Bien collectif | EPFL Graph Search

En économie publique, un bien collectif est un bien dont la consommation par un individu supplémentaire ne réduit pas la satisfaction des autres. Le bien collectif existant ou créé, est accessible à tous sans condition : comme l'air pur, une justice intègre et rapide, un réseau de téléphone qui fonctionne bien, une bonne route. Il existe aussi des services collectifs rendus par un bien privé tel que la radio ou les autoroutes. Les exemples typiques sont les forêts, les espaces maritimes de pêche, les nappes d'eau souterraines ou les systèmes d'irrigation. Un ouvrage portant sur les transferts massifs d'eau remet en question la qualification de l'eau au titre de bien collectif, cette ressource ne pouvant effectivement pas être exploitée sans limite. Un bien de club est un type particulier de bien collectif dont la consommation respecte le principe de non-rivalité mais qui a la possibilité d'être rendu excluable. Un exemple de bien de club est la télévision chiffrée : il faut payer po

Scalable Multi-agent Coordination and Resource Sharing

Panayiotis Danassis

A plethora of real world problems consist of a number of agents that interact, learn, cooperate, coordinate, and compete with others in ever more complex environments. Examples include autonomous vehicles, robotic agents, intelligent infrastructure, IoT devices, and so on. As more and more autonomous agents are deployed in the real-world, it will bring forth the need for novel algorithms, theory, and tools to enable coordination on a massive scale. In this thesis, we develop such tools to tackle two central challenges in multi-agent coordination research: solving allocation problems, and resource sharing, focusing on solutions that are scalable, practical, and applicable to real-world problems.In the first part of the thesis we tackle the problem of allocating resources to agents, i.e., solving a weighted matching problem. Real-world matching problems may occur in massively large systems, they are distributed and information-restrictive, and individuals have to reveal their preferences over the possible matches in order to get a high quality match, which brings forth significant privacy risks. As such, there are three main challenges: complexity, communication, and privacy.Our proposed approach, ALMA, is a practical heuristic designed for real-world, large-scale (

10^6

agents) applications. It is based on a simple altruistic behavioral convention: agents have a higher probability to back-off from contesting a resource if they have good alternatives, potentially freeing the resource for some agent that does not. ALMA tackles all of the aforementioned challenges: it is decentralized, runs on-device, requires no inter-agent communication, converges in constant time -- under reasonable assumptions --, and provides strong, worst-case, privacy guarantees. Moreover, by incorporating learning we can mitigate the loss in social welfare and increase fairness. Finally, rational agents can use such simple conventions, along with an arbitrary signal from the environment, to learn a correlated equilibrium for accessing a set resources, under high congestion.In the second part of the thesis we focus on a critical open problem: the question of cooperation in socio-ecological and socio-economical systems, and sustainability in the use of common-pool resources. In recent years, learning agents, especially deep reinforcement learning agents, have become ubiquitous in such systems. Yet, scaling to environments with a large number of agents and low observability continues to be a challenge. In our work, we focus on common-pool resources. Individuals face strong incentives to appropriate, which results in overuse and even the depletion of the resources. Our goal is to apply simple interventions to steer the population to desirable states.We propose a simple, yet powerful, and robust technique: allow agents to observe an arbitrary common signal from the environment. The agents learn to couple their policies, and avoid depletion in a wider range of settings, while achieving higher social welfare and convergence speed.Finally, we propose a practical approach to computing market prices and allocations via a deep reinforcement learning policymaker agent. Compared to the idealized market equilibrium outcome -- which can not always be efficiently computed -- our policymaker is much more flexible, allowing us to tune the prices with regard to diverse objectives such as sustainability and resource wastefulness, fairness, buyers' and sellers' welfare, etc.

EPFL2022

Exploiting environmental signals to enable policy correlation in large-scale decentralized systems

Panayiotis Danassis, Zeki Doruk Erden, Boi Faltings

Can artificial agents benefit from human conventions? Human societies manage to successfully self-organize and resolve the tragedy of the commons in common-pool resources, in spite of the bleak prediction of non-cooperative game theory. On top of that, real-world problems are inherently large-scale and of low observability. One key concept that facilitates human coordination in such settings is the use of conventions. Inspired by human behavior, we investigate the learning dynamics and emergence of temporal conventions, focusing on common-pool resources. Extra emphasis was given in designing a realistic evaluation setting: (a) environment dynamics are modeled on real-world fisheries, (b) we assume decentralized learning, where agents can observe only their own history, and (c) we run large-scale simulations (up to 64 agents). Uncoupled policies and low observability make cooperation hard to achieve; as the number of agents grow, the probability of taking a correct gradient direction decreases exponentially. By introducing an arbitrary common signal (e.g., date, time, or any periodic set of numbers) as a means to couple the learning process, we show that temporal conventions can emerge and agents reach sustainable harvesting strategies. The introduction of the signal consistently improves the social welfare (by 258% on average, up to 3306%), the range of environmental parameters where sustainability can be achieved (by 46% on average, up to 300%), and the convergence speed in low abundance settings (by 13% on average, up to 53%).

SPRINGER2022

The organisation of science: topics, incentives and funding.

Omar Ballester Gonzalez

Ever since the links between the development of new technologies and economic growth became evident, researchers have attempted to study how the creation of knowledge fosters progress. If pushing the frontier of knowledge has an impact on progress and well-being, it is essential to pursue some form of science policy. Policymakers rely on the scholarly work of researchers in order to understand the likely impact of new policies and investments, and evaluate the state of the art in science and innovation policy. Therefore, the work of social scientists, economists of science and information scientists, among others, is vital to the characterisation, understanding and management of science. In recent years, the availability and quality of science data (including bibliographic data and metadata, funding, relational databases, ontologies and classifications) has boosted the empirical work in the depiction of the organisational structure of science. In turn, policy analysis has been able to accurately identify many unsuspected effects of past investments and policy decisions both at the macro and micro level. Using topic models, we develop a novel method for evaluating the robustness of different text-to-text similarity models. Employing that procedure, we find that the neural-network-based paragraph embeddings approach seems capable of providing statistically robust estimates of document--document similarities. Finding methods to estimate the similarity between individual publications is an area of long-standing interest in the information science and scientometrics communities. These techniques enable researchers to build indicators and classification methods based on the analysis of large text corpora. We show that the most widely used techniques suffer from inconsistencies upon retraining, and provide a procedure to evaluate and compare the quality of different methods, regardless of the data. Next, we present a game-theoretic model of rewards to scientific contributions. Our model of science may help explain the resulting social organisation of science from a simple social dilemma model. We model a researcher's payoff as a common-pool resource game, intrinsically connecting the appropriability of scientific output to a scientist's optimal strategy. This simple model of reward allocation sheds new light on a variety of behaviours that have been observed amongst researchers. Finally, we propose an empirical analysis of the relationship between basic knowledge generation and spillovers to innovation. Using the United States' 2001 ban on federal funding of human embryonic stem cells (hESC), we disentangle the effect that policy had on downstream innovation. We employ recently developed data on patent-to-scientific-article citations to measure the spillovers, and we characterise the causal impact of the policy on subsequent innovation with a difference-in-differences estimator. Our estimates suggest that in the years following the policy, scholarly publications subject to the ban received 65 to 80 per-cent fewer patent citations than the control group. We then apply topic modelling techniques to examine changes in the direction of science. In particular we build a topic-variety metric. Our findings show that variety decreased in the aftermath of the policy. Our results suggest that even the most modest policy changes have a profound impact on downstream innovation and the advancement at the frontier.

EPFL2020