International Legal Scholarship and the Challenge of Digitalization

Thank you very much, Lotte, for these very kind words. I feel honored to speak here to you today and to discuss a question that keeps me up at night. And this question is, what challenges does digitalization pose to our discipline to international legal scholarship? Now, note that I will not talk about international law as such, and Lotte has already warned you. So be warned if you want to talk about international law proper, then you're probably not in the right place here. I'm talking about international scholarship, or if you wish, some would even call it international legal science. So this is my topic, and what is digitalization doing to our discipline, to the discipline of international legal scholarship? Now, I want to argue that the main challenge of digitalization to international legal scholarship is a methodological one. And so the methodological challenge I want to put up front here. Now, the structure of my talk is as follows. I want to take up two questions from my paper, which are, first, what does digitalization mean for legal scholarship in general? So not just international legal scholarship, but legal scholarship in general. And the second part, I will look at the question of how data analysis can be gainfully, usefully applied to questions of international legal scholarship. Now, I come to my first part. So what does digitalization mean for legal scholarship in general? And here, I would like to raise three points. So digitalization is or has turned, I would maybe even say has turned, into a key scientific impulse also for legal scholarship. And as I said, I want to raise three points related to digitalization in our context. And the first is that we see processes of datafication or digitalization of law. The second point is we see a gradual rise in the automated information retrieval on law. And thirdly, as a result of all of this, and this I consider most important, is legal knowledge is becoming more and more differentiated. Or you could put this maybe in easier words, our discipline is becoming more and more diverse, pluralistic in its approach. Now, on the first point, on the datafication of law, first of all, digitalization has to do with the massive production and the massive availability of digital data. Digital data is, as you all know, is computer readable data. Now, the question that I need to answer first as a lawyer is, does it make sense at all to talk about something such as legal data? Does it make sense to talk about legal data at all? And what the hell would legal data then be? Now, it is highly controversial whether there can be such a thing as legal data. Some legal scholars would reject outright the idea that legal data exists. And these critics usually point to the fact that law is not simply out there. It has no objective reality. Rather, they see law as a social construct. Law, in their view, is not observer-independent. It is not measurable. Now, while I think that this has credit, this view, I think it's an oversimplification. Law is, in my view and in that of others, multidimensional. For example, the German legal philosopher Robert Alexy speaks of the double nature of law. Law, of course, has this critical dimension, this ideal or critical dimension. Any legal law makes a legitimacy claim. But law certainly also has this other feature of a social or factual dimension. And this also applies, of course, to international law. International law also is socially effective. And this social effectiveness of international law is also empirically observable. Law always is also a social practice. Take the decision of a court. Take even the stretching of hands at a wine auction. These are all manifestations of the social practice of law. Now, this social practice of law manifests itself in legal texts, in legal relationships, and in legal decisions. In other words, texts, relationships, and relationships and decisions contain observable legal data. Here's a definition that I find useful by a legal sociologist, Hubert Rotloidner. And he says, legal data are law-related social facts. Now, the datafication of the law. At present, the datafication of law takes place at a rather modest scale with rather modest objectives. Various actors are currently pushing for the datafication of law. Take publishing companies that also publish e-books. Take Google Books and libraries. They're all involved here. But of course, and even more important, the central actors of the legal systems, such as courts, such as the legislators, and of course, the administration, all of them offer their decisions, their texts, online as computer-readable data. In Germany, starting from 2020, all files with the federal administration must always be in electronic form. I consider this is a huge step. Once all files are also or only available only in electronic form, this is a game-changer, in my view. And this is happening as we talk. Finally, and this is my main focus here, legal scholarship is also increasingly involved in producing legal data, involved as an actor in the datafication of law. Scholars collect legal data. For example, we in my research group, we code judgments by the European Court of Human Rights and we turn these texts into computer-readable data. Now, once you have these computer-readable data, what happens next is you apply methods other than the traditional hermeneutic text-based methods to these data. And here, I will talk about the automated information retrieval about the law. And these methods obviously come from quantitative methods, such as news and statistics and econometrics. I will come to a few examples in a minute. Now, I said that the most important point for me is that the third aspect of digitalization, and this is that the automated information retrieval, by which we now gain data-based legal knowledge, we get a new form of legal knowledge, which we didn't have before. And what is the added value of such a data-based legal knowledge? Data-based legal knowledge allows us to better understanding how international law works. The expectation is that an improved knowledge of the social dimension of international law is meant to provide a better basis for improvements in the lawmaking and decision-making processes. And I will come to examples later on. Now, I admit that much of what we are at the moment expecting, what would come from out of these legal data, is only speculation. All we can probably say at this point is that we can expect to see new correlations, new patterns, by looking at these data, which then triggers new, innovative legal scholarship. And what I found most striking is that this idea is all but new. It was actually coined 100 years ago by the American legal scholar Herman Olyfund, who said that, and I quote, our case material is a gold mine for scientific work. It has not been scientifically exploited. We should critically examine all the methods now used in any of the social sciences and having any useful degree of objectivity. Now, let me come to the second part of my talk. And this is about the question, how can we apply data analysis to questions that we ask in international law and international legal scholarship? For us legal scholars, it probably makes most sense to start by digitizing legal texts and see what we get out of this. For example, we might want to digitalize international treaties and then examine the resulting textual data. Text mining is a quantitative method that is used for information retrieval on digitized legal text. Text mining can be used to identify meaningful structures in complex data. Here you can see an example of a so-called word cloud. And this is often the starting point for further research. This is just a very simple thing, very simple starting point. If you have lots of treaties or unstructured texts, then text mining is the first useful step to get into the patterns that are in there. One question that interests international legal scholarship, for example, is this. What is the influence of one legal regime on another? And how closely are the two related already on a textual level? And text mining can be used here. One example of a fine study is a recent study by Ali and others that uses text mining for the study of WTO law. And they find that contrary to what is sometimes suspected in the literature, multilateral WTO law is very much present in preferential trade agreements of individual states. So here this text mining method is a useful tool to get a better grip at the fragmentation discussion that is ongoing in the study of WTO law. The second point is legal relationships as data. Not only text, but also legal relationships are leading in this respect. And they can be used for obtaining database information. Relations between actors or relations between acts. And here's an example from my own research, which is a joint project by the University of Zurich and the Max Planck Institute of International Law in Heidelberg. Now, the project asks the question, basically, which violations of the European Convention on Human Rights typically occur together. So if you read hundreds of decisions by the European Court of Human Rights, you will see that there are violations that typically go together, that hang in a way together. You always see the same pattern. But you see it. You somehow know it. But you need to have other tools than just the hermeneutic text reading to actually say, well, these are typical connections or typical structures in the data. And here, the so-called cluster analysis, which is a statistical method, can be gainfully used, I think. This dendrogram, so it is called, shows four clusters in the data and the very first data that we collected. And I will only briefly outline one of the clusters that are interesting here. You see that in the bottom part of the dendrogram, you see that freedom of expression and independent tribunal violations go together in a way. And the data shows that if the freedom of expression is violated, the right to an independent court is also violated in 30% of the cases of our sample. This is our data. For example, one of the cases that is in the sample here is that the complainant distributed leaflets on the subject of conscientious objection to military service before the state security court in Ankara. And the military judge was later involved in the conviction of the complainant. And the Strasbourg court came to the conclusion, Article 10 is violated and Article 6 is violated. So this is one of the clusters that we find in our data here. And we can then talk about, OK, what happens if we know that there are these clusters? You can, for example, cluster on the basis of the state. And I find this very important. If you know the typical violations that occur in your country, then you can possibly train prison staff. You can think of all sorts of policy implications by knowing the clusters of violations in your country. OK, so this is as much as I wanted to say on the cluster analysis. Then a final example here for automated information retrieval is about legal decisions. There are many, many studies on legal decision making in an empirical dimension. And these studies typically examine the statistical relationship between a response variable and one of various explanatory variables. So they try to find associations between these two categories of variables. And these can be all sorts of things, gender. OK, so I will hurry up here. How much time did you say? A minute. A minute, OK. So what these studies do is a so-called regression analysis. Typical research questions are, and here I will take another one of our studies, we were interested in the question, how does the European Court of Human Rights arrive at its sums of non-pecuniary damage? And what are the factors related to the amount of money that you get out of these judgments when there is one or more violations? So we, in a way, tested empirical doctrine here. And this is what I find important. We can talk about this in the discussion. So we tested empirical doctrine here. We tested legal doctrine empirically. One point on, a final point on, a final point on what's on differentiation of legal knowledge. Legal knowledge is becoming more differentiated because what you can do with this is you can do predictions. If there is a client coming to you and says, these are my violations, how much do I get for non-pecuniary damage? You can pretty accurately, if the data is good, say this is the amount of money you get. So I conclude the problem, the key problem to me by post-digitalization on international legal scholarship is one of method. And I think in the resulting discussion, we should talk about, OK, what happens for teaching? What happens for academia, for in the classroom, if we accept that there is a problem of method involved here? Thank you very much. Thank you.