GAM Models M1 and M2.

The emergence of social order on darknet markets presents social scientists with a unique puzzle. Because these markets operate outside of conventional regulatory frameworks, there is a lack of legitimate oversight to monitor transactions and protect users from opportunistic behaviour. While existing literature often examines the role of reputation in increasing sales, little attention has been paid to mechanisms that mitigate fraud. This study fills this gap by examining one of the largest known darknet platforms, Alphabay, which was operational from December 2014 to July 2017. Using two Generalised Additive Models (GAMs), results show that costly signals, such as a positive reputation, sellers’ seniority and escrow services, are inversely associated with fraudulent activity on darknet markets. Conversely, cheap signals, such as long product descriptions characterised by complex vocabulary and a positive tone, correlate positively with opportunistic behaviour. The study provides empirical support for signalling theory, by showing that costly signals are more difficult to fake or manipulate and can reduce fraud. Conversely, the study also demonstrates empirically that cheap signals, while potentially effective in initially generating trust among buyers, are associated with an increase in fraud and opportunistic behaviour.暗网市场中社会秩序的涌现，为社会科学家带来了一道独特的谜题。由于此类市场运行于传统监管框架之外，缺乏合法监管以监控交易、防范用户遭遇机会主义行为。尽管既有研究多探讨声誉对提升销量的作用，但针对诈骗行为抑制机制的关注却十分匮乏。本研究以2014年12月至2017年7月间运营的全球知名大型暗网平台之一阿尔法贝伊（Alphabay）为研究对象，填补了这一研究空白。本研究采用两类广义加性模型（Generalised Additive Models, GAMs），结果显示：正向声誉、卖家资历与托管服务等高成本信号，与暗网市场中的诈骗活动呈负相关关系。与之相反，诸如用词复杂且语调积极的冗长商品描述这类低成本信号，则与机会主义行为呈正相关关系。本研究为信号理论提供了实证支撑：高成本信号更难伪造或操纵，能够有效降低诈骗行为发生率。与此同时，本研究也通过实证表明：低成本信号虽可在初始阶段有效获取买家信任，但却会伴随诈骗与机会主义行为的增加。