Calculating case law

  
 

Latest update May 31st, 2023

 
 

English summary 

A quantitative legal study, which can be seen as a large-scale structural analysis of WIPO arbitral decisions on domain names, is described in this book. In Chapter 1, the legal subject analysed and the research questions are discussed. The subject involves disputes about registration and the use of Internet domain names involving infringement of trade marks, trade names, or other intellectual rights. Since the emergence of the Internet, the phenomenon of cybersquatting has developed rapidly, resulting in an increasing number of infringements.

ICANN, the organisation that governs the progression of the Internet, established a system for alternative dispute resolution because parties were dissatisfied with the outcomes of national court proceedings. This system is called the Uniform Dispute Resolution Policy (UDRP) and is intended to serve the interests of all parties concerned. To date, most UN Member States have entered the TRIPS Treaty, in which cooperation with WIPO at an international level has also been established.

Through the incorporation of treaties by a large number of Member States into national law systems, registrants and registrars of Internet domain names from the countries concerned must comply with a particular clause in the registration agreement. In the event that a dispute about the domain name arises, legal proceedings are to be taken to an institution that applies the UDRP procedure.

Meanwhile, since its establishment in 1999, WIPO has become one of the largest institutions to provide the service, and over 13,000 disputes have resulted in a decision being handed down. In this study, about 4000 WIPO decisions dating from 2000-2003 were analysed, using a list of 235 factors provided by WIPO.

The research method applied was of a quantitative legal nature, and it is briefly described in Chapter 2. Examples of large-scale structural analysis of judicial decisions and UDRP decisions are mentioned as well. Quantitative legal research entails using statistical calculations in order to undertake a systematic analysis of the collected decisions. A forecast on the outcome of pending cases can be made on the basis of the results. The analysis also included the percentage of affirmed and dismissed cases; the cases in which an irregular decision was made; which factors were important or not; and whether the procedural rules were applied fairly. This demonstrates immediately the resemblance to the traditional method of studying case law.

Because of the large data set, it was clear that parts of the analysis would have to be automated rather than executed manually as was formerly done (see Chapter 3). To identify the WIPO factors in the decisions, a semi-automatic search engine that combines Boolean search techniques and similarity was used. Compared to a manual identification of over one hundred factors in thousands of decisions, the search process was simplified and substantially accelerated.

A description of the search engine findings is given in Chapter 4. As well as advantages in using the search engine, however, there also existed relative disadvantages. Certain factors were difficult to locate because they were either not unique, were not explicit or unambiguous, or were simply not recognisable because of double negatives in the legal formulation. The search process had to be refined on several occasions in order to find the correct term that fit the factor. Moreover, terms identified in the decisions as indicated by the search engine had to be checked manually most of the time because it was uncertain whether the factor was exactly the one that was supposed to be retrieved. This was the case if a term was less unique and hence more difficult to locate easily.

The retrieved results were converted into code and imported to a spreadsheet by using a software module. It is recommended that future researchers use more sophisticated and more powerful material at this point to obtain even faster and better results.

It was proven in the identification process that the list of factors from WIPO lacked objectivity. Therefore, these factors had to be removed, adjusted, or added to, leaving intact slightly more than half of the original list. The basis for the logistic regression analysis to determine the prediction function, the significant factors, and the deviant cases (outliers) of the data set was formed by this final list consisting of 170 factors.

Because of the large number of available factors – considering even that all the available decisions (4067) would be used for the analysis – it was decided to investigate the entire available dataset instead of just a sample. In order to form a stable estimation of the model, a block-wise selection was made, resulting in a block complainant, respondent, panel, and miscellaneous factors. In this way, as many factors as possible could be involved in the research, whereby the analysis was subject to as little overinterpretation as possible.

Chapter 5 provides an overview of the statistical analysis. The results of the logistic regression analysis concerning the forecast function were minimal. The model did not predict much better (a 1.6% gain) than the null-model, with 82% affirmed decisions.

However, the forecast function through the block-system slightly increased, causing a decrease of 8% in the false forecasts. In comparison with the results of other researchers who had previously carried out a quantitative legal analysis of the UDRP procedure, the results concerning the significant factors were generally confirmed or nuanced and only in some cases denied. Thus, the status of the complainant also turned out to be crucial if he had a company. Instead of proving one of the three UDRP elements, certain circumstances for that, which also are enumerated in the procedure, were significant. It was remarkable, for example, that in this research the evidence as presented by the respondent was indeed significant (paragraph 5.3.2.2 table 5.2) in the case of WIPO decisions, while other researchers found that certain institutions were biased towards complainants (paragraph 2.3.2.2.3). In addition, the number of arbitrators deciding a case did not influence the outcome of the dispute.

Furthermore, the panel factors appeared to have more influence on the outcome of the decision than those of the complainant and the respondent. Only the significance of the miscellaneous factors could dominate the influence of the panel factors, among which the number of default cases.

The number of outliers in this analysis with 144 decisions covered only 3.5% of the total data set. It became clear from the contents of a number of these deviant decisions (see Chapter 6) that in those specific cases a certain number of exceptions to the rule occurred. Hence, a respondent being a distributor obtained a ruling in favour, and even a female respondent who was in default was deemed to be in the right.

It is recommended to investigate further the number of deviating cases concerning all the decisions (13,000) made thus far. In this manner, the following questions could be raised relating to, for example, the fairness of the decision, the taking of evidence, the motivations, and the development of case law. It might also be possible that patterns exist outside the legal rules – outside the context or the legal subject – which makes the decision predictable: for example, the presence of public policy (see paragraph 2.3.2.1.1 and 2.3.2.1.3).

This study also demonstrated that the criticism of previous researchers about the functioning of the UDRP system was unjustified on certain points: for example, bias towards complainants who filed a complaint at one of the large institutions (NAF and WIPO). The observation that the rules concerning the burden of proof and the standards to be applied in the assessment of the parties’ contentions were not clear was more plausible. Despite ICANN’s attempt to incorporate as many circumstances into the UDRP as possible, the legal substantive interpretation of the outliers concerning proving certain circumstances revealed that lacunas were filled by WIPO case law.

Moreover, it also showed that the substantiation of panel members was frequently poor or even incorrect and partially omitted. Furthermore, the parties’ contentions in the publication were summarised briefly, while the panel findings and discussions indicated that exactly the details omitted from the parties’ contentions showed detailed circumstances that directed to a specific decision. However, the results also showed (as found by other researchers) that the UDRP procedure still needs improvement in some areas. It is recommended that the points mentioned above be examined more closely. Generally speaking, the question suggests itself as to whether the fundamental rights for arbitration, such as fair trial, are likely to be sufficiently guaranteed. After all, the complaint is treated electronically – without the characteristically arbitral hearings (cross-examination). Further research on this subject is also desirable.

One of the most important research questions with regard to the execution of this large-scale structural analysis included the possibility of automating the process by which case characteristics that play an important role in the outcome of a legal decision can be discovered. By means of the answers and results in the aforementioned paragraphs and chapters, the answer to this question can be confirmed. Moreover, considering the scarce technical resources available for this research, this large-scale structural analysis was deemed to be a success.

The book and research as shown above is written in Dutch, only an  English summary is available. See academia.edu.

Relevant New Developments

Domain Name Hijacking of Scientific Journals

It has recently been discovered that expired domain names of defunct or renamed scientific journals are being purchased by hijackers and used for fake versions of these journals. This domain name hijacking enables quick publication of fake versions that can be included in reference databases such as Web of Science or Scopus, making it attractive for researchers who seek to advance their careers quickly. For the hijackers, this is a profitable business, with one fake publisher earning a million dollars in 2021 by “publishing” over 5,000 articles in a fake version of Annals of the Romanian Society for Cell Biology.1


Verwijzingen, bronnen en links

  • 1
    See the article of May 31st 2023 from NRC Handelsblad, “Expired domain names of scientific journals can be a goldmine”. The article discusses how hijackers are purchasing expired domain names of defunct or renamed scientific journals and using them for fake versions of the journals. This enables researchers to quickly achieve a higher status by publishing in these fake versions, which can be included in reference databases such as Web of Science or Scopus. However, this practice jeopardises the reliability of scientific research. For instance, in 2021, one fake publisher earned a million dollars by “publishing” over 5,000 articles in a fake version of Annals of the Romanian Society for Cell Biology. Weblink to the article: <https://www.nrc.nl/nieuws/2023/05/31/verlopen-domeinnamen-van-wetenschappelijke-tijdschriften-kunnen-een-goudmijn-zijn-a4165959>