How metrics matter: Mapping reactions to social media metrics in the public issues of Covid-19 misinformation and users' health and well-being

社交媒体指标，如计数或参与率，已经成为我们当代数据化社会的一个关键问题（Gerlitz & Helmond, 2013; Marres, 2017）。内容创作者、阴谋和极右翼团体、艺术家和活动家，在不同的议程下，正在展开不同的策略来操纵指标，并干预社交媒体平台上某些内容和账户的可见性、声誉和价值（Brunton & Nissenbaum, 2015; Bucher, 2018; Cotter, 2018; Marwick & Lewis, 2017; van der Nagel, 2018）。这些策略表明，“反应性”的形式在社交媒体中无处不在，或者“测量引起干预他们测量对象的人的反应”的想法（Espeland和Sauder， 2007，第2页）。社交媒体上的反应通常被定义为“不真实的行为”、“偏见的来源”或“欺诈”。出于这个原因，多个平台已经制定了法律政策、算法系统和界面重新设计，以打击“虚假粘性”和控制反应性。然而，响应式实践并不总是出乎意料的，也可能是平台故意设计的成果（Gerlitz & Lury, 2014）。这表明，在某些情况下，反应性不仅被接受，而且是社会媒体指标设计中的一种影响，但在其他情况下，它就变成了问题，并被定性为“滥用”或“有害”，不利于维持健康的社会媒体环境。这个博士研究项目将把对社交媒体指标的反应性的既定理解作为恶意操纵和方法偏见的必然问题形式，以调查它们最初是如何被认定为有问题的。借鉴数字社会学、科学技术研究和量化社会学等相关领域的工作，我的目标是探究社交媒体中对指标的反应所产生的独特问题。这个项目提出了以下问题：如何，为了谁，以及为了什么结束我们与社交媒体指标的关系成为一个值得关注的问题？我将在两个实质性领域探讨这个问题：a) Covid-19错误信息的放大或度量反应性如何促成阴谋和反疫苗团体的影响，b)社交媒体度量对用户健康和福祉的影响以及减轻这些影响的当代实验。我将进行多站点的数字民族志（Hine, 2015），其中包括使用数字方法进行跨平台分析（Marres, 2015; Rogers, 2013），以绘制两个选定的公共问题以及为解决社交媒体中的反应性而提出的证据、演示和干预形式；(b)与参与社交媒体反应性实践的内容创作者进行参与性观察和半结构化访谈，以修改其算法可见性，并与致力于检测和预防指标操纵的工程师和设计师进行访谈；(c)对有关社会媒体指标的实验和争议的二手材料的审查，如公共媒体、非政府组织报告、营销机构手册和政策文献；(d)开发原型或数字探针，以更投机和基于实践的探究方式，在研究本身中引出对社交媒体指标的反应。该研究项目将通过介绍关于社交媒体数据研究中测量的经典社会学辩论，为数字媒体研究和计算社会科学做出贡献。它建议将反应性实践重新思考为表现社交媒体数据的“原生人工”质量的情况（Marres, 2017; Marres & Gertliz, 2018），这些情况来自于测量设备和被测量对象之间的生动纠缠。

Social media metrics such as like counts or engagement rates have become a critical issue in our contemporary datafied societies (Gerlitz & Helmond, 2013; Marres, 2017). Content creators, conspiracy and far-right groups, artists and activists, under contrasting agendas, are unfolding diverse tactics to game the metrics and intervene in the (in)visibility, reputation and value of certain contents and accounts on social media platforms (Brunton & Nissenbaum, 2015; Bucher, 2018; Cotter, 2018; Marwick & Lewis, 2017; van der Nagel, 2018). These tactics indicate the ubiquity in social media of forms of 'reactivity' or the idea that 'measures elicit responses from people who intervene in the objects they measure' (Espeland and Sauder, 2007, p. 2). Reactivity in social media has been commonly framed as 'inauthentic behaviour', 'sources of bias', or 'fraud'. For this reason, multiple platforms have developed legal policies, algorithmic systems and interfaces redesigns to combat 'fake engagement' and control reactivity. However, reactive practices are not always unexpected, but can also be a deliberately designed achievement by platforms (Gerlitz & Lury, 2014). This suggests that, in certain situations, reactivity is not only accepted, but is an effect that is built into the design of social media metrics, but in others, it becomes problematic and characterized as "abusive" or "harmful" to the maintenance of a healthy social media environment.This doctoral research project will move established understandings of reactivity to social media metrics as necessarily problematic forms of malicious manipulation and methodological bias, in order to investigate how they come to be qualified as problematic in the first place. Drawing upon the work of interrelated fields of Digital Sociology, Science and Technology Studies and Sociology of Quantification, my aim is to inquiry into the distinctive problems that emerge from reactivity to metrics in social media. This project interrogates then the following: how, for whom, and for what ends our relations with metrics in social media becomes a matter of concern? I will explore this question in two substantive areas: a) the amplification of Covid-19 misinformation or how metric reactivity has contributed to the reach of conspiracy and anti-vaccine groups, b) the effects of social media metrics on health and well-being of users and contemporary experiments to mitigate these effects. I will conduct a multi-sited digital ethnography (Hine, 2015) that includes a cross-platform analysis with digital methods (Marres, 2015; Rogers, 2013) to map the two selected public issues and the forms of evidence, demonstration, and intervention put forward to address reactivity in social media; (b) participant observation and semi-structured interviews with content creators involved in reactive practices in social media to modify their algorithmic visibility, as well as with engineers and designers dedicated to the detection and prevention of metrics manipulation; (c) a review of secondary materials about experiments and controversies around social media metrics such as public media, NGO reports, marketing agencies brochures and policy literature; (d) the development of prototypes or digital probes to elicit reactions to social media metrics within the research itself, in a more speculative and practice-based line of inquiry. This research project will contribute to Digital Media Studies and Computational Social Science by introducing classical sociological debates about measurement in the study of social media data. It proposes to rethink reactive practices as situations that manifest the "natively artificial" quality of social media data (Marres, 2017; Marres & Gertliz, 2018) that emerge from the lively entanglements between measuring devices and measured objects.