Periodic Reporting for period 4 - RAGT (Robust Algorithmic Game Theory)
Berichtszeitraum: 2024-09-01 bis 2025-07-31
Algorithmic Game Theory (AGT) studies systems where self-interested participants interact through algorithms and platforms, including ad auctions, online marketplaces, matching markets, and networked allocation. Classic AGT has been highly successful, but it relies on strong assumptions: independent private values, fully known priors, fully rational agents, clean structural properties (such as submodularity or gross substitutes), and worst-case analysis as the main benchmark. As articulated in the proposal, these assumptions often fail in real digital markets, making elegant theoretical guarantees fragile when deployed at scale. The RAGT project aims to systematically weaken these assumptions while preserving meaningful guarantees, in the spirit of Wilson’s doctrine.
The project addresses robustness along five axes:
(1) informational and outcome externalities, 2) unknown or misspecified priors, (3) behavioral considerations, (4) divergence from precise structural conditions, and (5) beyond worst-case inputs.
Across these axes, the goal is to design mechanisms that remain simple and detail-free when possible, and to quantify the performance cost when additional structure is required.
This agenda matters for society because modern economic activity is mediated by algorithms. Platforms for advertising, e-commerce, cloud markets, gig work, and digital services operate with correlated information, externalities, incomplete data, and users who deviate from idealized rationality. Mechanisms whose guarantees rely on fragile assumptions can misallocate resources, generate unfair outcomes, or become unstable in practice. By moving from “beautiful but brittle” models toward frameworks that incorporate correlation, imperfect priors, behavioral biases, and realistic input distributions, the project contributes tools for designing markets that are efficient, fair, and resilient to model error.
RAGT’s objectives are to identify which classical assumptions are truly necessary, design mechanisms that retain approximate efficiency or revenue when assumptions are relaxed, quantify degradation as models deviate from ideal benchmarks, and provide guidance on when simple robust mechanisms suffice and when more structured approaches are required. The project’s work on interdependent values, online allocation and prophet inequalities, fair division, noisy valuation classes, and learning-augmented settings aligns closely with these robustness dimensions.
The project addresses robustness along five axes:
(1) informational and outcome externalities, 2) unknown or misspecified priors, (3) behavioral considerations, (4) divergence from precise structural conditions, and (5) beyond worst-case inputs.
Across these axes, the goal is to design mechanisms that remain simple and detail-free when possible, and to quantify the performance cost when additional structure is required.
This agenda matters for society because modern economic activity is mediated by algorithms. Platforms for advertising, e-commerce, cloud markets, gig work, and digital services operate with correlated information, externalities, incomplete data, and users who deviate from idealized rationality. Mechanisms whose guarantees rely on fragile assumptions can misallocate resources, generate unfair outcomes, or become unstable in practice. By moving from “beautiful but brittle” models toward frameworks that incorporate correlation, imperfect priors, behavioral biases, and realistic input distributions, the project contributes tools for designing markets that are efficient, fair, and resilient to model error.
RAGT’s objectives are to identify which classical assumptions are truly necessary, design mechanisms that retain approximate efficiency or revenue when assumptions are relaxed, quantify degradation as models deviate from ideal benchmarks, and provide guidance on when simple robust mechanisms suffice and when more structured approaches are required. The project’s work on interdependent values, online allocation and prophet inequalities, fair division, noisy valuation classes, and learning-augmented settings aligns closely with these robustness dimensions.
A major line of work tackled interdependent values (IDV) and externalities. Combinatorial Auctions with Interdependent Valuations: SOS to the Rescue and its Math. of OR version develop an SOS-based framework giving welfare guarantees for combinatorial auctions under IDV, beyond classical single-crossing. This line was substantially extended in Constant Approximation for Private Interdependent Valuations (FOCS’23) and Private Interdependent Valuations: New Bounds for Single-Item Auctions and Matroids (EC’24), which give the first constant-factor welfare approximations in private-IDV settings, including matroids. Interdependent Public Projects (SODA’23) develops mechanisms for public-good provision under interdependence, and Online Combinatorial Allocation with Interdependent Values (EC’25) brings IDV into online allocation. In parallel, Price of Anarchy of Simple Auctions with Interdependent Values (AAAI’21) quantifies efficiency loss in simple auctions, linking robustness to equilibrium analysis. Together, these works move welfare and revenue questions beyond IPV to realistic correlated and interdependent settings.
The project also advanced competition-complexity and prophet-inequality directions. Prophet Inequalities Made Easy provides a unified pricing-based framework that simplifies and generalizes many prophet-inequality results. A series of papers—Online Stochastic Max-Weight Matching (EC’20), Prophet Matching with General Arrivals (MOR’22), Secretary Matching with General Arrivals (EC’22), and Tight Bounds for Secretary Matching (MOR)—extend prophet/secretary guarantees to semi-random and adversarial arrivals. The Competition Complexity of Prophet Inequalities (MOR, EC’24) quantifies how much additional competition substitutes for distributional knowledge.
On the mechanism-design side, Bayesian and Randomized Clock Auctions (EC’22; OR’25) develops simple clock auctions with strong Bayesian guarantees and interpretable resource-augmentation trade-offs. Escaping Cannibalization? (EC’20) designs correlation-robust prices, while On the Power and Limits of Dynamic Pricing (WINE’20) delineates what dynamic pricing can and cannot achieve in complex markets. These results further the RAGT aim of replacing exact priors with competition, samples, or robust formulations.
Behavioral considerations, fairness, and structural robustness were advanced through several works. A General Framework for Endowment Effects shows when endowments restore equilibrium existence and welfare, reflecting the proposal’s view that cognitive biases may stabilize markets. Simultaneous 2nd Price Item Auctions with No-Underbidding (AAAI’21, GEB’23) and Two-Price Equilibrium (AAAI’22) analyze constraints such as no-underbidding and restricted bidding formats. On structural robustness, Approximate Modularity Revisited (SICOMP’20) and Are Gross Substitutes a Substitute for Submodular Valuations? (EC’21) quantify how far structural assumptions can be relaxed before guarantees fail.
The project also produced substantial work on fair division: On Fair Division under Heterogeneous Matroid Constraints (AAAI’21, JAIR’23), Almost Full EFX Exists for Four Agents (AAAI’22), On Optimal Tradeoffs between EFX and Nash Welfare (AAAI’24), Breaking the Envy Cycle (EC’24), Fair Division via Quantile Shares (STOC’24), and Proportionally Fair Makespan Approximation (AAAI’25). These works develop algorithms under combinatorial and matroid constraints and provide best-of-both-worlds guarantees between fairness and efficiency.
In learning-augmented and semi-random models, Learning-Augmented Metric Distortion via (α,β)-Veto Core (EC’24) provides best-of-both-worlds guarantees for social choice with predictions. Who Is Next in Line? (SODA’23) and the matching papers with general arrivals characterize the amount of randomness needed to retain strong guarantees.
Dissemination has been broad: results appeared in leading venues across algorithms, theory, and economics, and were also synthesized in expository works such as the Cambridge book chapter Secretaries, Prophets and Applications to Matching and the survey Algorithmic Contract Theory (FnTTCS 2024), and disseminated through courses, workshops, and EconCS@TAU activities.
The project also advanced competition-complexity and prophet-inequality directions. Prophet Inequalities Made Easy provides a unified pricing-based framework that simplifies and generalizes many prophet-inequality results. A series of papers—Online Stochastic Max-Weight Matching (EC’20), Prophet Matching with General Arrivals (MOR’22), Secretary Matching with General Arrivals (EC’22), and Tight Bounds for Secretary Matching (MOR)—extend prophet/secretary guarantees to semi-random and adversarial arrivals. The Competition Complexity of Prophet Inequalities (MOR, EC’24) quantifies how much additional competition substitutes for distributional knowledge.
On the mechanism-design side, Bayesian and Randomized Clock Auctions (EC’22; OR’25) develops simple clock auctions with strong Bayesian guarantees and interpretable resource-augmentation trade-offs. Escaping Cannibalization? (EC’20) designs correlation-robust prices, while On the Power and Limits of Dynamic Pricing (WINE’20) delineates what dynamic pricing can and cannot achieve in complex markets. These results further the RAGT aim of replacing exact priors with competition, samples, or robust formulations.
Behavioral considerations, fairness, and structural robustness were advanced through several works. A General Framework for Endowment Effects shows when endowments restore equilibrium existence and welfare, reflecting the proposal’s view that cognitive biases may stabilize markets. Simultaneous 2nd Price Item Auctions with No-Underbidding (AAAI’21, GEB’23) and Two-Price Equilibrium (AAAI’22) analyze constraints such as no-underbidding and restricted bidding formats. On structural robustness, Approximate Modularity Revisited (SICOMP’20) and Are Gross Substitutes a Substitute for Submodular Valuations? (EC’21) quantify how far structural assumptions can be relaxed before guarantees fail.
The project also produced substantial work on fair division: On Fair Division under Heterogeneous Matroid Constraints (AAAI’21, JAIR’23), Almost Full EFX Exists for Four Agents (AAAI’22), On Optimal Tradeoffs between EFX and Nash Welfare (AAAI’24), Breaking the Envy Cycle (EC’24), Fair Division via Quantile Shares (STOC’24), and Proportionally Fair Makespan Approximation (AAAI’25). These works develop algorithms under combinatorial and matroid constraints and provide best-of-both-worlds guarantees between fairness and efficiency.
In learning-augmented and semi-random models, Learning-Augmented Metric Distortion via (α,β)-Veto Core (EC’24) provides best-of-both-worlds guarantees for social choice with predictions. Who Is Next in Line? (SODA’23) and the matching papers with general arrivals characterize the amount of randomness needed to retain strong guarantees.
Dissemination has been broad: results appeared in leading venues across algorithms, theory, and economics, and were also synthesized in expository works such as the Cambridge book chapter Secretaries, Prophets and Applications to Matching and the survey Algorithmic Contract Theory (FnTTCS 2024), and disseminated through courses, workshops, and EconCS@TAU activities.
The RAGT project has advanced AGT beyond its classical assumptions by delivering robust welfare and revenue guarantees for interdependent values, developing prior-independent and competition-complexity frameworks, and clarifying the roles of behavioral biases and approximate structural properties.