Core stability is a natural and well-studied notion for group fairness in multi-winner voting, where the task is to select a committee from a pool of candidates. We study the setting where voters either approve or disapprove of each candidate; here, it remains a major open problem whether a core-stable committee always exists. In this work, we develop an approach based on mixed-integer linear programming for deciding whether and when core-stable committees are guaranteed to exist. In contrast to SAT-based approaches popular in computational social choice, our method can produce proofs for a specific number of candidates independent of the number of voters. In addition to these computational gains, our program lends itself to a novel duality-based reformulation of the core stability problem, from which we obtain new existence results in special cases. Further, we use our framework to reveal previously unknown relationships between core stability and other desirable properties, such as notions of priceability.

Alignment methods in moral domains seek to elicit moral preferences of human stakeholders and incorporate them into AI. This presupposes moral preferences as static targets, but such preferences often evolve over time. Proper alignment of AI to dynamic human preferences should ideally account for "legitimate" changes to moral reasoning, while ignoring changes related to attention deficits, cognitive biases, or other arbitrary factors. However, common AI alignment approaches largely neglect temporal changes in preferences, posing serious challenges to proper alignment, especially in high-stakes applications of AI, e.g., in healthcare domains, where misalignment can jeopardize the trustworthiness of the system and yield serious individual and societal harms. This work investigates the extent to which people's moral preferences change over time, and the impact of such changes on AI alignment. Our study is grounded in the kidney allocation domain, where we elicit responses to pairwise comparisons of hypothetical kidney transplant patients from over 400 participants across 3-5 sessions. We find that, on average, participants change their response to the same scenario presented at different times around 6-20% of the time (exhibiting "response instability"). Additionally, we observe significant shifts in several participants' retrofitted decision-making models over time (capturing "model instability"). Predictive performance of simple AI models decreases as a function of both response and model instability. Moreover, predictive performance diminishes over time, highlighting the importance of accounting for temporal changes in preferences during training. These findings raise fundamental normative and technical challenges relevant to AI alignment, highlighting the need to better understand the object of alignment (what to align to) when user preferences change significantly over time, including the mechanisms underlying these changes.

Imperfect-recall games—in which players may forget previously acquired information—have found many practical applications, ranging from game abstractions to team games and testing AI agents. In this paper, we quantify the utility gain by endowing a player with perfect recall, which we call the value of recall (VoR). While VoR can be unbounded in general, we parameterize it in terms of various game properties, namely the structure of chance nodes and the degree of absentmindedness (the number of successive times a player enters the same information set). Further, we identify several pathologies that arise with VoR, and show how to circumvent them. We also study the complexity of computing VoR, and how to optimally apportion partial recall. Finally, we connect VoR to other previously studied concepts in game theory, including the price of anarchy. We use that connection in conjunction with the celebrated smoothness framework to characterize VoR in a broad class of games.

In Tennenholtz’s program equilibrium, players of a game submit programs to play on their behalf. Each program receives the other programs’ source code and outputs an action. This can model interactions involving AI agents, mutually transparent institutions, or commitments. Tennenholtz 2004 (https://doi.org/10.1016/j.geb.2004.02.002) proves a folk theorem for program games, but the equilibria constructed are very brittle. We therefore consider simulation-based programs – i.e., programs that work by running opponents’ programs. These are relatively robust (in particular, two programs that act the same are treated the same) and are more practical than proof-based approaches. Oesterheld’s (2019, https://doi.org/10.1007/s11238-018-9679-3) epsilon-Grounded-pi-Bot is such an approach. Unfortunately, it is not generally applicable to games of three or more players, and only allows for a limited range of equilibria in two player games. In this paper, we propose a generalisation to Oesterheld’s (2019) epsilon-Grounded-pi-Bot. We prove a folk theorem for our programs in a setting with access to a shared source of randomness. We then characterise their equilibria in a setting without shared randomness. Both with and without shared randomness, we achieve a much wider range of equilibria than Oesterheld’s (2019) epsilon-Grounded-pi-Bot. Finally, we explore the limits of simulation-based program equilibrium, showing that the Tennenholtz folk theorem cannot be attained by simulation-based programs without access to shared randomness.

Strategic interactions can be represented more concisely, and analyzed and solved more efficiently, if we are aware of the symmetries within the multiagent system. Symmetries also have conceptual implications, for example for equilibrium selection. We study the computational complexity of identifying and using symmetries. Using the classical framework of normal-form games, we consider game symmetries that can be across some or all players and/or actions. We find a strong connection between game symmetries and graph automorphisms, yielding graph automorphism and graph isomorphism completeness results for characterizing the symmetries present in a game. On the other hand, we also show that the problem becomes polynomial-time solvable when we restrict the consideration of actions in one of two ways. Next, we investigate when exactly game symmetries can be successfully leveraged for Nash equilibrium computation. We show that finding a Nash equilibrium that respects a given set of symmetries is PPAD- and CLS-complete in general-sum and team games respectively---that is, exactly as hard as Brouwer fixed point and gradient descent problems. Finally, we present polynomial-time methods for the special cases where we are aware of a vast number of symmetries, or where the game is two-player zero-sum and we do not even know the symmetries.

A growing body of work in Ethical AI attempts to capture human moral judgments through simple computational models.The key question we address in this work is whether such simple AI models capture the critical nuances of moral decision-making by focusing on the use case of kidney allocation.We conducted twenty interviews where participants explained their rationale for their judgments about who should receive a kidney.We observe participants: (a) value patients' morally-relevant attributes to different degrees; (b) use diverse decision-making processes, citing heuristics to reduce decision complexity; (c) can change their opinions; (d) sometimes lack confidence in their decisions (e.g., due to incomplete information); and (e) express enthusiasm and concern regarding AI assisting humans in kidney allocation decisions.Based on these findings, we discuss challenges of computationally modeling moral judgments as a stand-in for human input, highlight drawbacks of current approaches, and suggest future directions to address these issues.

We study partially observable assistance games (POAGs), a model of the human-AI value alignment problem which allows the human and the AI assistant to have partial observations. Motivated by concerns of AI deception, we study a qualitatively new phenomenon made possible by partial observability: would an AI assistant ever have an incentive to interfere with the human’s observations? First, we prove that sometimes an optimal assistant must take observation-interfering actions, even when the human is playing optimally, and even when there are otherwise-equivalent actions available that do not interfere with observations. Though this result seems to contradict the classic theorem from single-agent decision making that the value of perfect information is nonnegative, we resolve this seeming contradiction by developing a notion of interference defined on entire policies. This can be viewed as an extension of the classic result that the value of perfect information is nonnegative into the cooperative multiagent setting. Second, we prove that if the human is simply making decisions based on their immediate outcomes, the assistant might need to interfere with observations as a way to query the human’s preferences. We show that this incentive for interference goes away if the human is playing optimally, or if we introduce a communication channel for the human to communicate their preferences to the assistant. Third, we show that if the human acts according to the Boltzmann model of irrationality, this can create an incentive for the assistant to interfere with observations. Finally, we use an experimental model to analyze tradeoffs faced by the AI assistant in practice when considering whether or not to take observation-interfering actions.

*Variational inequalities (VIs)* encompass many fundamental problems in diverse areas ranging from engineering to economics and machine learning. However, their considerable expressivity comes at the cost of computational intractability. In this paper, we introduce and analyze a natural relaxation—which we refer to as *expected variational inequalities (EVIs)*—where the goal is to find a distribution that satisfies the VI constraint in expectation. By adapting recent techniques from game theory, we show that, unlike VIs, EVIs can be solved in polynomial time under general (nonmonotone) operators. EVIs capture the seminal notion of correlated equilibria, but enjoy a greater reach beyond games. We also employ our framework to capture and generalize several existing disparate results, including from settings such as smooth games, and games with coupled constraints or nonconcave utilities.

This position paper argues for two claims regarding AI testing and evaluation. First, to remain informative about deployment behaviour, evaluations need account for the possibility that AI systems understand their circumstances and reason strategically. Second, game-theoretic analysis can inform evaluation design by formalising and scrutinising the reasoning in evaluation-based safety cases. Drawing on examples from existing AI systems, a review of relevant research, and formal strategic analysis of a stylised evaluation scenario, we present evidence for these claims and motivate several research directions.

We study two axioms for social choice functions that capture the impact of similar candidates: independence of clones (IoC) and composition consistency (CC). We clarify the relationship between these axioms by observing that CC is strictly more demanding than IoC, and investigate whether common voting rules that are known to be independent of clones (such as Single Transferable Vote, Ranked Pairs, Schulze Method, and Split Cycle) are composition-consistent. While for most of these rules the answer is negative, we identify a variant of Ranked Pairs that satisfies CC. Further, we show how to efficiently modify any (neutral) social choice function so that it satisfies CC, while maintaining its other desirable properties. Our transformation relies on the hierarchical representation of clone structures via PQ-trees. We extend our analysis to social preference functions. Finally, we interpret IoC and CC as measures of robustness against strategic manipulation by candidates, with IoC corresponding to strategy-proofness and CC corresponding to obvious strategy-proofness.

Φ-equilibria—and the associated notion of Φ-regret—are a powerful and flexible framework at the heart of online learning and game theory, whereby enriching the set of deviations Φ begets stronger notions of rationality. Recently, Daskalakis, Farina, Fishelson, Pipis, and Schneider (STOC '24) settled the existence of efficient algorithms when Φ contains only linear maps under a general, d-dimensional convex constraint set X. In this paper, we significantly extend their work by resolving the case where Φ is k-dimensional; degree-ℓ polynomials constitute a canonical such example with k = dO(ℓ). In particular, we obtain two main positive results: i) a poly(n, d, k, log(1/ϵ))-time algorithm for computing ϵ-approximate Φ-equilibria in n-player multilinear games, and ii) an efficient online algorithm that incurs average Φ-regret at most ϵ using poly(d, k)/ϵ2 rounds.

Usually, to apply game-theoretic methods, we must specify utilities precisely, and we run the risk that the solutions we compute are not robust to errors in this specification. Ordinal games provide an attractive alternative: they require specifying only which outcomes are preferred to which other ones. Unfortunately, they provide little guidance for how to play unless there are pure Nash equilibria; evaluating mixed strategies appears to fundamentally require cardinal utilities. In this paper, we observe that we can in fact make good use of mixed strategies in ordinal games if we consider settings that allow for folk theorems. These allow us to find equilibria that are robust, in the sense that they remain equilibria no matter which cardinal utilities are the correct ones -- as long as they are consistent with the specified ordinal preferences. We analyze this concept and study the computational complexity of finding such equilibria in a range of settings.

Bilateral trade is one of the most natural and important forms of economic interaction: A seller has a single, indivisible item for sale, and a buyer is potentially interested. The two parties typically have different, privately known valuations for the item, and ideally, they would like to trade if the buyer values the item more than the seller. The celebrated impossibility result by Myerson and Satterthwaite shows that any mechanism for this setting must violate at least one important desideratum. In this paper, we investigate a richer paradigm of bilateral trade, with many self-interested buyers and sellers on both sides of a single trade who cannot be excluded from the trade. We show that this allows for more positive results. In fact, we establish a dichotomy in the possibility of trading efficiently. If in expectation, the buyers value the item more, we can achieve efficiency in the limit. If this is not the case, then efficiency cannot be achieved in general. En route, we characterize trading mechanisms that encourage truth-telling, which may be of independent interest. We also evaluate our trading mechanisms experimentally, and the experiments align with our theoretical results.

Preference elicitation frameworks feature heavily in the research on participatory ethical AI tools and provide a viable mechanism to enquire and incorporate the moral values of various stakeholders. As part of the elicitation process, surveys about moral preferences, opinions, and judgments are typically administered only once to each participant. This methodological practice is reasonable if participants’ responses are stable over time such that, all other things being held constant, their responses today will be the same as their responses to the same questions at a later time. However, we do not know how often that is the case. It is possible that participants’ true moral preferences change, are subject to temporary moods or whims, or are influenced by environmental factors we don’t track. If participants’ moral responses are unstable in such ways, it would raise important methodological and theoretical issues for how participants’ true moral preferences, opinions, and judgments can be ascertained. We address this possibility here by asking the same survey participants the same moral questions about which patient should receive a kidney when only one is available ten times in ten different sessions over two weeks, varying only presentation order across sessions. We measured how often participants gave different responses to simple (Study One) and more complicated (Study Two) controversial and uncontroversial repeated scenarios. On average, the fraction of times participants changed their responses to controversial scenarios (i.e., were unstable) was around 10-18% (±14-15%) across studies, and this instability is observed to have positive associations with response time and decision-making difficulty. We discuss the implications of these results for the efficacy of common moral preference elicitation methods, highlighting the role of response instability in potentially causing value misalignment between the stakeholders and AI tools trained on their moral judgments.

Computational preference elicitation methods are tools used to learn people’s preferences quantitatively in a given context. Recent works on preference elicitation advocate for active learning as an efficient method to iteratively construct queries (framed as comparisons between context-specific cases) that are likely to be most informative about an agent’s underlying preferences. In this work, we argue that the use of active learning for moral preference elicitation relies on certain assumptions about the underlying moral preferences, which can be violated in practice. Specifically, we highlight the following common assumptions (a) preferences are stable over time and not sensitive to the sequence of presented queries, (b) the appropriate hypothesis class is chosen to model moral preferences, and (c) noise in the agent’s responses is limited. While these assumptions can be appropriate for preference elicitation in certain domains, prior research on moral psychology suggests they may not be valid for moral judgments. Through a synthetic simulation of preferences that violate the above assumptions, we observe that active learning can have similar or worse performance than a basic random query selection method in certain settings. Yet, simulation results also demonstrate that active learning can still be viable if the degree of instability or noise is relatively small and when the agent’s preferences can be approximately represented with the hypothesis class used for learning. Our study highlights the nuances associated with effective moral preference elicitation in practice and advocates for the cautious use of active learning as a methodology to learn moral preferences.

Foundation models such as GPT-4 are fine-tuned to avoid unsafe or otherwise problematic behavior, such as helping to commit crimes or producing racist text. One approach to fine-tuning, called reinforcement learning from human feedback, learns from humans’ expressed preferences over multiple outputs. Another approach is constitutional AI, in which the input from humans is a list of high-level principles. But how do we deal with potentially diverging input from humans? How can we aggregate the input into consistent data about “collective” preferences or otherwise use it to make collective choices about model behavior? In this paper, we argue that the field of social choice is well positioned to address these questions, and we discuss ways forward for this agenda, drawing on discussions in a recent workshop on Social Choice for AI Ethics and Safety held in Berkeley, CA, USA in December 2023.

Multi-agent AI research promises a path to develop human-like and human-compatible intelligent technologies that complement the solipsistic view of other approaches, which mostly do not consider interactions between agents. Aiming to make progress in this direction, the Melting Pot contest 2023 focused on the problem of cooperation among interacting agents and challenged researchers to push the boundaries of multi-agent reinforcement learning (MARL) for mixed-motive games. The contest leveraged the Melting Pot environment suite to rigorously evaluate how well agents can adapt their cooperative skills to interact with novel partners in unforeseen situations. Unlike other reinforcement learning challenges, this challenge focused on social rather than environmental generalization. In particular, a population of agents performs well in Melting Pot when its component individuals are adept at finding ways to cooperate both with others in their population and with strangers. Thus Melting Pot measures cooperative intelligence.The contest attracted over 600 participants across 100+ teams globally and was a success on multiple fronts: (i) it contributed to our goal of pushing the frontiers of MARL towards building more cooperatively intelligent agents, evidenced by several submissions that outperformed established baselines; (ii) it attracted a diverse range of participants, from independent researchers to industry affiliates and academic labs, both with strong background and new interest in the area alike, broadening the field’s demographic and intellectual diversity; and (iii) analyzing the submitted agents provided important insights, highlighting areas for improvement in evaluating agents' cooperative intelligence. This paper summarizes the design aspects and results of the contest and explores the potential of Melting Pot as a benchmark for studying Cooperative AI. We further analyze the top solutions and conclude with a discussion on promising directions for future research.

Quantitative Relative Judgment Aggregation (QRJA) is a new research topic in (computational) social choice. In the QRJA model, agents provide judgments on the relative quality of different candidates, and the goal is to aggregate these judgments across all agents. In this work, our main conceptual contribution is to explore the interplay between QRJA in a social choice context and its application to ranking prediction. We observe that in QRJA, judges do not have to be people with subjective opinions; for example, a race can be viewed as a ``judgment'' on the contestants' relative abilities. This allows us to aggregate results from multiple races to evaluate the contestants' true qualities. At a technical level, we introduce new aggregation rules for QRJA and study their structural and computational properties. We evaluate the proposed methods on data from various real races and show that QRJA-based methods offer effective and interpretable ranking predictions.

A mediator observes no-regret learners playing an extensive-form game repeatedly across T rounds. The mediator attempts to steer players toward some desirable predetermined equilibrium by giving (nonnegative) payments to players. We call this the steering problem. The steering problem captures problems several problems of interest, among them equilibrium selection and information design (persuasion). If the mediator's budget is unbounded, steering is trivial because the mediator can simply pay the players to play desirable actions. We study two bounds on the mediator's payments: a total budget and a per-round budget. If the mediator's total budget does not grow with T, we show that steering is impossible. However, we show that it is enough for the total budget to grow sublinearly with T, that is, for the average payment to vanish. When players' full strategies are observed at each round, we show that constant per-round budgets permit steering. In the more challenging setting where only trajectories through the game tree are observable, we show that steering is impossible with constant per-round budgets in general extensive-form games, but possible in normal-form games or if the per-round budget may itself depend on T. We also show how our results can be generalized to the case when the equilibrium is being computed online while steering is happening. We supplement our theoretical positive results with experiments highlighting the efficacy of steering in large games.

AI systems can interact in unexpected ways, sometimes with disastrous consequences. As AI gets to control more of our world, these interactions will become more common and have higher stakes. As AI becomes more advanced, these interactions will become more sophisticated, and game theory will provide the tools for analyzing these interactions. However, AI agents are in some ways unlike the agents traditionally studied in game theory, introducing new challenges as well as opportunities. We propose a research agenda to develop the game theory of highly advanced AI agents, with a focus on achieving cooperation.

Game-theoretic interactions with AI agents could differ from traditional human-human interactions in various ways. One such difference is that it may be possible to simulate an AI agent (for example because its source code is known), which allows others to accurately predict the agent's actions. This could lower the bar for trust and cooperation. In this paper, we first formally define games in which one player can simulate another at a cost, and derive some basic properties of such games. Then, we prove a number of results for such games, including: (1) introducing simulation into generic-payoff normal-form games makes them easier to solve; (2) if the only obstacle to cooperation is a lack of trust in the possibly-simulated agent, simulation enables equilibria that improve the outcome for both agents; and (3) however, there are settings where introducing simulation results in strictly worse outcomes for both players.

We study single-player extensive-form games with imperfect recall, such as the Sleeping Beauty problem or the Absentminded Driver game. For such games, two natural equilibrium concepts have been proposed as alternative solution concepts to ex-ante optimality. One equilibrium concept uses generalized double halving (GDH) as a belief system and evidential decision theory (EDT), and another one uses generalized thirding (GT) as a belief system and causal decision theory (CDT). Our findings relate those three solution concepts of a game to solution concepts of a polynomial maximization problem: global optima, optimal points with respect to subsets of variables and Karush–Kuhn–Tucker (KKT) points. Based on these correspondences, we are able to settle various complexity-theoretic questions on the computation of such strategies. For ex-ante optimality and (EDT,GDH)-equilibria, we obtain NP-hardness and inapproximability, and for (CDT,GT)-equilibria we obtain CLS-completeness results.

As machine learning agents act more autonomously in the world, they will increasingly interact with each other. Unfortunately, in many social dilemmas like the one-shot Prisoner’s Dilemma, standard game theory predicts that ML agents will fail to cooperate with each other. Prior work has shown that one way to enable cooperative outcomes in the one-shot Prisoner’s Dilemma is to make the agents mutually transparent to each other, i.e., to allow them to access one another’s source code (Rubinstein, 1998; Tennenholtz, 2004) – or weights in the case of ML agents. However, full transparency is often unrealistic, whereas partial transparency is commonplace. Moreover, it is challenging for agents to learn their way to cooperation in the full transparency setting. In this paper, we introduce a more realistic setting in which agents only observe a single number indicating how similar they are to each other. We prove that this allows for the same set of cooperative outcomes as the full transparency setting. We also demonstrate experimentally that cooperation can be learned using simple ML methods.

We introduce a new approach for computing optimal equilibria via learning in games. It applies to extensive-form settings with any number of players, including mechanism design, information design, and solution concepts such as correlated, communication, and certification equilibria. We observe that optimal equilibria are minimax equilibrium strategies of a player in an extensive-form zero-sum game. This reformulation allows to apply techniques for learning in zero-sum games, yielding the first learning dynamics that converge to optimal equilibria, not only in empirical averages, but also in iterates. We demonstrate the practical scalability and flexibility of our approach by attaining state-of-the-art performance in benchmark tabular games, and by computing an optimal mechanism for a sequential auction design problem using deep reinforcement learning.

We study equilibrium computation with extensive-form correlation in two-player turn-taking stochastic games. Our main results are two-fold: (1) We give an algorithm for computing a Stackelberg extensive-form correlated equilibrium (SEFCE), which runs in time polynomial in the size of the game, as well as the number of bits required to encode each input number. (2) We give an efficient algorithm for approximately computing an optimal extensive-form correlated equilibrium (EFCE) up to machine precision, i.e., the algorithm achieves approximation error ε in time polynomial in the size of the game, as well as log(1/ε).

In conference peer review, reviewers are often asked to provide “bids” on each submitted paper that express their interest in reviewing that paper. A paper assignment algorithm then uses these bids (along with other data) to compute a high-quality assignment of reviewers to papers. However, this process has been exploited by malicious reviewers who strategically bid in order to unethically manipulate the paper assignment, crucially undermining the peer review process. For example, these reviewers may aim to get assigned to a friend’s paper as part of a quid-pro-quo deal. A critical impediment towards creating and evaluating methods to mitigate this issue is the lack of any publicly-available data on malicious paper bidding. In this work, we collect and publicly release a novel dataset to fill this gap, collected from a mock conference activity where participants were instructed to bid either honestly or maliciously. We further provide a descriptive analysis of the bidding behavior, including our categorization of different strategies employed by participants. Finally, we evaluate the ability of each strategy to manipulate the assignment, and also evaluate the performance of some simple algorithms meant to detect malicious bidding. The performance of these detection algorithms can be taken as a baseline for future research on detecting malicious bidding.

We study the problem of learning influence adoption in networks. In this problem, a communicable entity (such as an infectious disease, a computer virus, or a social media meme) propagates through a network, and the goal is to learn the state of each individual node by sampling only a small number of nodes and observing/testing their states. We study this problem in heterogeneous networks, in which each individual node has a set of distinct features that determine how it is affected by the propagating entity. We give an efficient algorithm with nearly optimal sample complexity for two variants of this learning problem, corresponding to symptomatic and asymptomatic spread. In each case, the optimal sample complexity naturally generalizes the complexity of learning how nodes are affected in isolation, and the complexity of learning influence adoption in a homogeneous network.

We pose and study the problem of planning in Markov decision processes (MDPs), subject to participation constraints as studied in mechanism design. In this problem, a planner must work with a self-interested agent on a given MDP. Each action in the MDP provides an immediate reward to the planner and a (possibly different) reward to the agent. The agent has no control in choosing the actions, but has the option to end the entire process at any time. The goal of the planner is to find a policy that maximizes her cumulative reward, taking into consideration the agent's ability to terminate. We give a fully polynomial-time approximation scheme for this problem. En route, we present polynomial-time algorithms for computing (exact) optimal policies for important special cases of this problem, including when the time horizon is constant, or when the MDP exhibits a "definitive decisions" property. We illustrate our algorithms with two different game-theoretic applications: the problem of assigning rides in ride-sharing and the problem of designing screening policies. Our results imply efficient algorithms for computing (approximately) optimal policies in both applications.

Many scientific conferences employ a two-phase paper review process, where some papers are assigned additional reviewers after the initial reviews are submitted. Many conferences also design and run experiments on their paper review process, where some papers are assigned reviewers who provide reviews under an experimental condition. In this paper, we consider the question: how should reviewers be divided between phases or conditions in order to maximize total assignment similarity? We make several contributions towards answering this question. First, we prove that when the set of papers requiring additional review is unknown, a simplified variant of this problem is NP-hard. Second, we empirically show that across several datasets pertaining to real conference data, dividing reviewers between phases/conditions uniformly at random allows an assignment that is nearly as good as the oracle optimal assignment. This uniformly random choice is practical for both the two-phase and conference experiment design settings. Third, we provide explanations of this phenomenon by providing theoretical bounds on the suboptimality of this random strategy under certain natural conditions. From these easily-interpretable conditions, we provide actionable insights to conference program chairs about whether a random reviewer split is suitable for their conference.

Although it has been known since the 1970s that a globally optimal strategy profile in a common-payoff game is a Nash equilibrium, global optimality is a strict requirement that limits the result’s applicability. In this work, we show that any locally optimal symmetric strategy profile is also a (global) Nash equilibrium. Furthermore, we show that this result is robust to perturbations to the common payoff and to the local optimum. Applied to machine learning, our result provides a global guarantee for any gradient method that finds a local optimum in symmetric strategy space. While this result indicates stability to unilateral deviation, we nevertheless identify broad classes of games where mixed local optima are unstable under joint, asymmetric deviations. We analyze the prevalence of instability by running learning algorithms in a suite of symmetric games, and we conclude by discussing the applicability of our results to multi-agent RL, cooperative inverse RL, and decentralized POMDPs.

We consider the problem of planning with participation constraints introduced in[24]. In this problem, a principal chooses actions in a Markov decision process, resulting in separate utilities for the principal and the agent. However, the agent can and will choose to end the process whenever his expected onward utility becomes negative. The principal seeks to compute and commit to a policy that maximizes her expected utility, under the constraint that the agent should always want to continue participating. We provide the first polynomial-time exact algorithm for this problem for finite-horizon settings, where previously only an additive ε-approximation algorithm was known. Our approach can also be extended to the (discounted) infinite-horizon case, for which we give an algorithm that runs in time polynomial in the size of the input and log(1/ε), and returns a policy that is optimal up to an additive error of ε.

Abstract A set of players delegate playing a game to a set of representatives, one for each player. We imagine that each player trusts their respective representative’s strategic abilities. Thus, we might imagine that per default, the original players would simply instruct the representatives to play the original game as best as they can. In this paper, we ask: are there safe Pareto improvements on this default way of giving instructions? That is, we imagine that the original players can coordinate to tell their representatives to only consider some subset of the available strategies and to assign utilities to outcomes differently than the original players. Then can the original players do this in such a way that the payoff is guaranteed to be weakly higher than under the default instructions for all the original players? In particular, can they Pareto-improve without probabilistic assumptions about how the representatives play games? In this paper, we give some examples of safe Pareto improvements. We prove that the notion of safe Pareto improvements is closely related to a notion of outcome correspondence between games. We also show that under some specific assumptions about how the representatives play games, finding safe Pareto improvements is NP-complete.

Machine learning techniques can be useful in applications such as credit approval and college admission. However, to be classified more favorably in such contexts, an agent may decide to strategically withhold some of her features, such as bad test scores. This is a missing data problem with a twist: which data is missing depends on the chosen classifier, because the specific classifier is what may create the incentive to withhold certain feature values. We address the problem of training classifiers that are robust to this behavior. We design three classification methods: MINCUT, Hill-Climbing (HC) and Incentive-Compatible Logistic Regression (IC-LR). We show that MINCUT is optimal when the true distribution of data is fully known. However, it can produce complex decision boundaries, and hence be prone to overfitting in some cases. Based on a characterization of truthful classifiers (i.e., those that give no incentive to strategically hide features), we devise a simpler alternative called HC which consists of a hierarchical ensemble of out-of-the-box classifiers, trained using a specialized hill-climbing procedure which we show to be convergent. For several reasons, MINCUT and HC are not effective in utilizing a large number of complementarily informative features. To this end, we present IC-LR, a modification of Logistic Regression that removes the incentive to strategically drop features. We also show that our algorithms perform well in experiments on real-world data sets, and present insights into their relative performance in different settings.

AI systems are often used to make or contribute to important decisions in a growing range of applications, including criminal justice, hiring, and medicine. Since these decisions impact human lives, it is important that the AI systems act in ways which align with human values. Techniques for preference modeling and social choice help researchers learn and aggregate peoples' preferences, which are used to guide AI behavior; thus, it is imperative that these learned preferences are accurate. These techniques often assume that people are willing to express strict preferences over alternatives; which is not true in practice. People are often indecisive, and especially so when their decision has moral implications. The philosophy and psychology literature shows that indecision is a measurable and nuanced behavior---and that there are several different reasons people are indecisive. This complicates the task of both learning and aggregating preferences, since most of the relevant literature makes restrictive assumptions on the meaning of indecision. We begin to close this gap by formalizing several mathematical indecision models based on theories from philosophy, psychology, and economics; these models can be used to describe (indecisive) agent decisions, both when they are allowed to express indecision and when they are not. We test these models using data collected from an online survey where participants choose how to (hypothetically) allocate organs to patients waiting for a transplant.

We study the problem of automated mechanism design with partial verification, where each type can (mis)report only a restricted set of types (rather than any other type), induced by the principal's limited verification power. We prove hardness results when the revelation principle does not necessarily hold, as well as when types have even minimally different preferences. In light of these hardness results, we focus on truthful mechanisms in the setting where all types share the same preference over outcomes, which is motivated by applications in, e.g., strategic classification. We present a number of algorithmic and structural results, including an efficient algorithm for finding optimal deterministic truthful mechanisms, which also implies a faster algorithm for finding optimal randomized truthful mechanisms via a characterization based on the notion of convexity. We then consider a more general setting, where the principal's cost is a function of the combination of outcomes assigned to each type. In particular, we focus on the case where the cost function is submodular, and give generalizations of essentially all our results in the classical setting where the cost function is additive. Our results provide a relatively complete picture for automated mechanisms design with partial verification.

We study test-based binary classification, where a principal either accepts or rejects agents based on the outcomes they get in a set of tests. The principal commits to a policy, which consists of all sets of outcomes that lead to acceptance. Each agent is modeled by a distribution over the space of possible outcomes. When an agent takes a test, he pays a cost and receives an independent sample from his distribution as the outcome. Agents can always choose between taking another test and stopping. They maximize their expected utility, which is the value of acceptance if the principal's policy accepts the set of outcomes they have and 0 otherwise, minus the total cost of tests taken. We focus on the case where agents can be either "good" or "bad" (corresponding to their distribution over test outcomes), and the principal's goal is to accept good agents and reject bad ones. We show, roughly speaking, that as long as the good and bad agents have different distributions (which can be arbitrarily close to each other), the principal can always achieve perfect accuracy, meaning good agents are accepted with probability 1, and bad ones are rejected with probability 1. Moreover, there is a policy achieving perfect accuracy under which the maximum number of tests any agent needs to take is constant — in sharp contrast to the case where the principal directly observes samples from agents' distributions. The key technique is to choose the policy so that agents self-select into taking tests.

We study PAC learning in the presence of strategic manipulation, where data points may modify their features in certain predefined ways in order to receive a better outcome. We show that the vanilla ERM principle fails to achieve any nontrivial guarantee in this context. Instead, we propose an incentive-aware version of the ERM principle which has asymptotically optimal sample complexity. We then focus our attention on incentive-compatible classifiers, which provably prevent any kind of strategic manipulation. We give a sample complexity bound that is, curiously, independent of the hypothesis class, for the ERM principle restricted to incentive-compatible classifiers. This suggests that incentive compatibility alone can act as an effective means of regularization. We further show that it is without loss of generality to consider only incentive-compatible classifiers when opportunities for strategic manipulation satisfy a transitivity condition. As a consequence, in such cases, our hypothesis-class-independent sample complexity bound applies even without incentive compatibility. Our results set the foundations of incentive-aware PAC learning.

AI has the potential to revolutionize many areas of healthcare. Radiology, dermatology, and ophthalmology are some of the areas most likely to be impacted in the near future, and they have received significant attention from the broader research community. But AI techniques are now also starting to be used in in vitro fertilization (IVF), in particular for selecting which embryos to transfer to the woman. The contribution of AI to IVF is potentially significant, but must be done carefully and transparently, as the ethical issues are significant, in part because this field involves creating new people. We first give a brief introduction to IVF and review the use of AI for embryo selection. We discuss concerns with the interpretation of the reported results from scientific and practical perspectives. We then consider the broader ethical issues involved. We discuss in detail the problems that result from the use of black-box methods in this context and advocate strongly for the use of interpretable models. Importantly, there have been no published trials of clinical effectiveness, a problem in both the AI and IVF communities, and we therefore argue that clinical implementation at this point would be premature. Finally, we discuss ways for the broader AI community to become involved to ensure scientifically sound and ethically responsible development of AI in IVF.

AI is increasingly making decisions, not only for us, but also about us -- from whether we are invited for an interview, to whether we are proposed as a match for someone looking for a date, to whether we are released on bail. Often, we have some control over the information that is available to the algorithm; we can self-report some information, and other information we can choose to withhold. This creates a potential circularity: the classifier used, mapping submitted information to outcomes, depends on the (training) data that people provide, but the (test) data depend on the classifier, because people will reveal their information strategically to obtain a more favorable outcome. This setting is not adversarial, but it is also not fully cooperative. Mechanism design provides a framework for making good decisions based on strategically reported information, and it is commonly applied to the design of auctions and matching mechanisms. However, the setting above is unlike these common applications, because in it, preferences tend to be similar across agents, but agents are restricted in what they can report. This creates both new challenges and new opportunities, as we demonstrate in our theoretical work and our initial experiments. This is joint work with Hanrui Zhang, Andrew Kephart, Yu Cheng, Anilesh Krishnaswamy, Haoming Li, and David Rein.

We study Bayesian automated mechanism design in unstructured dynamic environments, where a principal repeatedly interacts with an agent, and takes actions based on the strategic agent's report of the current state of the world. Both the principal and the agent can have arbitrary and potentially different valuations for the actions taken, possibly also depending on the actual state of the world. Moreover, at any time, the state of the world may evolve arbitrarily depending on the action taken by the principal. The goal is to compute an optimal mechanism which maximizes the principal's utility in the face of the self-interested strategic agent.We give an efficient algorithm for computing optimal mechanisms, with or without payments, under different individual-rationality constraints, when the time horizon is constant. Our algorithm is based on a sophisticated linear program formulation, which can be customized in various ways to accommodate richer constraints. For environments with large time horizons, we show that the principal's optimal utility is hard to approximate within a certain constant factor, complementing our algorithmic result. These results paint a relatively complete picture for automated dynamic mechanism design in unstructured environments. We further consider a special case of the problem where the agent is myopic, and give a refined efficient algorithm whose time complexity scales linearly in the time horizon. In the full version of the paper, we show that memoryless mechanisms, which are without loss of generality optimal in Markov decision processes without strategic behavior, do not provide a good solution for our problem, in terms of both optimality and computational tractability. Moreover, we present experimental results where our algorithms are applied to synthetic dynamic environments with different characteristics, which not only serve as a proof of concept for our algorithms, but also exhibit intriguing phenomena in dynamic mechanism design.

Given AI's growing role in modeling and improving decision-making, how and when to present users with feedback is an urgent topic to address. We empirically examined the effect of feedback from false AI on moral decision-making about donor kidney allocation. We found some evidence that judgments about whether a patient should receive a kidney can be influenced by feedback about participants' own decision-making perceived to be given by AI, even if the feedback is entirely random. We also discovered different effects between assessments presented as being from human experts and assessments presented as being from AI.

We study the problem of learning opinions in social networks. The learner observes the states of some sample nodes from a social network, and tries to infer the states of other nodes, based on the structure of the network. We show that sample-efficient learning is impossible when the network exhibits strong noise, and give a polynomial-time algorithm for the problem with nearly optimal sample complexity when the network is sufficiently stable.

We study problems where a learner aims to learn the valuations of an agent by observing which goods he buys under varying price vectors. More specifically, we consider the case of a $k$-demand agent, whose valuation over the goods is additive when receiving up to $k$ goods, but who has no interest in receiving more than $k$ goods. We settle the query complexity for the active-learning (preference elicitation) version, where the learner chooses the prices to post, by giving a \emph{biased binary search} algorithm, generalizing the classical binary search procedure. We complement our query complexity upper bounds by lower bounds that match up to lower-order terms. We also study the passive-learning version in which the learner does not control the prices, and instead they are sampled from some distribution. We show that in the PAC model for passive learning, any \emph{empirical risk minimizer} has a sample complexity that is optimal up to a factor of $\widetilde{O}(k)$.

We consider three important challenges in conference peer review: (i) reviewers maliciously attempting to get assigned to certain papers to provide positive reviews, possibly as part of quid-pro-quo arrangements with the authors; (ii) "torpedo reviewing," where reviewers deliberately attempt to get assigned to certain papers that they dislike in order to reject them; (iii) reviewer de-anonymization on release of the similarities and the reviewer-assignment code. On the conceptual front, we identify connections between these three problems and present a framework that brings all these challenges under a common umbrella. We then present a (randomized) algorithm for reviewer assignment that can optimally solve the reviewer-assignment problem under any given constraints on the probability of assignment for any reviewer-paper pair. We further consider the problem of restricting the joint probability that certain suspect pairs of reviewers are assigned to certain papers, and show that this problem is NP-hard for arbitrary constraints on these joint probabilities but efficiently solvable for a practical special case. Finally, we experimentally evaluate our algorithms on datasets from past conferences, where we observe that they can limit the chance that any malicious reviewer gets assigned to their desired paper to 50% while producing assignments with over 90% of the total optimal similarity.

We consider a combinatorial variant of the classical ski rental problem --- which we call combinatorial ski rental --- where multiple resources are available to purchase and to rent, and are demanded online. Moreover, the costs of purchasing and renting are potentially combinatorial. The dual problem of combinatorial ski rental, which we call combinatorial online bipartite matching, generalizes the classical online bipartite matching problem into a form where constraints, induced by both offline and online vertices, can be combinatorial. We give a 2-competitive (resp. e / (e - 1)-competitive) deterministic (resp. randomized) algorithm for combinatorial ski rental, and an e / (e - 1)-competitive algorithm for combinatorial online bipartite matching. All these ratios are optimal given simple lower bounds inherited from the respective well-studied special cases. We also prove information-theoretic impossibility of constant-factor algorithms when any part of our assumptions is considerably relaxed.

Research in artificial intelligence, as well as in economics and other related fields, generally proceeds from the premise that each agent has a well-defined identity, well-defined preferences over outcomes, and well-defined beliefs about the world. However, as we design AI systems, we in fact need to specify where the boundaries between one agent and another in the system lie, what objective functions these agents aim to maximize, and to some extent even what belief formation processes they use. The premise of this paper is that as AI is being broadly deployed in the world, we need well-founded theories of, and methodologies and algorithms for, how to design preferences, identities, and beliefs. This paper lays out an approach to address these problems from a rigorous foundation in decision theory, game theory, social choice theory, and the algorithmic and computational aspects of these fields.

We consider the problem of fairly dividing a collection of indivisible goods among a set of players. Much of the existing literature on fair division focuses on notions of individual fairness. For instance, envy-freeness requires that no player prefer the set of goods allocated to another player to her own allocation. We observe that an algorithm satisfying such individual fairness notions can still treat groups of players unfairly, with one group desiring the goods allocated to another. Our main contribution is a notion of group fairness, which implies most existing notions of individual fairness. Group fairness (like individual fairness) cannot be satisfied exactly with indivisible goods. Thus, we introduce two “up to one good” style relaxations. We show that, somewhat surprisingly, certain local optima of the Nash welfare function satisfy both relaxations and can be computed in pseudo-polynomial time by local search. Our experiments reveal faster computation and stronger fairness guarantees in practice.

Specifying the objective function that an AI system should pursue can be challenging. Especially when the decisions to be made by the system have a moral component, input from multiple stakeholders is often required. We consider approaches that query them about their judgments in individual examples, and then aggregate these judgments into a general policy. We propose a formal learning-theoretic framework for this setting. We then give general results on how to translate classical results from PAC learning into results in our framework. Subsequently, we show that in some settings, better results can be obtained by working directly in our framework. Finally, we discuss how our model can be extended in a variety of ways for future research.

In standard classification problems, the assumption is that the entity making the decision (the principal) has access to all the samples. However, in many contexts, she either does not have direct access to the samples, or can inspect only a limited set of samples and does not know which are the most relevant ones. In such cases, she must rely on another party (the agent) to either provide the samples or point out the most relevant ones. If the agent has a different objective, then the principal cannot trust the submitted samples to be representative. She must set a policy for how she makes decisions, keeping in mind the agent’s incentives. In this paper, we introduce a theoretical framework for this problem and provide key structural and computational results.

Often, a principal must make a decision based on data provided by an agent. Moreover, typically, that agent has an interest in the decision that is not perfectly aligned with that of the principal. Thus, the agent may have an incentive to select from or modify the samples he obtains before sending them to the principal. In other settings, the principal may not even be able to observe samples directly; instead, she must rely on signals that the agent is able to send based on the samples that he obtains, and he will choose these signals strategically. In this paper, we give necessary and sufficient conditions for when the principal can distinguish between agents of good'' andbad'' types, when the type affects the distribution of samples that the agent has access to. We also study the computational complexity of checking these conditions. Finally, we study how many samples are needed.

A paper by Deng and Conitzer in AAAI'17 introduces disarmament games, in which players alternatingly commit not to play certain pure strategies. However, in practice, disarmament usually does not consist in removing a strategy, but rather in removing a resource (and doing so rules out all the strategies in which that resource is used simultaneously). In this paper, we introduce a model of disarmament games in which resources, rather than strategies, are removed. We prove NP-completeness of several formulations of the problem of achieving desirable outcomes via disarmament. We then study the case where resources can be fractionally removed, and prove a result analogous to the folk theorem that all desirable outcomes can be achieved. We show that we can approximately achieve any desirable outcome in a polynomial number of rounds, though determining whether a given outcome can be obtained in a given number of rounds remains NP-complete.

The efficient allocation of limited resources is a classical problem in economics and computer science. In kidney exchanges, a central market maker allocates living kidney donors to patients in need of an organ. Patients and donors in kidney exchanges are prioritized using ad-hoc weights decided on by committee and then fed into an allocation algorithm that determines who get what—and who does not. In this paper, we provide an end-to-end methodology for estimating weights of individual participant profiles in a kidney exchange. We first elicit from human subjects a list of patient attributes they consider acceptable for the purpose of prioritizing patients (e.g., medical characteristics, lifestyle choices, and so on). Then, we ask subjects comparison queries between patient profiles and estimate weights in a principled way from their responses. We show how to use these weights in kidney exchange market clearing algorithms. We then evaluate the impact of the weights in simulations and find that the precise numerical values of the weights we computed matter little, other than the ordering of profiles that they imply. However, compared to not prioritizing patients at all, there is a significant effect, with certain classes of patients being (de)prioritized based on the human-elicited value judgments.

The efficient allocation of limited resources is a classical problem in economics and computer science. In kidney exchanges, a central market maker allocates living kidney donors to patients in need of an organ. Patients and donors in kidney exchanges are prioritized using ad-hoc weights decided on by committee and then fed into an allocation algorithm that determines who get what---and who does not. In this paper, we provide an end-to-end methodology for estimating weights of individual participant profiles in a kidney exchange. We first elicit from human subjects a list of patient attributes they consider acceptable for the purpose of prioritizing patients (e.g., medical characteristics, lifestyle choices, and so on). Then, we ask subjects comparison queries between patient profiles and estimate weights in a principled way from their responses. We show how to use these weights in kidney exchange market clearing algorithms. We then evaluate the impact of the weights in simulations and find that the precise numerical values of the weights we computed matter little, other than the ordering of profiles that they imply. However, compared to not prioritizing patients at all, there is a significant effect, with certain classes of patients being (de)prioritized based on the human-elicited value judgments.

AI systems are now or will soon be sophisticated enough to make consequential decisions. Although this technology has flourished, we also need public appraisals of AI systems playing these more important roles. This article reports surveys of preferences for and against AI systems making decisions in various domains as well as experiments that intervene on these preferences. We find that these preferences are contingent on subjects' previous exposure to computer systems making these kinds of decisions, and some interventions designed to mimic previous exposure successfully encourage subjects to be more hospitable to computer systems making these weighty decisions.

The problem of optimally scheduling the charging demand of electric vehicles within the constraints of the electricity infrastructure is called the charge scheduling problem. The models of the charging speed, horizon, and charging demand determine the computational complexity of the charge scheduling problem. We show that for about 20 variants the problem is either in P or weakly NP-hard and dynamic programs exist to compute optimal solutions. About 10 other variants of the problem are strongly NP-hard, presenting a potentially significant obstacle to their use in practical situations of scale. An experimental study establishes up to what parameter values the dynamic programs can determine optimal solutions in a couple of minutes.

This paper presents a new way of formalizing the coalition structure generation problem (CSG) so that we can apply constraint optimization techniques to it. Forming effective coalitions is a major research challenge in AI and multi-agent systems. CSG involves partitioning a set of agents into coalitions to maximize social surplus. Traditionally, the input of the CSG problem is a black-box function called a characteristic function, which takes a coalition as input and returns the value of the coalition. As a result, applying constraint optimization techniques to this problem has been infeasible. However, characteristic functions that appear in practice often can be represented concisely by a set of rules, rather than treating the function as a black box. Then we can solve the CSG problem more efficiently by directly applying constraint optimization techniques to this compact representation. We present new formalizations of the CSG problem by utilizing recently developed compact representation schemes for characteristic functions. We first characterize the complexity of CSG under these representation schemes. In this context, the complexity is driven more by the number of rules than by the number of agents. As an initial step toward developing efficient constraint optimization algorithms for solving the CSG problem, we also develop mixed integer programming formulations and show that an off-the-shelf optimization package can perform reasonably well.

We introduce a new class of mechanisms, robust mechanisms, that is an intermediary between ex-post mechanisms and Bayesian mechanisms. This new class of mechanisms allows the mechanism designer to incorporate imprecise estimates of the distribution over bidder valuations in a way that provides strong guarantees that the mechanism will perform at least as well as ex-post mechanisms, while in many cases performing better. We further extend this class to mechanisms that are with high probability incentive compatible and individually rational, ε-robust mechanisms. Using techniques from automated mechanism design and robust optimization, we provide an algorithm polynomial in the number of bidder types to design robust and ε-robust mechanisms. We show experimentally that this new class of mechanisms can significantly outperform traditional mechanism design techniques when the mechanism designer has an estimate of the distribution and the bidder’s valuation is correlated with an externally verifiable signal.

The generality of decision and game theory has enabled domain-independent progress in AI research. For example, a better algorithm for finding good policies in (PO)MDPs can be instantly used in a variety of applications. But such a general theory is lacking when it comes to moral decision making. For AI applications with a moral component, are we then forced to build systems based on many ad-hoc rules? In this paper we discuss possible ways to avoid this conclusion.

Much recent work in the AI community concerns algorithms for computing optimal mixed strategies to commit to, as well as the deployment of such algorithms in real security applications. Another possibility is to commit not to play certain actions. If only one player makes such a commitment, then this is generally less powerful than completely committing to a single mixed strategy. However, if players can alternatingly commit not to play certain actions and thereby iteratively reduce their strategy spaces, then desirable outcomes can be obtained that would not have been possible with just a single player committing to a mixed strategy. We refer to such a setting as a disarmament game. In this paper, we study disarmament for two-player normal-form games. We show that deciding whether an outcome can be obtained with disarmament is NP-complete (even for a fixed number of rounds), if only pure strategies can be removed. On the other hand, for the case where mixed strategies can be removed, we provide a folk theorem that shows that all desirable utility profiles can be obtained, and give an efficient algorithm for (approximately) obtaining them.

We study a dynamic social choice problem in which an alternative is chosen at each round according to the reported valuations of a set of agents. In the interests of obtaining a solution that is both efficient and fair, we aim to maximize the long-term Nash social welfare, which is the product of all agents' utilities. We present and analyze two greedy algorithms for this problem, including the classic Proportional Fair (PF) algorithm. We analyze several versions of the algorithms and how they relate, and provide an axiomatization of PF. Finally, we evaluate the algorithms on data gathered from a computer systems application.

We generalize the classic problem of fairly allocating indivisible goods to the problem of fair public decision making, in which a decision must be made on several social issues simultaneously, and, unlike the classic setting, a decision can provide positive utility to multiple players. We extend the popular fairness notion of proportionality (which is not guaranteeable) to our more general setting, and introduce three novel relaxations --- proportionality up to one issue, round robin share, and pessimistic proportional share --- that are also interesting in the classic goods allocation setting. We show that the Maximum Nash Welfare solution, which is known to satisfy appealing fairness properties in the classic setting, satisfies or approximates all three relaxations in our framework. We also provide polynomial time algorithms and hardness results for finding allocations satisfying these axioms, with or without insisting on Pareto optimality.

Conventionally, the questions on a test are assumed to be kept secret from test takers until the test. However, for tests that are taken on a large scale, particularly asynchronously, this is very hard to achieve. For example, TOEFL iBT and driver's license test questions are easily found online. This also appears likely to become an issue for Massive Open Online Courses (MOOCs, as offered for example by Coursera, Udacity, and edX). Specifically, the test result may not reflect the true ability of a test taker if questions are leaked beforehand. In this paper, we take the loss of confidentiality as a fact. Even so, not all hope is lost as the test taker can memorize only a limited set of questions' answers, and the tester can randomize which questions to let appear on the test. We model this as a Stackelberg game, where the tester commits to a mixed strategy and the follower responds. Informally, the goal of the tester is to best reveal the true ability of a test taker, while the test taker tries to maximize the test result (pass probability or score). We provide an exponential-size linear program formulation that computes the optimal test strategy, prove several NP-hardness results on computing optimal test strategies in general, and give efficient algorithms for special cases (scored tests and single-question tests). Experiments are also provided for those proposed algorithms to show their scalability and the increase of the tester's utility relative to that of the uniform-at-random strategy. The increase is quite significant when questions have some correlation---for example, when a test taker who can solve a harder question can always solve easier questions.

In a landmark paper in the mechanism design literature, Cremer and McLean (1985) (CM for short) show that when a bidder’s valuation is correlated with an external signal, a monopolistic seller is able to extract the full social surplus as revenue. In the original paper and subsequent literature, the focus has been on ex-post incentive compatible (or IC) mechanisms, where truth telling is an ex-post Nash equilibrium. In this paper, we explore the implications of Bayesian versus ex-post IC in a correlated valuation setting. We generalize the full extraction result to settings that do not satisfy the assumptions of CM. In particular, we give necessary and sufficient conditions for full extraction that strictly relax the original conditions given in CM. These more general conditions characterize the situations under which requiring ex-post IC leads to a decrease in expected revenue relative to Bayesian IC. We also demonstrate that the expected revenue from the optimal ex-post IC mechanism guarantees at most a (|Θ| + 1)/4 approximation to that of a Bayesian IC mechanism, where |Θ| is the number of bidder types. Finally, using techniques from automated mechanism design, we show that, for randomly generated distributions, the average expected revenue achieved by Bayesian IC mechanisms is significantly larger than that for ex-post IC mechanisms.

In many multiagent environments, a designer has some, but limited control over the game being played. In this paper, we formalize this by considering incompletely specified games, in which some entries of the payoff matrices can be chosen from a specified set. We show that it is NP-hard for the designer to make this choices optimally, even in zero-sum games. In fact, it is already intractable to decide whether a given action is (potentially or necessarily) played in equilibrium. We also consider incompletely specified symmetric games in which all completions are required to be symmetric. Here, hardness holds even in weak tournament games (symmetric zero-sum games whose entries are all -1, 0, or 1) and in tournament games (symmetric zero-sum games whose non-diagonal entries are all -1 or 1). The latter result settles the complexity of the possible and necessary winner problems for a social-choice-theoretic solution concept known as the bipartisan set. We finally give a mixed-integer linear programming formulation for weak tournament games and evaluate it experimentally.

We study the societal tradeoffs problem, where a set of voters each submit their ideal tradeoff value between each pair of activities (e.g., "using a gallon of gasoline is as bad as creating 2 bags of landfill trash"), and these are then aggregated into the societal tradeoff vector using a rule. We introduce the family of distance-based rules and show that these can be justified as maximum likelihood estimators of the truth. Within this family, we single out the logarithmic distance-based rule as especially appealing based on a social-choice-theoretic axiomatization. We give an efficient algorithm for executing this rule as well as an approximate hill climbing algorithm, and evaluate these experimentally.

Extensive-form games constitute the standard representation scheme for games with a temporal component. But do all extensive-form games correspond to protocols that we can implement in the real world? We often rule out games with imperfect recall, which prescribe that an agent forget something that she knew before. In this paper, we show that even some games with perfect recall can be problematic to implement. Specifically, we show that if the agents have a sense of time passing (say, access to a clock), then some extensive-form games can no longer be implemented; no matter how we attempt to time the game, some information will leak to the agents that they are not supposed to have. We say such a game is not exactly timeable. We provide easy-to-check necessary and sufficient conditions for a game to be exactly timeable. Most of the technical depth of the paper concerns how to approximately time games, which we show can always be done, though it may require large amounts of time. Specifically, we show that some games require time proportional to the power tower of height proportional to the number of players, which in practice would make them untimeable. We hope to convince the reader that timeability should be a standard assumption, just as perfect recall is today. Besides the conceptual contribution to game theory, we show that timeability has implications for onion routing protocols.

The revelation principle is a key tool in mechanism design. It allows the designer to restrict attention to the class of truthful mechanisms, greatly facilitating analysis. This is also borne out in an algorithmic sense, allowing certain computational problems in mechanism design to be solved in polynomial time. Unfortunately, when not every type can misreport every other type (the partial verification model), or, more generally, misreporting can be costly, the revelation principle can fail to hold. This also leads to NP-hardness results. The primary contribution of this paper consists of characterizations of conditions under which the revelation principle still holds when misreporting can be costly. (These are generalizations of conditions given earlier for the partial verification case [Green and Laffont 1986; Yu 2011].) We also study associated computational problems.

In a classic result in the mechanism design literature, Cremerand McLean (1985) show that if buyers’ valuations are sufficiently correlated, a mechanism exists that allows the seller to extract the full surplus from efficient allocation as revenue. This result is commonly seen as “too good to be true” (in practice), casting doubt on its modeling assumptions. In this paper, we use an automated mechanism design approach to assess how sensitive the Cremer-McLean result is to relaxing its main technical assumption. That assumption implies that each valuation that a bidder can have results in a unique conditional distribution over the external signal(s). We relax this, allowing multiple valuations to be consistent with the same distribution over the external signal(s). Using similar insights to Cremer-McLean, we provide a highly efficient algorithm for computing the optimal revenue in this more general case. Using this algorithm, we observe that indeed, as the number of valuations consistent with a distribution grows, the optimal revenue quickly drops to that of a reserve-price mechanism. Thus, automated mechanism design allows us to gain insight into the precise sense in which Cremer-McLean is “too good to be true.”

We consider approval-based committee voting, i.e., the setting where each voter approves a subset of candidates, and these votes are then used to select a fixed-size set of winners (committee). We propose a natural axiom for this setting, which we call justified representation (JR). This axiom requires that if a large enough group of voters exhibits agree- ment by supporting the same candidate, then at least one voter in this group has an approved candidate in the winning committee. We show that for every list of ballots it is possible to select a committee that provides JR. We then check if this axiom is fulfilled by well-known approval-based voting rules. We show that the answer is negative for most of the rules we consider, with notable exceptions of PAV (Proportional Approval Voting), an extreme version of RAV (Reweighted Approval Voting), and, for a restricted preference domain, MAV (Minimax Approval Voting). We then introduce a stronger version of the JR axiom, which we call extended justified representation (EJR), and show that PAV satisfies EJR, while other rules do not. We also consider several other questions related to JR and EJR, including the relationship between JR/EJR and unanimity, and the complexity of the associated algorithmic problems.

Models of strategic candidacy analyze the incentives of candidates to run in an election. Most work on this topic assumes that strategizing only takes place among candidates, whereas voters vote truthfully. In this paper, we extend the analysis to also include strategic behavior on the part of the voters. (We also study cases where only candidates or only voters are strategic.) We consider two settings in which strategic voting is well-defined and has a natural interpretation: majority-consistent voting with single-peaked preferences and voting by successive elimination. In the former setting, we analyze the type of strategic behavior required in order to guarantee desirable voting outcomes. In the latter setting, we determine the complexity of computing the set of potential outcomes if both candidates and voters act strategically.

In cooperative game theory, it is typically assumed that the value of each coalition is known. We depart from this, assuming that v(S) is only a noisy estimate of the true value V (S), which is not yet known. In this context, we investigate which solution concepts maximize the probability of ex-post stability (after the true values are revealed). We show how various conditions on the noise characterize the least core and the nucleolus as optimal. Modifying some aspects of these conditions to (arguably) make them more realistic, we obtain characterizations of new solution concepts as being optimal, including the partial nucleolus, the multiplicative least core, and the multiplicative nucleolus.

We study the problem where a task (or multiple unrelated tasks) must be executed, there are multiple machines/agents that can potentially perform the task, and our objective is to minimize the expected sum of the agents' processing times. Each agent does not know exactly how long it will take him to finish the task; he only knows the distribution from which this time is drawn. These times are independent across agents and the distributions fulfill the monotone hazard rate condition. Agents are selfish and will lie about their distributions if this increases their expected utility. We study different variations of the Vickrey mechanism that take as input the agents' reported distributions and the players' realized running times and that output a schedule that minimizes the expected sum of processing times, as well as payments that make it an ex-post equilibrium for the agents to both truthfully report their distributions and exert full effort to complete the task. We devise the ChPE mechanism, which is uniquely tailored to our problem, and has many desirable properties including: not rewarding agents that fail to finish the task and having non-negative payments.

Runoff voting rules such as single transferable vote (STV) and Baldwin's rule are of particular interest in computational social choice due to their recursive nature and hardness of manipulation, as well as in (human) practice because they are relatively easy to understand. However, they are not known for their compliance with desirable axiomatic properties, which we attempt to rectify here. We characterize runoff rules that are based on scoring rules using two axioms: a weakening of local independence of irrelevant alternatives and a variant of population-consistency. We then show, as our main technical result, that STV is the only runoff scoring rule satisfying an independence-of-clones property. Furthermore, we provide axiomatizations of Baldwin's rule and Coombs' rule.

Gambles in casinos are usually set up so that the casino makes a profit in expectation -- as long as gamblers play honestly. However, some gamblers are able to cheat, reducing the casino’s profit. How should the casino address this? A common strategy is to selectively kick gamblers out, possibly even without being sure that they were cheating. In this paper, we address the following question: Based solely on a gambler’s track record,when is it optimal for the casino to kick the gambler out? Because cheaters will adapt to the casino’s policy, this is a game-theoretic question. Specifically, we model the problem as a Bayesian game in which the casino is a Stackelberg leader that can commit to a (possibly randomized) policy for when to kick gamblers out, and we provide efficient algorithms for computing the optimal policy. Besides being potentially useful to casinos, we imagine that similar techniques could be useful for addressing related problems -- for example, illegal trades in financial markets.

Among the many deployment areas of Stackelberg Security games, a major area involves games played out in space and time, which includes applications in multiple mobile defender resources protecting multiple mobile targets. Previous algorithms for such spatio-temporal security games fail to scale-up and little is known ofthe computational complexity properties of these problems.This paper provides a novel oracle-based algorithmic framework for a systematic study of different problem variants of computing optimal (minimax) strategies in spatio-temporal security games. Our framework enables efficient computation of a minimax strategy when the problem admits a polynomial-time oracle. Furthermore,for the cases in which efficient oracles are difficultto find, we propose approximations or prove hardness results.

It is known that an equilibrium of an infinitely repeated two-player game (with limit average payoffs) can be computed in polynomial time, as follows: according to the folk theorem, we compute minimax strategies for both players to calculate the punishment values, and subsequently find a mixture over outcomes that exceeds these punishment values. However, for very large games, even computing minimax strategies can be prohibitive. In this paper, we propose an algorithmic framework for computing equilibria of repeated games that does not require linear programming and that does not necessarily need to inspect all payoffs of the game. This algorithm necessarily sometimes fails to compute an equilibrium, but we mathematically demonstrate that most of the time it succeeds quickly on uniformly random games, and experimentally demonstrate this for other classes of games. This also holds for games with more than two players, for which no efficient general algorithms are known.

Security games involving the allocation of multiple security resources to defend multiple targets generally have an exponential number of pure strategies for the defender. One method that has been successful in addressing this computational issue is to instead directly compute the marginal probabilities with which the individual resources are assigned (first pursued by Kiekintveld et al. (2009)). However, in sufficiently general settings, there exist games where these marginal solutions are not implementable, that is, they do not correspond to any mixed strategy of the defender. In this paper, we examine security games where the defender tries to monitor the vertices of a graph, and we show how the type of graph, the type of schedules, and the type of defender resources affect the applicability of this approach. In some settings, we show the approach is applicable and give a polynomial-time algorithm for computing an optimal defender strategy; in other settings, we give counterexample games that demonstrate that the approach does not work, and prove NP-hardness results for computing an optimal defender strategy.

The family of Groves mechanisms, which includes the well-known VCG mechanism (also known as the Clarke mechanism), is a family of efficient and strategy-proof mechanisms. Unfortunately, the Groves mechanisms are generally not budget balanced. That is, under such mechanisms, payments may flow into or out of the system of the agents, resulting in deficits or reduced utilities for the agents. We consider the following problem: within the family of Groves mechanisms, we want to identify mechanisms that give the agents the highest utilities, under the constraint that these mechanisms must never incur deficits. We adopt a prior-free approach. We introduce two general measures for comparing mechanisms in prior-free settings. We say that a non-deficit Groves mechanism M individually dominates another non-deficit Groves mechanism M' if for every type profile, every agent's utility under M is no less than that under M', and this holds with strict inequality for at least one type profile and one agent. We say that a non-deficit Groves mechanism M collectively dominates another non-deficit Groves mechanism M' if for every type profile, the agents' total utility under M is no less than that under M', and this holds with strict inequality for at least one type profile. The above definitions induce two partial orders on non-deficit Groves mechanisms. We study the maximal elements corresponding to these two partial orders, which we call the individually undominated mechanisms and the collectively undominated mechanisms, respectively.

The multiagent systems community has adopted game theory as a framework for the design of systems of multiple self-interested agents. For this to be effective, efficient algorithms must be designed to compute the solutions that game theory prescribes. In this paper, I summarize some of the state of the art on this topic, focusing particularly on how this line of work has contributed to several highly visible deployed security applications, developed at the University of Southern California.

Significant progress has been made recently in the following two lines of research in the intersection of AI and game theory: (1) the computation of optimal strategies to commit to (Stackelberg strategies), and (2) the computation of correlated equilibria of stochastic games. In this paper, we unite these two lines of research by studying the computation of Stackelberg strategies in stochastic games. We provide theoretical results on the value of being able to commit and the value of being able to correlate, as well as complexity results about computing Stackelberg strategies in stochastic games. We then modify the QPACE algorithm (MacDermed et al. 2011) to compute Stackelberg strategies, and provide experimental results.

In many mechanisms (especially online mechanisms), a strategic agent can influence the outcome by creating multiple false identities. We consider voting settings where the mechanism designer cannot completely prevent false-name manipulation, but may use false-name-limiting methods such as CAPTCHAs to influence the amount and characteristics of such manipulation. Such a designer would prefer, first, a high probability of obtaining the “correct” outcome, and second, a statistical method for evaluating the correctness of the outcome. In this paper, we focus on settings with two alternatives. We model voters as independently drawing a number of identities from a distribution that may be influenced by the choice of the false-name-limiting method. We give a criterion for the evaluation and comparison of these distributions. Then, given the results of an election in which false-name manipulation may have occurred, we propose and justify a statistical test for evaluating the outcome.

The standard approach to computing an optimal mixed strategy to commit to is based on solving a set of linear programs, one for each of the follower's pure strategies. We show that these linear programs can be naturally merged into a single linear program; that this linear program can be interpreted as a formulation for the optimal correlated strategy to commit to, giving an easy proof of a result by von Stengel and Zamir that the leader's utility is at least the utility she gets in any correlated equilibrium of the simultaneous-move game; and that this linear program can be extended to compute optimal correlated strategies to commit to in games of three or more players. (Unlike in two-player games, in games of three or more players, the notions of optimal mixed and correlated strategies to commit to are truly distinct.) We give examples, and provide experimental results that indicate that for 50x50 games, this approach is usually significantly faster than the multiple-LPs approach.

We consider manipulation problems when the manipulator only has partial information about the votes of the non-manipulators. Such partial information is described by an {\em information set}, which is the set of profiles of the non-manipulators that are indistinguishable to the manipulator. Given such an information set, a {\em dominating manipulation} is a non-truthful vote that the manipulator can cast which makes the winner at least as preferable (and sometimes more preferable) as the winner when the manipulator votes truthfully. When the manipulator has full information, computing whether or not there exists a dominating manipulation is in P for many common voting rules (by known results). We show that when the manipulator has no information, there is no dominating manipulation for many common voting rules. When the manipulator's information is represented by partial orders and only a small portion of the preferences are unknown, computing a dominating manipulation is NP-hard for many common voting rules. Our results thus throw light on whether we can prevent strategic behavior by limiting information about the votes of other voters.

Algorithms for finding game-theoretic solutions are now used in several real-world security applications. This work has generally assumed a Stackelberg model where the defender commits to a mixed strategy first. In general two-player normal-form games, Stackelberg strategies are easier to compute than Nash equilibria, though it has recently been shown that in many security games, Stackelberg strategies are also Nash strategies for the defender. However, the work on security games so far assumes that the attacker attacks only a single target. In this paper, we generalize to the case where the attacker attacks multiple targets simultaneously. Here, Stackelberg and Nash strategies for the defender can be truly different. We provide a polynomial-time algorithm for finding a Nash equilibrium. The algorithm gradually increases the number of defender resources and maintains an equilibrium throughout this process. Moreover, we prove that Nash equilibria in security games with multiple attackers satisfy the interchange property, which resolves the problem of equilibrium selection in such games. On the other hand, we show that Stackelberg strategies are actually NP-hard to compute in this context. Finally, we provide experimental results. 1

In many settings, a group of voters must come to a joint decision on multiple issues. In practice, this is often done by voting on the issues in sequence. We model sequential voting in multi-issue domains as a complete-information extensive-form game, in which the voters are perfectly rational and their preferences are common knowledge. In each step, the voters simultaneously vote on one issue, and the order of the issues is given exogenously before the process. We call this model strategic sequential voting.

There has been significant recent interest in game-theoretic approaches to security, with much of the recent research focused on utilizing the leader-follower Stackelberg game model. Among the major applications are the ARMOR program deployed at LAX Airport and the IRIS program in use by the US Federal Air Marshals (FAMS). The foundational assumption for using Stackelberg games is that security forces (leaders), acting first, commit to a randomized strategy; while their adversaries (followers) choose their best response after surveillance of this randomized strategy. Yet, in many situations, a leader may face uncertainty about the follower’s surveillance capability. Previous work fails to address how a leader should compute her strategy given such uncertainty. We provide five contributions in the context of a general class of security games. First, we show that the Nash equilibria in security games are interchangeable, thus alleviating the equilibrium selection problem. Second, under a natural restriction on security games, any Stackelberg strategy is also a Nash equilibrium strategy; and furthermore, the solution is unique in a class of security games of which ARMOR is a key exemplar. Third, when faced with a follower that can attack multiple targets, many of these properties no longer hold. Fourth, we show experimentally that in most (but not all) games where the restriction does not hold, the Stackelberg strategy is still a Nash equilibrium strategy, but this is no longer true when the attacker can attack multiple targets. Finally, as a possible direction for future research, we propose an extensive-form game model that makes the defender’s uncertainty about the attacker’s ability to observe explicit.

Usually a voting rule requires agents to give their preferences as linear orders. However, in some cases it is impractical for an agent to give a linear order over all the alternatives. It has been suggested to let agents submit partial orders instead. Then, given a voting rule, a profile of partial orders, and an alternative (candidate) c, two important questions arise: first, is it still possible for c to win, and second, is c guaranteed to win? These are the possible winner and necessary winner problems, respectively. Each of these two problems is further divided into two sub-problems: determining whether c is a unique winner (that is, c is the only winner), or determining whether c is a co-winner (that is, c is in the set of winners). We consider the setting where the number of alternatives is unbounded and the votes are unweighted. We completely characterize the complexity of possible/necessary winner problems for the following common voting rules: a class of positional scoring rules (including Borda), Copeland, maximin, Bucklin, ranked pairs, voting trees, and plurality with runoff.

In many multiagent settings, a decision must be made based on the preferences of multiple agents, and agents may lie about their preferences if this is to their benefit. In mechanism design, the goal is to design procedures (mechanisms) for making the decision that work in spite of such strategic behavior, usually by making untruthful behavior suboptimal. In automated mechanism design, the idea is to computationally search through the space of feasible mechanisms, rather than to design them analytically by hand. Unfortunately, the most straightforward approach to automated mechanism design does not scale to large instances, because it requires searching over a very large space of possible functions. In this paper, we describe an approach to automated mechanism design that is computationally feasible. Instead of optimizing over all feasible mechanisms, we carefully choose a parameterized subfamily of mechanisms. Then we optimize over mechanisms within this family, and analyze whether and to what extent the resulting mechanism is suboptimal outside the subfamily. We demonstrate the usefulness of our approach with two case studies.

Recently, algorithms for computing game-theoretic solutions have been deployed in real-world security applications, such as the placement of checkpoints and canine units at Los Angeles International Airport. These algorithms assume that the defender (security personnel) can commit to a mixed strategy, a so-called Stackelberg model. As pointed out by Kiekintveld et al. (2009), in these applications, generally, multiple resources need to be assigned to multiple targets, resulting in an exponential number of pure strategies for the defender. In this paper, we study how to compute optimal Stackelberg strategies in such games, showing that this can be done in polynomial time in some cases, and is NP-hard in others.

In computational social choice, one important problem is to take the votes of a subelectorate (subset of the voters), and summarize them using a small number of bits. This needs to be done in such a way that, if all that we know is the summary, as well as the votes of voters outside the subelectorate, we can conclude which of the m alternatives wins. This corresponds to the notion of compilation complexity, the minimum number of bits required to summarize the votes for a particular rule, which was introduced by Chevaleyre et al. [IJCAI-09]. We study three different types of compilation complexity. The first, studied by Chevaleyre et al., depends on the size of the subelectorate but not on the size of the complement (the voters outside the subelectorate). The second depends on the size of the complement but not on the size of the subelectorate. The third depends on both. We first investigate the relations among the three types of compilation complexity. Then, we give upper and lower bounds on all three types of compilation complexity for the most prominent voting rules. We show that for l-approval (when l ≤ m/2), Borda, and Bucklin, the bounds for all three types are asymptotically tight, up to a multiplicative constant; for l-approval (when l > m/2), plurality with runoff, all Condorcet consistent rules that are based on unweighted majority graphs (including Copeland and voting trees), and all Condorcet consistent rules that are based on the order of pairwise elections (including ranked pairs and maximin), the bounds for all three types are asymptotically tight up to a multiplicative constant when the sizes of the subelectorate and its complement are both larger than m1+ε for some ε > 0.

We consider settings in which voters vote in sequence, each voter knows the votes of the earlier voters and the preferences of the later voters, and voters are strategic. This can be modeled as an extensive-form game of perfect information, which we call a Stackelberg voting game. We first propose a dynamic-programming algorithm for finding the backward-induction outcome for any Stackelberg voting game when the rule is anonymous; this algorithm is efficient if the number of alternatives is no more than a constant. We show how to use compilation functions to further reduce the time and space requirements. Our main theoretical results are paradoxes for the backward-induction outcomes of Stackelberg voting games. We show that for any n ≥ 5 and any voting rule that satisfies nonimposition and with a low domination index, there exists a profile consisting of n voters, such that the backward-induction outcome is ranked somewhere in the bottom two positions in almost every voter’s preferences. Moreover, this outcome loses all but one of its pairwise elections. Furthermore, we show that many common voting rules have a very low (= 1) domination index, including all majority-consistent voting rules. For the plurality and nomination rules, we show even stronger paradoxes. Finally, using our dynamic-programming algorithm, we run simulations to compare the backward-induction outcome of the Stackelberg voting game to the winner when voters vote truthfully, for the plurality and veto rules. Surprisingly, our experimental results suggest that on average, more voters prefer the backward-induction outcome.

Computing optimal strategies to commit to in general normal-form or Bayesian games is a topic that has recently been gaining attention, in part due to the application of such algorithms in various security and law enforcement scenarios. In this paper, we extend this line of work to the more general case of commitment in extensive-form games. We show that in some cases, the optimal strategy can be computed in polynomial time; in others, computing it is NP-hard.

The coalitional manipulation problem is one of the central problems in computational social choice. In this paper, we focus on solving the problem under the important family of positional scoring rules, in an approximate sense that was advocated by Zuckerman et al. [SODA 2008, AIJ 2009]. Our main result is a polynomial-time algorithm with (roughly speaking) the following theoretical guarantee: given a manipulable instance with m alternatives, the algorithm finds a successful manipulation with at most m - 2 additional manipulators. Our technique is based on a reduction to the scheduling problem known as Q|pmtn|Cmax, along with a novel rounding procedure. We demonstrate that our analysis is tight by establishing a new type of integrality gap. We also resolve a known open question in computational social choice by showing that the coalitional manipulation problem remains (strongly) NP-complete for positional scoring rules even when votes are unweighted. Finally, we discuss the implications of our results with respect to the question: "Is there a prominent voting rule that is usually hard to manipulate?"

In this paper, we consider the problem of designing incentive compatible auctions for multiple (homogeneous) units of a good, when bidders have private valuations and private budget constraints. When only the valuations are private and the budgets are public, Dobzinski et al [8] show that the adaptive clinching auction is the unique incentive-compatible auction achieving Pareto-optimality. They further show that this auction is not truthful with private budgets, so that there is no deterministic Pareto-optimal auction with private budgets. Our main contribution is to show the following Budget Monotonicity property of this auction: When there is only one infinitely divisible good, a bidder cannot improve her utility by reporting a budget smaller than the truth. This implies that the adaptive clinching auction is incentive compatible when over-reporting the budget is not possible (for instance, when funds must be shown upfront). We can also make reporting larger budgets suboptimal with a small randomized modification to the auction. In either case, this makes the modified auction Pareto-optimal with private budgets. We also show that the Budget Monotonicity property does not hold for auctioning indivisible units of the good, showing a sharp contrast between the divisible and indivisible cases. The Budget Monotonicity property also implies other improved results in this context. For revenue maximization, the same auction improves the best-known competitive ratio due to Abrams [1] by a factor of 4, and asymptotically approaches the performance of the optimal single-price auction. Finally, we consider the problem of revenue maximization (or social welfare) in a Bayesian setting. We allow the bidders have public size constraints (on the amount of good they are willing to buy) in addition to private budget constraints. We show a simple poly-time computable 5.83-approximation to the optimal Bayesian incentive compatible mechanism, that is implementable in dominant strategies. Our technique again crucially needs the ability to prevent bidders from over-reporting budgets via randomization. We show the approximation result via designing a rounding scheme for an LP relaxation of the problem, which may be of independent interest.

Restricting the preferences of the agents by assuming that their utility functions linearly depend on a payment allows for the positive results of the Vickrey auction and the Vickrey-Clarke-Groves mechanism. These results, however, are limited to settings where there is some commonly desired commodity or numeraire--money, shells, beads, etcetera--which is commensurable with utility. We propose a generalization of the Vickrey auction that does not assume that the agents' preferences are quasilinear, but nevertheless retains some of the Vickrey auction's desirable properties. In this auction, a bid can be any alternative, rather than just a monetary offer. As a consequence, the auction is also applicable to situations where there is a fixed budget, or no numeraire is available at all (or it is undesirable to use payments for other reasons)--such as, for example, in the allocation of the task of contributing a module to an open-source project. We show that in two general settings, this qualitative Vickrey auction has a dominant-strategy equilibrium, invariably yields a weakly Pareto efficient outcome in this equilibrium, and is individually rational. In the first setting, the center has a linear preference order over a finite set of alternatives, and in the second setting, the bidders' preferences can be represented by continuous utility functions over a closed metric space of alternatives and the center's utility is equipeaked. The traditional Vickrey auction turns out to be a special case of the qualitative Vickrey auction in this second setting.

Voting is a general method for aggregating the preferences of multiple agents. Each agent ranks all the possible alternatives, and based on this, an aggregate ranking of the alternatives (or at least a winning alternative) is produced. However, when there are many alternatives, it is impractical to simply ask agents to report their complete preferences. Rather, the agents' preferences, or at least the relevant parts thereof, need to be elicited. This is done by asking the agents a (hopefully small) number of simple queries about their preferences, such as comparison queries, which ask an agent to compare two of the alternatives. Prior work on preference elicitation in voting has focused on the case of unrestricted preferences. It has been shown that in this setting, it is sometimes necessary to ask each agent (almost) as many queries as would be required to determine an arbitrary ranking of the alternatives. In contrast, in this paper, we focus on single-peaked preferences. We show that such preferences can be elicited using only a linear number of comparison queries, if either the order with respect to which preferences are single-peaked is known, or at least one other agent's complete preferences are known. We show that using a sublinear number of queries does not suffice. We also consider the case of cardinally single-peaked preferences. For this case, we show that if the alternatives' cardinal positions are known, then an agent's preferences can be elicited using only a logarithmic number of queries; however, we also show that if the cardinal positions are not known, then a sublinear number of queries does not suffice. We present experimental results for all elicitation algorithms. We also consider the problem of only eliciting enough information to determine the aggregate ranking, and show that even for this more modest objective, a sublinear number of queries per agent does not suffice for known ordinal or unknown cardinal positions. Finally, we discuss whether and how these techniques can be applied when preferences are almost single-peaked.

For the problem of allocating one or more items among a group of competing agents, the Vickrey-Clarke-Groves (VCG) mechanism is strategy-proof and efficient. However, the VCG mechanism is not strongly budget balanced: in general, value flows out of the system of agents in the form of VCG payments, which reduces the agents' utilities. In many settings, the objective is to maximize the sum of the agents' utilities (taking payments into account). For this purpose, several VCG redistribution mechanisms have been proposed that redistribute a large fraction of the VCG payments back to the agents, in a way that maintains strategy-proofness and the non-deficit property. Unfortunately, sometimes even the best VCG redistribution mechanism fails to redistribute a substantial fraction of the VCG payments. This results in a low total utility for the agents, even though the items are allocated efficiently. In this paper, we study strategy-proof allocation mechanisms that do not always allocate the items efficiently. It turns out that by allocating inefficiently, more payment can sometimes be redistributed, so that the net effect is an increase in the sum of the agents' utilities.

The Gibbard-Satterthwaite Theorem states that (in unrestricted settings) any reasonable voting rule is manipulable. Recently, a quantitative version of this theorem was proved by Ehud Friedgut, Gil Kalai, and Noam Nisan: when the number of alternatives is three, for any neutral voting rule that is far from any dictatorship, there exists a voter such that a random manipulation---that is, the true preferences and the strategic vote are all drawn i.i.d., uniformly at random---will succeed with a probability of Ω(1/n), where n is the number of voters. However, it seems that the techniques used to prove this theorem can not be fully extended to more than three alternatives. In this paper, we give a more limited result that does apply to four or more alternatives. We give a sufficient condition for a voting rule to be randomly manipulable with a probability of Ω(1/n) for at least one voter, when the number of alternatives is held fixed. Specifically, our theorem states that if a voting rule r satisfies 1. homogeneity, 2. anonymity, 3. non-imposition, 4. a canceling-out condition, and 5. there exists a stable profile that is still stable after one given alternative is uniformly moved to different positions; then there exists a voter such that a random manipulation for that voter will succeed with a probability of Ω(1/n). We show that many common voting rules satisfy these conditions, for example any positional scoring rule, Copeland, STV, maximin, and ranked pairs.

We introduce a class of voting rules called generalized scoring rules. Under such a rule, each vote generates a vector of k scores, and the outcome of the voting rule is based only on the sum of these vectors---more specifically, only on the order (in terms of score) of the sum's components. This class is extremely general: we do not know of any commonly studied rule that is not a generalized scoring rule.

For allocation problems with one or more items, the well-known Vickrey-Clarke-Groves (VCG) mechanism is efficient, strategy-proof, individually rational, and does not incur a deficit. However, the VCG mechanism is not (strongly) budget balanced: generally, the agents' payments will sum to more than 0. If there is an auctioneer who is selling the items, this may be desirable, because the surplus payment corresponds to revenue for the auctioneer. However, if the items do not have an owner and the agents are merely interested in allocating the items efficiently among themselves, any surplus payment is undesirable, because it will have to flow out of the system of agents. In 2006, Cavallo [3] proposed a mechanism that redistributes some of the VCG payment back to the agents, while maintaining efficiency, strategy-proofness, individual rationality, and the non-deficit property. In this paper, we extend this result in a restricted setting. We study allocation settings where there are multiple indistinguishable units of a single good, and agents have unit demand. (For this specific setting, Cavallo's mechanism coincides with a mechanism proposed by Bailey in 1997 [2].) Here we propose a family of mechanisms that redistribute some of the VCG payment back to the agents. All mechanisms in the family are efficient, strategy-proof, individually rational, and never incur a deficit. The family includes the Bailey-Cavallo mechanism as a special case. We then provide an optimization model for finding the optimal mechanism--that is, the mechanism that maximizes redistribution in the worst case--inside the family, and show how to cast this model as a linear program. We give both numerical and analytical solutions of this linear program, and the (unique) resulting mechanism shows significant improvement over the Bailey-Cavallo mechanism (in the worst case). Finally, we prove that the obtained mechanism is optimal among all anonymous deterministic mechanisms that satisfy the above properties.

In open, anonymous environments such as the Internet, mechanism design is complicated by the fact that a single agent can participate in the mechanism under multiple identifiers. One way to address this is to design false-name-proof mechanisms, which choose the outcome in such a way that agents have no incentive to use more than one identifier. Unfortunately, there are inherent limitations on what can be achieved with false-name-proof mechanisms, and at least in some cases, these limitations are crippling. An alternative approach is to verify the identities of all agents. This imposes significant overhead and removes any benefits from anonymity.

In multiagent settings where the agents have different preferences, preference aggregation is a central issue. Voting is a general method for preference aggregation, but seminal results have shown that all general voting protocols are manipulable. One could try to avoid manipulation by using protocols where determining a beneficial manipulation is hard. Especially among computational agents, it is reasonable to measure this hardness by computational complexity. Some earlier work has been done in this area, but it was assumed that the number of voters and candidates is unbounded. Such hardness results lose relevance when the number of candidates is small, because manipulation algorithms that are exponential only in the number of candidates (and only slightly so) might be available. We give such an algorithm for an individual agent to manipulate the Single Transferable Vote (STV) protocol, which has been shown hard to manipulate in the above sense. This motivates the core of this article, which derives hardness results for realistic elections where the number of candidates is a small constant (but the number of voters can be large). The main manipulation question we study is that of coalitional manipulation by weighted voters. (We show that for simpler manipulation problems, manipulation cannot be hard with few candidates.) We study both constructive manipulation (making a given candidate win) and destructive manipulation (making a given candidate not win). We characterize the exact number of candidates for which manipulation becomes hard for the plurality , Borda , STV , Copeland , maximin , veto , plurality with runoff , regular cup , and randomized cup protocols. We also show that hardness of manipulation in this setting implies hardness of manipulation by an individual in unweighted settings when there is uncertainty about the others' votes (but not vice-versa). To our knowledge, these are the first results on the hardness of manipulation when there is uncertainty about the others' votes.

In a principal-agent problem, a principal seeks to motivate an agent to take a certain action beneficial to the principal, while spending as little as possible on the reward. This is complicated by the fact that the principal does not know the agent's utility function (or type). We study the online setting where at each round, the principal encounters a new agent, and the principal sets the rewards anew. At the end of each round, the principal only finds out the action that the agent took, but not his type. The principal must learn how to set the rewards optimally. We show that this setting generalizes the setting of selling a digital good online.We study and experimentally compare three main approaches to this problem. First, we show how to apply a standard bandit algorithm to this setting. Second, for the case where the distribution of agent types is fixed (but unknown to the principal), we introduce a new gradient ascent algorithm. Third, for the case where the distribution of agents' types is fixed, and the principal has a prior belief (distribution) over a limited class of type distributions, we study a Bayesian approach.

In multiagent systems, strategic settings are often analyzed under the assumption that the players choose their strategies simultaneously. However, this model is not always realistic. In many settings, one player is able to commit to a strategy before the other player makes a decision. Such models are synonymously referred to as leadership, commitment, or Stackelberg models, and optimal play in such models is often significantly different from optimal play in the model where strategies are selected simultaneously.The recent surge in interest in computing game-theoretic solutions has so far ignored leadership models (with the exception of the interest in mechanism design, where the designer is implicitly in a leadership position). In this paper, we study how to compute optimal strategies to commit to under both commitment to pure strategies and commitment to mixed strategies, in both normal-form and Bayesian games. We give both positive results (efficient algorithms) and negative results (NP-hardness results).

We determine the communication complexity of the common voting rules. The rules (sorted by their communication complexity from low to high) are plurality, plurality with runoff, single transferable vote (STV), Condorcet, approval, Bucklin, cup, maximin, Borda, Copeland, and ranked pairs. For each rule, we first give a deterministic communication protocol and an upper bound on the number of bits communicated in it; then, we give a lower bound on (even the nondeterministic) communication requirements of the voting rule. The bounds match for all voting rules except STV and maximin.

We study various computational aspects of solving games using dominance and iterated dominance. We first study both strict and weak dominance (not iterated), and show that checking whether a given strategy is dominated by some mixed strategy can be done in polynomial time using a single linear program solve. We then move on to iterated dominance. We show that determining whether there is some path that eliminates a given strategy is NP-complete with iterated weak dominance. This allows us to also show that determining whether there is a path that leads to a unique solution is NP-complete. Both of these results hold both with and without dominance by mixed strategies. (A weaker version of the second result (only without dominance by mixed strategies) was already known [7].) Iterated strict dominance, on the other hand, is path-independent (both with and without dominance by mixed strategies) and can therefore be done in polynomial time.We then study what happens when the dominating strategy is allowed to place positive probability on only a few pure strategies. First, we show that finding the dominating strategy with minimum support size is NP-complete (both for strict and weak dominance). Then, we show that iterated strict dominance becomes path-dependent when there is a limit on the support size of the dominating strategies, and that deciding whether a given strategy can be eliminated by iterated strict dominance under this restriction is NP-complete (even when the limit on the support size is 3).Finally, we study Bayesian games. We show that, unlike in normal form games, deciding whether a given pure strategy is dominated by another pure strategy in a Bayesian game is NP-complete (both with strict and weak dominance); however, deciding whether a strategy is dominated by some mixed strategy can still be done in polynomial time with a single linear program solve (both with strict and weak dominance). Finally, we show that iterated dominance using pure strategies can require an exponential number of iterations in a Bayesian game (both with strict and weak dominance).

A fast-growing body of research in the AI and machine learning communities addresses learning in games, where there are multiple learners with different interests. This research adds to more established research on learning in games conducted in economics. In part because of a clash of fields, there are widely varying requirements on learning algorithms in this domain. The goal of this paper is to demonstrate how communication complexity can be used as a lower bound on the required learning time or cost. Because this lower bound does not assume any requirements on the learning algorithm, it is universal, applying under any set of requirements on the learning algorithm.We characterize exactly the communication complexity of various solution concepts from game theory, namely Nash equilibrium, iterated dominant strategies (both strict and weak), and backwards induction. This gives the tighest lower bounds on learning in games that can be obtained with this method.

When donating money to a (say, charitable) cause, it is possible touse the contemplated donation as negotiating material to induce other parties interested in the charity to donate more. Such negotiation is usually done in terms of matching offers, where one party promises to pay a certain amount if others pay a certain amount. However, in their current form, matching offers allow for only limited negotiation. For one, it is not immediately clear how multiple parties can make matching offers at the same time without creating circular dependencies. Also, it is not immediately clear how to make adonation conditional on other donations to multiple charities, when the donator has different levels of appreciation for the different charities. In both these cases, the limited expressiveness of matching offers causes economic loss: it may happen that an arrangement that would have made all parties (donators as well as charities) better off cannot be expressed in terms of matching offers and will therefore notoccur.In this paper, we introduce a bidding language for expressing very general types of matching offers over multiple charities. We formulate the corresponding clearing problem (deciding how much each bidder pays, and how much each charity receives), and show that it is NP-complete to approximate to any ratio even in very restricted settings. We givea mixed-integer program formulation of the clearing problem, and show that for concave bids, the program reduces to a linear program. We then show that the clearing problem for a subclass of concave bids is at least as hard as the decision variant of linear programming. Subsequently, we show that the clearing problem is much easier when bids are quasilinear---for surplus, the problem decomposes across charities, and for payment maximization, a greedy approach isoptimal if the bids are concave (although this latter problem is weakly NP-complete when the bids are not concave). For the quasilinear setting, we study the mechanism design question. We show that anex-post efficient mechanism is impossible even with only one charity and a very restricted class of bids. We also show that there may bebenefits to linking the charities from a mechanism design stand point.

Often, an outcome must be chosen on the basis of the preferences reported by a group of agents. The key difficulty is that the agents may report their preferences insincerely to make the chosen outcome more favorable to themselves. Mechanism design is the art of designing the rules of the game so that the agents are motivated to report their preferences truthfully, and a desirable outcome is chosen. In a recently proposed approach---called automated mechanism design ---a mechanism is computed for the preference aggregation setting at hand. This has several advantages, but the downside is that the mechanism design optimization problem needs to be solved anew each time. Unlike the earlier work on automated mechanism design that studied a benevolent designer, in this paper we study automated mechanism design problems where the designer is self-interested. In this case, the center cares only about which outcome is chosen and what payments are made to it. The reason that the agents' preferences are relevant is that the center is constrained to making each agent at least as well off as the agent would have been had it not participated in the mechanism. In this setting, we show that designing optimal deterministic mechanisms is NP-complete in two important special cases: when the center is interested only in the payments made to it, and when payments are not possible and the center is interested only in the outcome chosen. We then show how allowing for randomization in the mechanism makes problems in this setting computationally easy. Finally, we show that the payment-maximizing AMD problem is closely related to an interesting variant of the optimal (revenue-maximizing) combinatorial auction design problem, where the bidders have "best-only" preferences. We show that here, too, designing an optimal deterministic auction is NP-complete, but designing an optimal randomized auction is easy.

No abstract available.

No abstract available.

The aggregation of conflicting preferences is a central problem in multiagent systems. The key difficulty is that the agents may report their preferences insincerely. Mechanism design is the art of designing the rules of the game so that the agents are motivated to report their preferences truthfully and a (socially) desirable outcome is chosen. We propose an approach where a mechanism is automatically created for the preference aggregation setting at hand. This has several advantages, but the downside is that the mechanism design optimization problem needs to be solved anew each time. Hence the computational complexity of mechanism design becomes a key issue. In this paper we analyze the single-agent mechanism design problem, whose simplicity allows for elegant and generally applicable results.We show that designing an optimal deterministic mechanism that does not use payments is NP-complete even if there is only one agent whose type is private information---even when the designer's objective is social welfare. We show how this hardness result extends to settings with multiple agents with private information. We then show that if the mechanism is allowed to use randomization, the design problem is solvable by linear programming (even for general objectives) and hence in P. This generalizes to any fixed number of agents. We then study settings where side payments are possible and the agents' preferences are quasilinear. We show that if the designer's objective is social welfare, an optimal deterministic mechanism is easy to construct; in fact, this mechanism is also ex post optimal. We then show that designing an optimal deterministic mechanism with side payments is NP-complete for general objectives, and this hardness extends to settings with multiple agents. Finally, we show that an optimal randomized mechanism can be designed in polynomial time using linear programming even for general objective functions. This again generalizes to any fixed number of agents.

Coalition formation is a key problem in automated negotiation among self-interested agents, and other electronic commerce applications. A coalition of agents can sometimes accomplish things that the individual agents cannot, or can do things more efficiently. However, motivating the agents to abide to a solution requires careful analysis: only some of the solutions are stable in the sense that no group of agents is motivated to break off and form a new coalition. This constraint has been studied extensively in cooperative game theory. However, the computational questions around this constraint have received less attention. When it comes to coalition formation among software agents (that represent real-world parties), these questions become increasingly explicit.In this paper we define a concise general representation for games in characteristic form that relies on superadditivity, and show that it allows for efficient checking of whether a given outcome is in the core. We then show that determining whether the core is nonempty is NP-complete both with and without transferable utility. We demonstrate that what makes the problem hard in both cases is determining the collaborative possibilities (the set of outcomes possible for the grand coalition), by showing that if these are given, the problem becomes tractable in both cases. However, we then demonstrate that for a hybrid version of the problem, where utility transfer is possible only within the grand coalition, the problem remains NP-complete even when the collaborative possibilities are given.

Often, an outcome must be chosen on the basis of the preferences reported by a group of agents. The key difficulty is that the agents may report their preferences insincerely to make the chosen outcome more favorable to themselves. Mechanism design is the art of designing the rules of the game so that the agents are motivated to report their preferences truthfully, and a desirable outcome is chosen. We recently proposed an approach---called automated mechanism design---where a mechanism is computed for the preference aggregation setting at hand. This has several advantages, but the downside is that the mechanism design optimization problem needs to be solved anew each time. Unlike the earlier work on automated mechanism design that studied a benevolent designer, in this paper we study automated mechanism design problems where the designer is self-interested. In this case, the center cares only about which outcome is chosen and what payments are made to it. The reason that the agents' preferences are relevant is that the center is constrained to making each agent at least as well off as the agent would have been had it not participated in the mechanism. In this setting, we show that designing optimal deterministic mechanisms is NP-complete in two important special cases: when the center is interested only in the payments made to it, and when payments are not possible and the center is interested only in the outcome chosen. We then show how allowing for randomization in the mechanism makes problems in this setting computationally easy.

In multiagent settings where the agents have different preferences, preference aggregation is a central issue. Voting is a general method for preference aggregation, but seminal results have shown that all general voting protocols are manipulable. One could try to avoid manipulation by using voting protocols where determining a beneficial manipulation is hard computationally. The complexity of manipulating realistic elections where the number of candidates is a small constant was recently studied [4], but the emphasis was on the question of whether or not a protocol becomes hard to manipulate for some constant number of candidates. That work, in many cases, left open the question: How many candidates are needed to make elections hard to manipulate? This is a crucial question when comparing the relative manipulability of different voting protocols. In this paper we answer that question for the voting protocols of the earlier study: plurality, Borda, STV, Copeland, maximin, regular cup, and randomized cup. We also answer that question for two voting protocols for which no results on the complexity of manipulation have been derived before: veto and plurality with runoff. It turns out that the voting protocols under study become hard to manipulate at 3 candidates, 4 candidates, 7 candidates, or never.