Algorithmic Justice and Hostile Solidarity: An interdisciplinary examination of Parole Board decisions mediated by automated decision-making systems

在惩罚研究中占主导地位的实证主义思维一直忽视了惩罚的排他性方面，忽视了敌对情绪（如恐惧）在构建压迫性刑事司法实践中所起的作用。在过去的几十年里，英国的惩罚性话语和实践在很大程度上以明确的报复性话语为特征，这反过来又加强了更严格法律的制定。再加上在当前社会条件下普遍缺乏社会团结意识，在惩罚理论和实践中产生了根深蒂固的敌对情绪，形成了一种“敌对团结”的形式，这种“敌对团结”通过“疏远”来促进“认同”——因为“守法公民”对罪犯的敌意变得团结起来。由于康复问题主要服从于报复主义，考虑到使用数据来提高刑事司法效率，导致越来越多地部署自动决策系统（ADM）来预测再犯风险。尽管有证据表明ADM可能是不准确和有偏见的，这与自由主义司法原则的内在矛盾，但英国刑事法院在假释程序中采用了一种名为罪犯评估系统（OASys）的技术。然而，平庸化OASys的部署可能不仅(i)将其决策歪曲为“理性的”（因此加强了对情感与刑事司法相关的忽视），而且（ii）加强了刑事司法决策者和罪犯之间的隔阂（从而助长了该领域现有的他者感）。为了解决这些问题，这个项目将调查ADM在刑事程序中的部署（以及它所促进的“客观性”的伪装）是否会加强英国社会的“敌对团结”，并导致决策者对其决策后果的情感疏离（因此混淆了他们的情感在量刑过程中所起的作用）。这样的调查是相关的，因为它可以弥合关于定罪和惩罚的情感方面的知识差距，这是一个很少被探索的话题。此外，“算法治理”在刑事司法中基本上没有得到解决，因此，鉴于这一领域的高风险，提出的分析不仅相关，而且迫切。在微观环境中，该项目将允许重新评估决策者与OASys之间的相互作用。在宏观环境中，它可以提供知识，以反映和审查全球刑事司法系统中使用其他决策技术的广泛背景下的人类互动。“敌对团结”模式将作为一个框架，以了解周围的犯罪行为的当前条件的概念基础，因为它提供了一个合适的理论背景，以欣赏在英国刑事司法环境的复杂性。跨学科将是追求与惩罚、情感和刑事程序的犯罪学、心理学和社会学方面的理论接触的关键。此外，对假释决定依赖于OASys的法庭程序进行定性的社会法律评估，将有助于理解决策者遵循的情感理由。这种评估将能够批判性地评估OASys的部署是否（以及如何）加强了“敌对团结”的感觉，并导致决策者在情绪上疏远其决策的后果。

Algorithmic Justice and Hostile Solidarity: An interdisciplinary examination of Parole Board decisions mediated by automated decision-making systems in the UKThe dominant positivist mindset in the study of punishment consistently disregards punishment's exclusionary aspect, neglecting the role played by hostile emotions (like fear) in constructing repressive criminal justice practices. Punitive discourse and practice in the last few decades in the UK have been largely characterised by explicitly retributive discourses that in turn reinforce the enactment of more strict laws. This combined with the lack of a prevalent sense of social solidarity in current social conditions has ingrained hostile emotions within punishment theory and practice, fostering a form of 'hostile solidarity' which feeds 'identification through estrangement' - as 'law abiding citizens' become solidary in their animosity towards offenders. With rehabilitation concerns mostly subordinated to retributivism, considerations about the use of data to enhance efficacy in criminal justice led to the increasing deployment of automated decision-making systems (ADM) to predict recidivism risks. Despite the inherent contradiction with liberal justice principles of prospectively assessing punishment and the evidence that ADM can be inaccurate and biased, UK criminal courts deploy a technology named Offender Assessment System (OASys) in parole proceedings. However, banalizing OASys deployment may not only (i) misrepresent its decisions as 'rational' (therefore reinforcing the neglect of emotions' relevance to criminal justice), but also (ii) strengthen estrangement between criminal justice decision-makers and offenders (thus contributing to the existing sense of othering in this field). To address such problems, this project will investigate whether ADM deployment in criminal procedure (and the pretence to 'objectivity' it promotes) may reinforce 'hostile solidarity' in the UK society and contribute to decision-makers' emotional alienation from the consequences of their decisions (therefore obfuscating the role their emotions play in the sentencing process). Such investigation is pertinent as it can bridge the knowledge gap over the emotional facet of criminalisation and punishment, a scarcely explored topic. Moreover, 'algorithmic governance' continues essentially unaddressed in criminal justice, hence the proposed analysis is not only relevant, but also urgent, given the high stakes in this area. In a micro context, this project will allow the reassessment of interactions decision-makers have with OASys. In a macro setting, it could contribute with knowledge to reflect over and review human interactions in a wide range of contexts where other decision-making technologies are used in criminal justice systems globally. The 'hostile solidarity' model will serve as a framework to understand the conceptual foundations of current conditions surrounding criminal practices, as it provides a suitable theoretical context to appreciate the complexity of the criminal justice environment in the UK. Interdisciplinarity will be key to pursue theoretical engagement with criminological, psychological, and sociological aspects of punishment, emotions, and criminal procedure. Moreover, a qualitative socio-legal assessment of court proceedings where OASys is relied upon for parole decisions will help understand the emotional rationale followed by decision-makers. Such assessment will enable the critical assessment of whether (and how) the deployment of OASys has reinforced the sense of 'hostile solidarity' and contributed to the emotional alienation of decision-makers from the consequences of their decisions.