Computer-Aided Rationalization of Sentencing

Although the statutory sentencing ranges prescribed by the German Federal Criminal Code (Strafgesetzbuch) apply equally nationwide, there is consid­er­a­ble interregional variation in sentencing. In addition, judges display consid­er­a­ble differences in punitiveness, depending on their specific role in the judiciary and their professional experi­ence. Consequently, there are repeated calls for standardization. Experience has shown, however, that the “margin” or “leeway” theory (Spielraumtheorie) developed by the German Federal Court of Justice (Bundesgerichtshof) in the 1950s is unsuitable for this purpose; it serves only to limit appellate review of sentencing decisions.
The profound societal transformation in recent years due to technological progress could contribute to standardi­za­tion. In other legal systems, numerous efforts have been undertaken to structure the sentencing process. In the Anglo-American legal world, the use of sentencing guidelines has been growing since the 1980s, and their intro­duc­tion was discussed in Germany. Additionally, in the United States, for example, computer-based decision sup­port systems, such as sentencing information systems and decision algorithms, have already been tested. With the incorporation of machine learning, these developments could go even further, leading to an increase in tech­nical autonomy. From today’s perspective, however, a complete relinquishing of human decision-making in the sentencing process seems to be neither possible nor desirable, at least in Germany.
The focus of this research project lies therefore on the potential advantages of a partial rationalization of the sen­tencing process. This kind of intervention in the existing system would lead to a reduction of judicial discretion and would raise the question of the role human decision-making should play in the sentencing process. While hu­man decisions may be prone to error, they may nevertheless display characteristics worth preserving in the con­text of a rationalization of the sentencing process. The best possible outcome might be a synthesis of computer-based objectivity and final decisions taken by human beings. In addition to questions of technical feasibility, this research project addresses ethical and moral objections as well as statutory obstacles.