Early in my academic life, as a doctoral student at the University of St. Gallen in Switzerland and influenced by the St. Gallen School of Information Law developed by Jean Nicolas Druey and Herbert Burkert in the late ’90s and early 2000, I began studying an issue that has received baseline attention in Europe since the World Wars, and more recently takes center stage in the intense US debates about the state and future trajectory of the digital information and media ecosystem: the concept of information quality (“IQ”). Looking at my work from back then (*) and reflecting on key developments over the past decade-and-a-half since my initial focus on this topic, four paradigm shifts emerge that have shaped my thinking about IQ as a theoretical concept and empirical phenomenon.
Information does not have an inherently positive value. An initial conceptual (and, among proponents of a “marketplace of ideas,” still controversial) shift that has profoundly impacted the understanding of IQ was prominently expressed by my early academic teacher and mentor Jean Nicolas Druey: Despite the enormous positive value that Western societies have attributed to information (“more information is better”) and gave name to the “information society,” information should not be seen as something that has an intrinsic positive value. Rather, information in the human context is what Druey calls a “Wert-Chance”: an opportunity to create value. A rich set of case studies and examples in which information in its syntactic, semantic, or pragmatic interpretation lacks quality and leads to negative effects and harms — some of them addressed by law, ranging from norms about Truth-in-Advertising to examples such as the Information Quality Act or data protection (keyword: RTBF) or anti-hate speech legislation — has been at the core of the argument that information does not have any value per se. The insight that more information is not necessarily “something good” leads to the another key point, which marks a second paradigm shift when considering the role of information in society and issues about quality: its relational and contextual nature.
Information quality is subjective and context-dependent. For decades, the application of Shannon and Weaver’s influential mathematical theory of communication back to the human communication context (notably against the intentions of the authors) led to a widespread mental model of communication as a process in which a sender ships a container with a piece of information through a channel towards a destination, where a receiver unpacks the same piece of information transmitted by the sender. Across disciplines, the sender-receiver model of communication contributed to an understanding of information as an object — with qualities that can be determined objectively: its credibility. In the late 80ies and throughout the 90ies, however, media and communication scholars, building upon a long history of constructivist theories, departed from such a “container model” and emphasized the active role of the actors involved (particularly what was formerly known as the recipient) as well as the importance of context when dealing with information. This new understanding has had a deep impact on the notion of IQ, which under such a paradigm is seen as inherently subject-bound and shaped by contextual factors — resulting in a multitude of interrelated (and at times incommensurable) quality criteria and considerations at different levels. One of the most comprehensive resulting frameworks on IQ can be found in management literature.
Information practices on the Internet shape the notion of quality. In contrast to the two preceding paradigm shifts, which are conceptual in nature, the third game-changer in this very short and incomplete modern history of IQ is networked technology: The Internet, at least in its first decades of existence and interacting with non-technical enablers, has fundamentally changed the ways in which information is created, distributed, accessed, and (re-)used. These tectonic shifts have not only led to an unprecedented amount of information, but also to a qualitative change in the information ecosystem. Particularly, traditional gatekeepers such editorial boards and professional routines, have become less important as guarantors of certain levels of information quality. The rise of new gatekeepers such as search engines and information aggregators and the introduction of new quality “control” mechanisms — ranging from reputation and rating systems to filters — offer a richer set of issues to consider when exploring IQ in the digitally connected environment when compared to the analog world (this unfinished book manuscript offers my earlier thoughts on these issues). Furthermore, the Internet-enabled turn from passive receivers of information to active users has additional ramifications: As we discussed here, the entire process in which users interact with information, including co-creation and re-use of information, is relevant when experiencing and determining the quality of information.
The quality of information is paramount in the world of AI. Big data — the availability of large-volume data sets containing of both structured and unstructured data — in tandem with recent breakthroughs in Artificial Intelligence research, particularly in machine learning, make the quality of information from which patterns and insights are extracted more important than ever. At the same time, big data quality assurance as a prerequisite of IQ becomes more challenging given the variety of data, the way it is repurposed, and the multitude of (internal and external) data sources that are typically combined. IQ in the context of data-driven technologies such as AI matters not only from a business perspective (e.g. more accurate predictions about consumer behavior or more reliable anomalies detection) or from the viewpoint of users who, for instance, benefit from better machine translation or more accurate recommendations by their digital personal assistants. It also increasingly matters from a societal perspective given the scale and growing importance of AI-based platforms across many domains and areas of life — from health and education to news media. While some of the shifts mentioned before have indicated a “turn to the human” in terms of conzeptualiations and notions of IQ, the rise of AI-based technologies fueled by big data introduces a new world in which autonomous technical systems are going to play an increasingly important role not only in the production, dissemination, and processing of information, but also in the acting upon such information. The current lack of robust accountability, transparency, and explanatory mechanisms combined with the scope and scale of autonomous systems will turn the IQ challenge into a serious ecosystem-level problem.
Across these developments and over time, the complexity of the IQ maze has increased along a number of dimensions. The following issues seem particularly noteworthy:
- Contexts: The number of contexts in which IQ has emerged as a key concern has multiplied. While much of the debate focused on issues related to media manipulation and propaganda in the last century, IQ issues in the age of social media now span — and actually blur the lines between — the private and public spheres as well as the offline and online spaces, and include content categories such as financial information, advertisement, health information, public sector information, entertainment, civic engagement, etc.
- Actors: Under the conditions of the analog information ecosystem, the quality concern centered largely on a relatively small set of (professional) information producers or a (typically centralized) medium. In the digitally networked environment, the spectrum of relevant players has broadened dramatically, including the entrance of powerful new intermediaries as well as distributed users with a broad range of interests, intentions, and agendas, making information quality a complicated compound of decentralized actions and determinations.
- Criteria: For decades, the credibility of information has ranked among the key criteria for IQ. With the increased importance that information plays in society, many more quality criteria have been introduced — according to one account, over 80 different IQ criteria can be distinguished. The plethora of quality criteria demonstrates the many different and at times competing value considerations that are at play when assessing the quality of information.
- Norms: Not only the relevant actors and potential criteria, but also the norms based on which the quality of information can be evaluated have multiplied. Whether looking at legal norms, “soft law” in form of standards, professional codes of conduct, contractual agreements, societal and interaction-born norms, or most recently norms embedded in algorithms, the realm of possible quality requirements applicable within and/or across contexts has grown dramatically, with the relationship among them still to be sorted out.
- Perspectives: IQ is approached from a growing number of disciplines, including data and computer science, and scholarship can take various perspectives, including ethnographic, normative, systematic, and prescriptive viewpoints on the topic. These discourses, using different terminologies and methods, are still largely siloed and happen at low levels of interoperability.
- Levelers: Attempts to shape the quality of information can take many forms — ranging from human-centric approaches like digital literacy education, code-based filtering systems, to solutions like labeling or rating schemes — and intervene at different levels or places, for instance focusing on ecosystem-relevant information intermediaries/platforms or individual sources of information, to mention just a few examples. The distributed nature of potential solutions to a given IQ challenge leads to the question of appropriate governance models and the distribution of responsibility among actors.
In conclusion, these observations illustrate three meta points that are worth considering: (1) The IQ challenges of our time, both in terms of problem description and the solution space, require a more integrated interdisciplinary approach, with a particular need to bridge computer science and the humanities. (2) Any approach to information quality will have to take into account both the deeply human dimension of the phenomenon as well as the increasingly important role of technological systems and applications, particularly when entering the age of AI. (3) Addressing IQ challenges will likely require a multi-layered, multi-level governance approach that blends different instruments and coordinates cooperation among various actors, with law serving as the fundamental safeguard of human autonomy and flourishing.
(*) My first law review article on the topic of information quality was published in 2000 in the Zeitschrift für Schweizerisches Recht “On the Possibility of Regulating Media- and Information Quality by Law.” In 2002, I wrote a book chapter (Festschrift Druey, in German) with the title “Variations on ‘Information Quality.’” One year later, I published the working paper “Information Quality and the Law, or, How to Catch a Difficult Horse.” In 2004, I edited a book in English entitled “Information Quality Regulation: Foundations, Perspectives, and Applications” with international contributors from the field of information law. In the same year, I served as co-editor of a special volume of the Studies in Communication Sciences on Information Quality (introduction in English available here). In 2012, I co-authored the report “Youth and Digital Media: From Credibility to Information Quality.” One chapter of Born Digital (2010 and 2016, with John Palfrey) also addresses information quality issues.