply-Side Equilibria in Recommender Systems

This paper investigates the supply-side equilibria in content recommender systems, model users and content as D -dimensional vectors, and shows that producers can achieve positive profit at equilibrium, which is typically impossible under perfect competition.



Content Provider Dynamics and Coordination in Recommendation Ecosystems

This work investigates the dynamics of content creation using a game-theoretic lens, and shows that the dynamics will always converge to a pure Nash Equilibrium (PNE), but the convergence rate can be exponential.

From Recommendation Systems to Facility Location Games

This work proposes the widely studied framework of facility location games to study recommendation systems with strategic content providers, and proposes a mediator that implements the socially optimal strategy profile as the unique equilibrium profile, and shows a tight bound on its intervention cost.

A/B Testing for Recommender Systems in a Two-sided Marketplace

The proposed UniCoRn (Unifying Counterfactual Rankings) approach is agnostic to the density of the producer-consumer network and does not rely on any treatment propagation assumption, making this widely applicable to the industrial setting where the underlying network is unknown and challenging to predict a priori due to its dynamic nature.

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

This work explores settings in which content providers cannot remain viable unless they receive a certain level of user engagement, and forms the recommendation problem in this setting as one of equilibrium selection in the induced dynamical system, and shows that it can be solved as an optimal constrained matching problem.

Convergence of Learning Dynamics in Information Retrieval Games

It is proved that under the probability ranking principle (PRP), which forms the basis of the current state of the art ranking methods, any better-response learning dynamics converges to a pure Nash equilibrium.

Regulating algorithmic filtering on social media

This work proposes a unifying framework that considers the key stakeholders of AF regulation, and mathematically formalizes this framework, using it to construct a data-driven, statistically sound regulatory procedure that satisfies several important criteria.

Online certification of preference-based fairness for personalized recommender systems

A sample-efficient algorithm with theoretical guarantees that it does not deteriorate user experience is proposed to audit for envy-freeness, a more granular criterion aligned with individual preferences.

Quantifying Availability and Discovery in Recommender Systems via Stochastic Reachability

This work proposes an evaluation procedure based on stochastic reachability to quantify the maximum probability of recommending a target piece of content to an user for a set of allowable strategic modifications and shows that this metric can be computed efficiently as a convex program for a variety of practical settings.

Strategic Ranking

It is found that randomization in the ranking reward design can mitigate two measures of disparate impact, welfare gap and access, whereas non-randomization may induce a high level of competition that systematically excludes a disadvantaged group.

Towards Content Provider Aware Recommender Systems: A Simulation Study on the Interplay between User and Provider Utilities

A REINFORCE recommender agent is built to optimize a joint objective of user utility and the counterfactual utility lift of the content provider associated with the recommended content, which it is shown to be equivalent to maximizing overall user Utility and the utilities of all content providers on the platform under some mild assumptions.