Algorithmenkompatibles Verwaltungsrecht?

Die Übersetzung von Recht in (Computer–)‌Code ist derzeit in aller Munde. Lawrence Lessigs berühmtes Diktum, „Code is Law“ wird neuerdings dahingehend reformuliert, dass „Law“ auch „Code“ sei, dass man bei der Rechtsetzung also zugleich seine rechentechnische Umsetzbarkeit mitzudenken habe. Einen Ansatzpunkt für eine derartige „Algorithmisierbarkeit“ von Recht bietet § 35a des Verwaltungsverfahrensgesetzes des Bundes, wonach „automatisierte“ Entscheidungen in bestimmten Fällen zugelassen werden. Ein aktuelles Papier des Fraunhofer FOKUS-Instituts unter dem Titel „Recht Digital“ denkt dieses weiter und suggeriert, man müsse nur die passenden, eindeutigen Ausdrücke finden, dann sei Recht gleichsam „programmierbar“. Aber genau hier stellt sich das Problem: Rechtssprache ist eine Multi-Adressaten-Sprache, also eine Sprache, die sich ebenso sehr an ein Fachpublikum wie an Laien (Bürgerinnen und Bürger) wendet. Sie ist zudem kontextabhängig. Der aktuelle Hype um den Begriff der „Algorithmisierung“ von Gesetzen verbirgt zudem, dass es sich hierbei um ein Grundproblem von Rechtssprache handelt, das in den 1960er bis 1980er Jahren unter den Paradigmata „Rechts-/Verwaltungsautomation“ oder Rechtskybernetik verhandelt wurde. Wie kann man sich also dem Problem der Kontextabhängigkeit von Recht unter dem neuen Paradigma der Algorithmisierung nähern? Im Beitrag über „Algorithmenkompatibles Verwaltungsrecht? Juristische und sprachwissenschaftliche Überlegungen zu einer ‚Standardisierung von Rechtsbegriffen‘“ werden verschiedene Zugänge zur Schaffung einer „algorithmenkonformen“ Rechtssprache vorgestellt. Letztlich aber vermögen es noch so ausgefeilte technische Methoden nicht, das Problem demokratischer Deliberation zu verdrängen – über die fundamentalen Fragen einer Algorithmisierung der Rechtssprache muss der unmittelbar demokratisch legitimierte Gesetzgeber entscheiden. „Kontext“ und „Text“ geraten insoweit in ein wechselseitiges Abhängigkeitsverhältnis. The translation of law into (computer) code seems to be currently on everyone’s lips. Lawrence Lessigs’ famous dictum “Code is Law” has recently been rephrased saying that “Law” was also “Code”. This means that the wording of laws should directly take their “computer implementability” into consideration. A starting point for those postulations can be seen in the (relatively) new section 35a of the (Federal) Administrative Prodecure Act (Verwaltungsverfahrensgesetz), which allows “automatic” decisions in specific cases. A new paper of the Fraunhofer FOKUS institute takes this up and suggests that we have only to look for the appropriate, unambiguous term that corresponds with an unequivocal legal meaning. In doing so, law could be programmable. But this is exactly the point where the problem arises: laws have more than one addressee; they address lawyers as well as citizens (mostly laypeople). Furthermore, legal terminology is context dependent. The current hype regarding the “algorithmization” of legal terminology also hides the fact that this issue was – more or less – discussed once before under the paradigm “legal cybernetics” between 1960 and 1985. So how can we approach the problem of context-dependency of law under the new paradigm of algorithmization? In our contribution on “Algorithm-compatible administrative law? Legal and linguistic considerations concerning the ‘standardization’ of legal terminology”, we will introduce different approaches to safeguard the compatibility of law with computer technics. But how sophisticated a technical method can be: It is the democratically legitimised parliament that must make the fundamental decisions when it comes to an “algorithmization” of legal terminology, because there is no text without context.

Put Dialectics into the Machine: Protection against Automatic-decision-making through a Deeper Understanding of Contestability by Design

This paper endorses the idea that the right to contest provided for by art. 22, § 3 GDPR, actually is the apex of a progressive set of tools the data subject has at his disposal to cope with automatic decisions and it should work as an architectural principle to create contestable systems. But in order to achieve that important role, it cannot be reduce to the right of human intervention, also provided for by art.22, § 3, nor to a generic opposition to the outcome of the automatic processing. Thus, drawing from a thorough analysis of the relationships among the rights included in art. 22, § 3 GDPR as well as from the juridical proper meaning of “contestatio”, it is concluded that the right to contest has its own proper nature as a hybrid substantial-processual right that is able to give concrete shape to all the other rights indicated in art. 22, § 3, included the much discussed right to explanation.

Crowdfunding – the Interplay between Evolving Organization and Embedding Institutions

This paper explains the functioning of economic crowdfunding (equity- and loan-based). One of the angles from which crowdfunding can be analysed is institutional economics. Namely, it can be studied as a case where the spontaneous evolution of contractual forms is difficult and the need for legislative intervention arises. The paper explains the divergence of interests and risks of three groups of agents involved (platforms, fundraisers and funders). The radical asymmetry of power, to the advantage of the platform, is inherent in this type of business. The spontaneous development of strategies by the platforms (pricing and some non-price strategies relating to fundraisers) and recent developments in the direction of automatic decisions taken by the platform instead of funders, promotes the development of the crowdfunding business, but it does not ensure protection from risk of other parties. Still, the major source of risk for funders and partly also for fundraisers is the absence of liability of the platforms themselves. In these conditions, the spontaneous move towards the better alignment of the interests of parties seems unfeasible and external regulation is necessary.

The Sample Size-Richness Relation: The Relevance of Research Questions, Sampling Strategies, and Behavioral Variation

Recent discussions of the relation between sample size and assemblage richness suggest that researchers should control for that relation before behavioral interpretations of richness levels can proceed. We suggest that such automatic decisions should be avoided because a strong correlation between sample size and richness alone is not sufficient evidence for dismissing behavioral factors as the primary causes of the variation in richness. Instead, we must first explore the causes of the observed variation in sample size and the relation of those causes to our research questions.