Creating Better Disclosure Norms through Machine Learning Algorithms

Autor: Fabiana Di Porto
Rok vydání: 2020
Předmět:
Zdroj: SSRN Electronic Journal.
ISSN: 1556-5068
DOI: 10.2139/ssrn.3705967
Popis: During the past decade, a small but rapidly growing number of Law&Tech scholars have been applying algorithmic methods in their legal research. This Article does it too, for the sake of saving disclosure regulation failure: a normative strategy that has long been considered dead by legal scholars, but conspicuously abused by rulemakers. To illustrate in what way Machine Learning (ML) algorithms and Natural Language Processing (NLP) tools could reframe how to create better disclosure norms, this Article presents an innovative approach, which combines classical text analysis (Phase one), with a behavioral one (Phase two). While existing proposals mainly focus on the implementation of disclosure duties, seeking to simplify the policies framed by the industry to reduce consumers’ costs of reading, this Article develops a ‘comprehensive approach’, thus filling a void in the Law&Tech scholarship. It outlines how algorithmic tools can be used in a holistic manner to address the many failures of disclosures from the rulemaking in parliament to consumer screens. Hence, in Phase one, after conceptualizing both disclosure rulemaking and implementation (the policies used by the industry) as text datasets, termed de iure and de facto disclosures respectively, I propose to set a library of indexes to measure their failures. NLP and ML technologies are then used to combine and rank the texts to finally select those that ‘fail’ the least: these are the Best Available Disclosures, or BADs. I assign the case-law a special ‘linking’ role (between the de iure and de facto disclosure datasets): court decisions provide for constant updates of our libraries, so that new terms can raise that are disputed, while others can become settled and undisputed. Only the latter, among the case-law, would feed the BADs algorithm with fresh data to produce greater legal certainty. In Phase two these BADs are tested by real-life consumers and industry representatives in a ‘regulatory sandbox’, which will allow to generate behavioral data. This step is as innovative as it is crucial: even if a disclosure is well-understandable, coherent and complete from a theoretical point of view, this does not automatically mean that it will be well-received by all consumers or suitable for all the industry needs. In fact, different groups of consumers might have diverse capabilities and preferences with regards to how and when to receive information as well as to what information to receive. A regulatory sandbox opens the unique opportunity to identify different groups to tailor the way how they are presented with disclosures, which should make the latter significantly more efficient, as similar approaches used in online advertising suggest. With the help of ML algorithms, this data will be used to transform the tested BADs into the ‘best ever disclosures’ (BEDs) to be implemented automatically by the industry, thus significantly decreasing the cost of complying with disclosure duties.t must be noted that such ‘algorithmic rulemaking’ proposals are frequently met with well-justified skepticism regarding their compatibility with transparency requirements and other democratic principles. However, since the sandbox stage constitutes a transparent process offering the possibility of intervening with and correcting potentially problematic algorithmic norms, the above-illustrated proposal duly takes into account such concerns. As a whole, the Article presents a comprehensive, balanced and innovative approach aimed at ensuring that we do not drown in the sea of available information by transforming the source of the problem, the omnipresence of data, into its solution: data-driven, Algorithmic Disclosure Regulations.
Databáze: OpenAIRE
Externí odkaz: