Deciphering Algorithmic Collusion: Insights from Bandit Algorithms and Implications for Antitrust Enforcement
Author(s)
Marty, FrédéricContributor(s)
Groupe de Recherche en Droit, Economie et Gestion (GREDEG) ; Université Nice Sophia Antipolis (1965 - 2019) (UNS)-Centre National de la Recherche Scientifique (CNRS)-Université Côte d'Azur (UCA)PNCR
Keywords
Algorithmic CollusionBandit Algorithms
Antitrust Enforcement
Unilateral Signals
Pricing Strategies
Collusion algorithmique
algorithmes de bandits
application de la législation antitrust
signaux unilatéraux
stratégies de fixation des prix
Full record
Show full item recordOnline Access
https://shs.hal.science/halshs-04363106Abstract
This paper examines algorithmic collusion from legal and economic perspectives, highlighting the growing role of algorithms in digital markets and their potential for anti-competitive behavior. Using bandit algorithms as a model, traditionally applied in uncertain decision-making contexts, we illuminate the dynamics of implicit collusion without overt communication. Legally, the challenge is discerning and classifying these algorithmic signals, especially as unilateral communications. Economically, distinguishing between rational pricing and collusive patterns becomes intricate with algorithm-driven decisions. The paper emphasizes the imperative for competition authorities to identify unusual market behaviors, hinting at shifting the burden of proof to firms with algorithmic pricing. Balancing algorithmic transparency and collusion prevention is crucial. While regulations might address these concerns, they could hinder algorithmic development. As this form of collusion becomes central in antitrust, understanding through models like bandit algorithms is vital, since these last ones may converge faster towards an anticompetitive equilibrium.Cet article examine la collusion algorithmique du point de vue juridique et économique, mettant en évidence le rôle croissant des algorithmes dans les marchés numériques et leur potentiel comportement anticoncurrentiel. En utilisant les algorithmes de bandit comme modèle, traditionnellement appliqués dans des contextes de prise de décision incertaine, nous mettons en lumière la dynamique de la collusion implicite sans communication explicite. Sur le plan juridique, le défi réside dans le discernement et la classification de ces signaux algorithmiques, en particulier en tant que communications unilatérales. Sur le plan économique, la distinction entre une tarification rationnelle et des schémas collusifs devient complexe avec les décisions pilotées par des algorithmes. L'article met l'accent sur l'impératif pour les autorités de la concurrence d'identifier les comportements de marché inhabituels, laissant entendre un transfert du fardeau de la preuve aux entreprises pratiquant la tarification algorithmique. Équilibrer la transparence algorithmique et la prévention de la collusion est crucial. Bien que la réglementation puisse traiter ces préoccupations, elle pourrait entraver le développement des algorithmes. À mesure que cette forme de collusion devient centrale dans le domaine de la concurrence, la compréhension à travers des modèles tels que les algorithmes de bandit est essentielle, car ces derniers peuvent converger plus rapidement vers un équilibre anticoncurrentiel.
Date
2023-12-22Type
info:eu-repo/semantics/preprintIdentifier
oai:HAL:halshs-04363106v1halshs-04363106
https://shs.hal.science/halshs-04363106
Collections
Related items
Showing items related by title, author, creator and subject.
-
Against the Dehumanisation of Decision-Making-Algorithmic Decisions at the Crossroads of Intellectual Property, Data Protection, and Freedom of InformationNorthumbria University; Northumbria University; 0000-0001-6918-5398; Noto La Diega, Guido (Digital Peer Publishing, 2020-01-24)This work presents ten arguments against algorithmic decision-making. These re-volve around the concepts of ubiquitous discretionary interpretation, holistic intu-ition, algorithmic bias, the three black boxes, psychology of conformity, power of sanctions, civilising force of hypocrisy, pluralism, empathy, and technocracy. Nowadays algorithms can decide if one can get a loan, is allowed to cross a bor-der, or must go to prison. Artificial intelligence techniques (natural language pro-cessing and machine learning in the first place) enable private and public deci-sion-makers to analyse big data in order to build profiles, which are used to make decisions in an automated way. The lack of transparency of the algorithmic deci-sion-making process does not stem merely from the characteristics of the relevant techniques used, which can make it impossible to access the rationale of the deci-sion. It depends also on the abuse of and overlap between intellectual property rights (the “legal black box”). In the US, nearly half a million patented inventions concern algorithms; more than 67% of the algorithm-related patents were issued over the last ten years and the trend is increasing. To counter the increased mo-nopolisation of algorithms by means of intellectual property rights (with trade se-crets leading the way), this paper presents three legal routes that enable citizens to ‘open’ the algorithms. First, copyright and patent exceptions, as well as trade se-crets are discussed. Second, the EU General Data Protection Regulation is critical-ly assessed. In principle, data controllers are not allowed to use algorithms to take decisions that have legal effects on the data subject’s life or similarly significantly affect them. However, when they are allowed to do so, the data subject still has the right to obtain human intervention, to express their point of view, as well as to contest the decision. Additionally, the data controller shall provide meaningful in-formation about the logic involved in the algorithmic decision. Third, this paper critically analyses the first known case of a court using the access right under the freedom of information regime to grant an injunction to release the source code of the computer program that implements an algorithm. Only an integrated ap-proach – which takes into account intellectual property, data protection, and free-dom of information – may provide the citizen affected by an algorithmic decision of an effective remedy as required by the Charter of Fundamental Rights of the EU and the European Convention on Human Rights.
-
Comprendre les algorithmes, leurs risques et leurs promessesCohen-Addad, Vincent (2022)Après avoir défini ce qu’est un algorithme, nous proposons dans cet article un panorama du champ de recherche de l’algorithmique. Son évolution a conduit à ajouter à une propriété prescriptive une dimension descriptive, compte tenu de l’utilité des algorithmes à expliquer le monde qui nous entoure.  Nous identifions aussi différentes limites dans la pratique, dont la difficile formalisation de certains problèmes en langage mathématique ou l’incapacité à expliquer certains algorithmes obtenus au travers des méthodes d’apprentissage. Limitations que nous formalisons au travers de la notion de « contrat algorithmique ». Cet article se conclut en évoquant les récentes pistes de recherche apparues afin de limiter les biais occasionnés par les mésusages des algorithmes. Enfin, nous posons la question de l’encadrement normatif de ces derniers mais également celle de leur régulation.
-
Fast numerical algorithms for optimal robot motion planningLaValle, Steven M.; Heath, Michael T.; LaValle, Steven M.; Olson, Luke N.; Frazzoli, Emilio; Yershov, Dmytro (2014-01-16)Optimization of high-level autonomous tasks requires solving the optimal motion planning problem for a mobile robot. For example, to reach the desired destination on time, a self-driving car must quickly navigate streets and avoid hazardous obstacles such as buildings or other cars, as well as provide safety for pedestrians. Our approach to solve the optimal planning problem is to use the optimal control formalism borrowed from the control theory. In this setting, safety constraints for either a robot or its surroundings are defined as obstacles, which are penalized using infinite cost to guarantee that tasks are performed safely. Unlike in control theory, the complexity of real-world tasks in addition with safety constraints prohibit finding an analytic solution to the optimal motion planning problem. Hence, the application of numerical algorithms is necessary. In this thesis, we demonstrate that solutions to a general motion planning problem are computable, which permits the use of numerical algorithms to solve this problem. Moreover, we propose a numerical discretization of a general optimal motion planning problem and prove that this discretization is accurate. Numerical algorithms that use the proposed discretization are applied to several realistic motion planning problems. Using these algorithms, we demonstrate the practicality of the proposed numerical approach. In addition, we extend our consideration beyond the classical deterministic models of motion and apply the proposed numerical algorithms to solve a stochastic optimal motion planning problem.