Incentive Mechanisms for Users and Platforms of Social Networks

在过去的几十年里，经济学中关于社会网络的文献，无论是实证的还是理论的，都在大幅增长。这种增长在很大程度上可以归因于在线社交网络的激增。理论论文为网络形成过程和网络内影响力的传播提供了重要的模型（Jackson and Wolinsky, 1996; Mutuswami and Winter, 2002; Kempe et al., 2003; Ballester et al., 2006; Jackson, 2011）。实证论文利用现有数据和实地实验来研究行为如何在网络中扩散（Ichino和Maggi, 2000; Sacerdote, 2001； Mas和Moretti, 2009； Banerjee等人，2013）。关于社交媒体平台的文献也在不断涌现（Ghosh and McAfee, 2011; Ashish, and Ronaghi, 2012; Ghosh and Hummel, 2014）。大多数关于社交媒体的图论/博弈论文献都将网络结构视为给定和固定的，并将重点放在信息在网络中的流动和用户的战略考虑上。平台本身的角色通常没有建模。在这个项目中，我们的目标是开发和分析社交媒体模型，主要使用博弈论和图论中的工具。我们的方法与现有的许多文献不同，因为我们将平台的所有者/管理者视为社交媒体游戏中的主要参与者——一个对网络中发生的活动具有巨大影响的人。这种影响是通过对社交网络结构的一定程度的控制和对用户可用行为的几乎完全控制而产生的。我们的建模工作将不仅仅关注用户的动机和偏好。平台本身的动机和偏好同样重要。这些偏好很大程度上取决于平台的商业模式。在某些商业模式中，它们可以与用户的偏好保持一致，而在其他商业模式中，它们可能与用户的偏好相冲突。我们在这里注意到，平台的偏好和兴趣是渴望提供政策影响的模型的关键要素。如果平台和用户有相同的兴趣，就不需要干预。当这些利益发生冲突时，可能需要政府或国际组织在某些情况下进行干预。我们的第一个目标是确定平台和用户的利益冲突和一致的领域。我们还将研究社交媒体的机制和规则，这些机制和规则可以在不给平台带来重大收入损失的情况下实现SM内容质量的帕累托改进或用户福利的实质性改善。我们计划解决的更具体的问题是：从平台的角度来看，什么是最优的网络结构？这种结构可能对用户表达自己的权利的分配和他们的整体福利产生什么样的影响？基于无向图的社交媒体平台（如Facebook和LinkedIn）与基于有向图的社交媒体平台（如Twitter和Instagram）在用户福利方面有何不同？从平台和用户的角度来看，网络中内容流的最佳策略（例如，feed新闻策略）是什么？用户通过网络传播不良内容和假新闻背后的动机是什么？如何减轻这种不良动机？我们的项目旨在通过与Ofcom、FCC等组织的合作，产生实质性的政策影响。很有希望的是，这种合作的倡议是在我们关于该主题的讨论文件出现后由Ofcom提出的。我们被邀请来展示我们在Ofcom的工作，我们现在正在与Ofcom合作组织一个关于这个主题的会议，Ofcom、FCC、Facebook和b谷歌的主要政策人员已经确认参加。我们计划在项目实施期间与Ofcom的人员定期举行会议。

The literature on social networks in economics, both empirical and theoretical has been growing substantially in the last few decades. Much of this growth can be attributed to the proliferation of online social networks. Theory papers have provided important models for the process of network formation and the spread of influence within networks (Jackson and Wolinsky, 1996; Mutuswami and Winter, 2002; Kempe et al., 2003; Ballester et al., 2006; Jackson, 2011). Empirical papers used existing data as well as field experiments to study how behavior diffuses within a network (Ichino and Maggi, 2000; Sacerdote, 2001; Mas and Moretti, 2009; Banerjee et al., 2013). There is also emerging literature on social media platforms (Ghosh and McAfee, 2011; Ashish, and Ronaghi, 2012; Ghosh and Hummel, 2014). Most of the Graph/Game theory literature of social media treats the network structure as given and fixed and focuses on the flow of information through the network and on users' strategic considerations. The role of the platform itself is typically left un-modeled. Our objective in this project is to develop and analyze models of social media using primarily tools in Game Theory and Graph Theory. Our approach is different from much of the existing literature in that we treat the platform's owners/managers as a major player in the social media game - one that has an enormous influence on the activity that takes place in the network. This influence arises through a certain degree of control over the structure of the social network and almost full control over users' available actions. Our modeling efforts will not only focus on users' motives and preferences. The motives and preferences of the platform itself are equally important. These preferences critically depend on the platform's business model. They can be aligned with users' preferences in some business models and conflicting with users' preferences in others. We note here that platforms' preferences and interests are key elements in models that aspire to provide policy implications. If platforms and users have identical interests there is little need for intervention. It is when these interests are conflicting that intervention by governments or international organizations in certain circumstances might be required.Our first objective will be to identify areas in which the interests of the platform and its users are conflicting and areas in which they coincide. We will also study mechanisms and rules for social media that can bring about Pareto improvements in SM content quality or substantial improvement to users' welfare with no major revenue losses to the platform. More specific questions we plan to address are:What are the optimal network structures from the point of view of the platform and what sort of implications such structure might have on the distribution of users' right to express themselves and their overall welfare? How do social media platforms based on undirected graphs (e.g. Facebook and LinkedIn) differ from those based on directed graphs (Twitter and Instagram) in terms of users' welfare? What is the optimal policy for the flow of content in the network (e.g., feed news policy) from the point of view of both the platform and users? What are the incentives behind users' spreading abusive content and fake news through the network, and how can such ill incentives be mitigated? Our project is designed to have a substantial policy impact through collaboration with organizations such as Ofcom, FCC. Promisingly enough the initiative for such collaboration came from Ofcom following the appearance of our discussion paper on the topic. We were invited to present our work at Ofcom and we are now collaborating with Ofcom in organizing a conference on this subject where key policy people from Ofcom, the FCC, Facebook and Google have confirmed their participation. We plan to hold regular meetings with people at Ofcom during the implementation of the project.