IMBS Lunch talk: May 28, 2014
Lunch talk at the Institute for Mathematical Behavioral Sciences (IMBS), UC Irvine Wednesday, 28 May 2014
Can we quantify the role of social influence in individual decisions?
Frank Schweitzer, Chair of Systems Design, ETH Zurich, Switzerland
Social influence occurs in various ways in offline and online interactions of individuals. In most cases, however, it can be hardly quantified, which restricts empirical analyses and subsequent modeling to controlled setups where social interactions are traced and recorded. In this talk, I will discuss three such setups, both with respect to the empirical findings and their mathematical modeling. The first example focuses on the wisdom of crowds and how the prediction performance is affected if individuals get to know the guesses of others, either as aggregated or as detailed information. In the second example, I discuss the emotional influence between users of online platforms and how it affects their activity patterns. The third example tries to reconstruct cascades of users leaving an online network by focusing on the influence of the decisions of their friends. As all three examples demonstrate, social influence can result in better or worse outcome, i.e., stabilizing or destabilizing a collective dynamics, or improving or deteriorating the performance of systems.
Personal homepage (CV, Talks, Boards, Conferences/Schools, ...)
Nicolas Wider, Antonios Garas, Ingo Scholtes, Frank Schweitzer. 2015. An ensemble perspective on multi-layer networks
We study properties of multi-layered, interconnected networks from an ensemble perspec- tive, i.e. we analyze ensembles of multi-layer networks that share similar aggregate charac- teristics. Using a diffusive process that evolves on a multi-layer network, we analyze how the speed of diffusion depends on the aggregate characteristics of both intra- and inter-layer connectivity. Through a block-matrix model representing the distinct layers, we construct transition matrices of random walkers on multi-layer networks, and estimate expected prop- erties of multi-layer networks using a mean-field approach. In addition, we quantify and explore conditions on the link topology that allow to estimate the ensemble average by only considering aggregate statistics of the layers. Our approach can be used when only partial information is available, like it is usually the case for real-world multi-layer complex systems.
 The Rise and Fall of R&D Networks http://arxiv.org/abs/1304.3623
 Selection rules in alliance formation: strategic decisions or abundance of choice? http://arxiv.org/abs/1403.3298
 The role of endogenous and exogenous mechanisms in the formation of R&D networks http://arxiv.org/abs/1403.4106
Editor-in-Chief, "Advances in Complex Systems" (ACS)
Editor-in-Chief, "European Physical Journal B: Condensed Matter and Complex Systems" (EPJB) with responsibility for the section "Complex Systems"
Haven't heard from you, for a while.
Since you are the master of interorganisational collaborations, I thought you could be interested in our recent papers on the empirics and modelling of R&D networks.
 The Rise and Fall of R&D Networks http://arxiv.org/abs/1304.3623 Abstract: Drawing on a large database of publicly announced R&D alliances, we track the evolution of R&D networks in a large number of economic sectors over a long time period (1986-2009). Our main goal is to evaluate temporal and sectoral robustness of the main statistical properties of empirical R&D networks. We study a large set of indicators, thus providing a more complete description of R&D networks with respect to the existing literature. We find that most network properties are invariant across sectors. In addition, they do not change when alliances are considered independently of the sectors to which partners belong. Moreover, we find that many properties of R&D networks are characterized by a rise-and-fall dynamics with a peak in the mid-nineties. Finally, we show that the properties of empirical R&D networks support predictions of the recent theoretical literature on R&D network formation.
 Selection rules in alliance formation: strategic decisions or abundance of choice? http://arxiv.org/abs/1403.3298 Abstract: We study how firms select partners using a large database of publicly announced R&D alliances over a period of 25 years. We identify, for the first time, two distinct behavioral strategies of firms in forming these alliances. By reconstructing and analysing the temporal R&D network of 14,000 international firms and 21.000 publicly announced alliances, we find a "universal" behavior in firms changing between these strategies. In the first strategy, newcomers and nodes of low centrality initially establish links to nodes of similar or higher centrality. After these firms have consolidated their position and increased their centrality, they switch to the second strategy, and preferably form links to less central nodes. In addition, we show that k-core centrality can be established as a measure of firm's success that correlates e.g. with the number of patents (obtained from a dataset of 3 Mio patents). To synthesize our findings, we provide a network growth model based on k-core centrality which reproduces the strategic behavior of firms, as well as other properties of the empirical network.
 The role of endogenous and exogenous mechanisms in the formation of R&D networks http://arxiv.org/abs/1403.4106 Abstract: We develop an agent based model of strategic link formation in Research and Development (R&D) networks. In our model, firms are represented as nodes, and their R&D alliances as links. Empirical evidence has shown that the growth and evolution of these networks are driven by two types of mechanisms. Endogenous mechanisms depend on the previous history of the network, i.e. they take into account existing alliance patterns. Exogenous mechanisms do not depend on the properties of the network but on those of newcomers, i.e. they include the exploratory search for firms that are not part of the network yet. Despite the observation of these two mechanisms, research has focused only on either of the two. To overcome this limitation, we develop a general modeling framework that includes both mechanisms and allows to tune their relative importance in the formation of links. The model contains additional ingredients derived from empirical observations, such as the heterogeneous propensity to form alliances and the presence of circles of influence, i.e. clusters of firms in the network. We test our model against the Thomson Reuters SDC Platinum dataset, one of the most complete datasets available nowadays, listing cross-country R&D alliances from 1984 to 2009. Interestingly, by fitting only three macroscopic properties of the network, this framework is able to reproduce a number of microscopic measures characterizing the network topology, including the distributions of degree, local clustering, path length and component size, and the emergence of network clusters. Furthermore, by estimating the link probabilities towards newcomers and established firms from the available data, we find that endogenous mechanisms are predominant over the exogenous ones in the network formation. This quantifies the importance of existing network structures in selecting partners for R&D alliances.
 is a large scale analysis (actually the largest one we are aware of) of the structural and temporal features of empirical R&D networks.  takes a deeper (emprirical) look into how new entrants form links with incumbants and provides a model that also predicts the success of formed alliances.  provides a new agent-based model of strategic alliance formation that recaptures important features of empirical R&D networks.
I would be glad to receive your comments on these papers.
Kind regards, and see you hopefully in Berkeley in June (http://netsci2014.cpt.univ-mrs.fr/), Frank
-- Prof. Dr. Frank Schweitzer ETH Zurich, Chair of Systems Design, http://www.sg.ethz.ch
Frank Schweitzer, 2007, Multi-Agent Approach to the Self-Organization of Networks. In, F. Reed-Tsochas, N. F. Johnson, J. Efstathiou: Understanding and Managing Complex Agent-Based Dynamical Networks, Singapore: World Scientific.
Frank Schweitzer, Didier Sornette, A. Vespignani, G. Fagiolo, F. Vega-Redondo, D. R. White. 2009. Economic Networks: The New Challenges. Science 24 July: 422-425. Special Issue on Complex Systems and Networks http://www.sciencemag.org/cgi/reprint/325/5939/422.pdf.
- Schweitzer and colleagues present their research into the structure and dynamics of economic networks and the problems of finding a complex systems approach to facilitate the design of policies. The research into economic networks has been studied from two perspectives: more micro approaches from economics or sociology and more macro approaches from physics or computer science. The micro approaches focus on individual system elements and their detailed network of relations. Game theory is often used to model the dynamics of agent behavior. The context of behavior may be modeled by the more macro approaches. These focus on the statistical regularities of the network as a whole but taken alone would fail to link to the economic motivation of individual agents. Combinations of these approaches may describe what is needed in the challenges of predicting stabilities and instabilities and propose economic policies. With computational models, large-scale network data can be processed and can reflect complex agent interactions.
2009 Frank Schweitzer, Didier Sornette, G. Fagiolo, F. Vega-Redondo, D. R. White. Economic Networks: What do we know and what do we need to know? Advances in Complex Systems (ACS) 12(4-5):407-422. Preprint: Santa Fe Institute Working Paper 09-09-038.
- Abstract: We examine the emergent field of economic networks and explore its ability to shed light on the global and volatile economy where credit, ownership, innovation, investment, and virtually every other economic activity is carried at a scale and scope that respects no geographical, organizational, or political boundaries. In this context, the study of economic networks and their dynamics must reflect the vast complexity of the interaction patterns and integrate it with a realistic account of the incentives and information that govern agents' behavior. The interplay of both has been shown to produce metastabilities, system crashes, and emergent structures in ways that are yet only poorly understood. Meeting this exciting scientific challenge requires a combination of time series analysis, complexity theory, and simulation with the analytical tools that have been developed by game theory, as well as graph and matrix theories. We argue that this will help achieving a better integration of theory and data models and provide a better understanding of the potentials and risks of modern economic systems.
Prof. Dirk Helbing
Prof. Didier Sornette