Text and Data Mining exceptions and limitations in copyright

Do you remember last summer, when the war on memes took place? When you just simply had to sign that petition to #SaveYourInternet? I would like to take you back to the legal instrument that created so much controversy and focus on two articles that got very little attention yet that substantially change the legal situation they govern.

Articles 3 and 4 of the Directive on Copyright in the Digital Single Market (DSM Directive) did not create a grand upheaval in the public sphere for the simple reason that it does not affect your own daily internet experience. Instead, the provision impacts the attractiveness of the European Union in relation to certain start-ups, scientific research, private research establishments and journalism & information institutions among others.

In order to understand the implications, we must first define what Text and Data Mining (TDM) is. In recital 8 of the DSM Directive, we find that TDM is “the processing of large amounts of information with a view to gaining new knowledge and discovering new trends possible. Text and data mining technologies are prevalent across the digital economy […] However, in the Union, […] organisations and institutions are confronted with legal uncertainty as to the extent to which they can perform text and data mining of content.”

This was the first time a legal instrument from the European Union concerning copyright actually defined TDM, despite the technique having been used since the 1980s in government intelligence.

All that is very well and interesting, but you would be entitled to ask us what the link between TDM and copyright is. TDM is used in many different areas but let us take the example of scientific research as an example.

Research papers are copyright protected, as they are the tangible expression of an idea and the intellectual creation of the author. A new paper is published every 30 seconds in the global research community, making it basically impossible for a scientist to read all the relevant literature in the field before starting their own research! That is where TDM comes into play generally – when there are large swathes of data that need to be sorted and analysed. This technique has the potential to infringe some of the rights given by copyright because it often requires copying, downloading, using and maybe even replicating thousands of scientific articles – rights that may be reserved to the author and that require the author’s authorization. Asking for consent is simply not be a viable option when dealing with such huge numbers of potential copyright owners. If you are further interested in how copyright and TDM interact specifically with the scientific publication community, this video is great.

Coming back to the regulation of TDM in copyright more generally, we will first look into the previous Directive that regulated the situation. The Harmonisation of Copyright in the Information Society Directive (InfoSoc Directive) did not have any explicit mention of TDM, and instead the people wanting to make use of TDM for any purpose had to infer and guess whether they were within the law or not. Article 2 of the InfoSoc Directive gave the copyright owner the right to allow or refuse the reproduction of their work, including direct or indirect, temporary or permanent, in any means and any form, in whole or in part.

However, Article 5(1) of the same Directive contains the one mandatory exception in the whole Directive: when there is a temporary act of reproduction that is absolutely necessary for the only purpose of either a transmission in a network (think of your internet service provider having to reproduce the data for you to see this article), or a lawful use of a work. This exception is also only valid as long as the reproduction of the work does not have independent economic significance. The consequences of this exception in the InfoSoc Directive meant legal uncertainty, as found in Recital 8 of the new DSM Directive, for the many industries that would benefit from TDM. Any commercial entity such as Google or IBM trying to develop an AI through the use of TDM was however nearly assured to be infringing the law if they were going to benefit from it afterwards, as per the requirement of no independent economic significance.

The new DSM Directive changes things for the scientific research community, as Article 3 of the Directive now explicitly allows some forms of TDM for scientific research purposes. This includes techniques for reproduction and extraction, which must be made free by the copyright holders, as well as allowing the storage of the data – this last part is especially important in scientific research, as the results must be reproducible and explainable, which is a lot easier if you have the data you used at hand.

Article 4 of the DSM Directive deals with the rest of the industries that do not use TDM for scientific research purposes, but instead mostly for commercial purposes. There are of course other uses for TDM, such as Smart Disclosure Systems whose aim is to provide consumers with improved access to the data needed to make informed decisions for example. Users who do not have a scientific research purpose may therefore use TDM techniques to reproduce and extract works that they lawfully accessed and even keep the data as long as necessary for the purposes of TDM. This is very similar to Article 3 so far, except in relation to the data retention period.

The biggest difference between the two articles is that copyright holders can opt out and forbid the use of TDM for their work. An appropriate manner for a copyright holder to ‘reserve’ their right would be through the use of a robot.txt protocol (which very basically means that you make the webpage not readable for an algorithm, hence your work will not be included in the TDM process) or a technological protection measure. Other means of reserving copyright could include through contracts or licences, but apart from the first mentioned (robot.txt protocol) nothing is certain yet. This leaves TDM users falling within the scope of Article 4 at the mercy of copyright holders. For now however, the current uncertainty from the wording of the article also exposes content owners to TDM users who may claim that the owner’s reservation was not done in an appropriate manner. This matter will ultimately have to be dealt with and the appropriate manner required to reserve the right further defined in the Court of Justice of the European Union.

The Crime Prism: a visual tool for defining crime.

Have you ever wondered how a particular conduct ought to become qualified as a crime? What factors are taking into account by the various interest groups involved in the criminalisation process, when assessing such a thing? Those questions are precisely the ones that criminologists and legislators have to reflect on.

A common-sense answer would be to look at the legal definition of crime, as criminal law specifically gives behaviour its quality of criminality. It defines crime as an act/omission prohibited by criminal law, committed without defence/justification, and liable for sanctioning by the criminal justice system, or actually sanctioned. This justification is not as clear-cut as it seems. Indeed, relying on a strict legal definition of crime would fail to encompass those harms that are covered by administrative law, such as white-collar crimes.

Moreover, another significant issue with the legal definition is that it ignores the truth that the law is contextual; what counts as a crime in one particular place at one specific point in time may not qualify as such in another legal system at another point in time. Take the example of prostitution; while some countries have adopted a prohibitionist approach, criminalising all the activities and parties involved in the conduct, others have accepted prostitution as a social fact, which needs to be strictly controlled through regulation for the protection of society. Thus, what is crime in the legal sense varies, depending on diverse contextual factors such as the historical context, the social setting, the location, and so on.

Last but not least, one should keep in mind that criminal law is largely shaped by the interests of powerful groups in society. Those groups (politicians, lobbyists, corporations, etc.) have the ability to influence legislation through their status and consequently, the so-called “crimes of the powerful” (ie: corporate crimes) often go unrecognised and their criminalisation may be considered as minimal. With these three major limitations in mind, it is easier to conceive why there is little agreement among criminologists about what really constitutes crime.

Various approaches are being used for this purpose, each of them presenting its own advantages and boundaries. This paper will submit one of those approaches as a powerful visual tool to, at the very least, argue whether a particular behaviour should be labelled as crime, or not. This conception was first provided by the criminologist John Hagan in his famous pyramid of crime. He integrated three dimensions in the pyramid; the degree of agreement among people about the wrongfulness of an act, the severity of the society’s response in law (reflected in the sentences set by the law) and finally, the social evaluation of the harm the act inflicted on others.

Even if this view has the advantage of encompassing more harms and contextual factors than a strict legal definition of crime, Lanier and Henry submitted that it is nonetheless incomplete. According to them, Hagan’s pyramid firstly neglects the public awareness of crime. It is not exceptional that some crimes, even if involving a certain degree of harm, may not be recognised as such because the victims do not realise that they actually have been harmed; this tendency is mostly reflected in victims of government and corporate crimes. A second missing dimension is the extent of victimisation as such. Indeed, it is easy to perceive that a society’s perception as to the seriousness of a crime can be influenced by the number of people harmed by the conduct. Lastly, as to Hagan’s dimension of the severity of the society’s response in law, Lanier and Henry argued that it failed to acknowledge the probability or likelihood that an offender will receive a serious response, even when the law sets such a penalty. Unfortunately, crimes of the powerless (ie: street crimes, rape, theft, etc.) are much more scrutinised by the various legal agencies involved in the process of criminalisation as compared to the crimes of the powerful, where the state is commonly involved.

In an attempt to resolve those inadequacies, Lanier and Henry have proposed a re-defined conception of Hagan’s pyramid: the Crime Prism. It is composed of five dimensions, each of them subjected to changes in degree depending on the particular conduct/crime. Those are the individual and social harm that resulted from the conduct (does it affect one specific individual or the society as a whole?), the visibility of the crime (what about invisible crimes, such as environmental crimes?), the degree of consensus in society about the wrongfulness of the crime, the extent of victimisation (many victims vs. victimless crimes) and finally, the probability of severe social response (is the criminal a repeated offender?). It is however important to keep in mind that the prism offers a theoretical conception of a crime’s definition. It definitely still rests on the challenge that two persons with different life experiences may locate a particular conduct on two different spatial points on the prism, as they may not agree on certain dimensions; for instance, the victim of a crime may perceive its seriousness in a much more absolute way than the one that has never been confronted to such a conduct. With the dimensions in mind, you can now consider the following illustration in order to better visualise the conception of the prism:

As you can see, toward the upper half of the prism are crimes that are obvious as to the seriousness of it, highly visible to the society, extremely harmful and noncontroversial as to the society’s response in law, as the punishments attached to those crimes are the most severe one can receive. Crimes such as terrorism would undeniably be located on this part of the prism. The more the degree of the different dimensions decrease, the lower the crime will be positioned on the prism.

In the middle of it, but still through the upper half, are violent acts of individual crimes: those crimes are also largely perceived as serious, the societal reaction is severe and so are the sanctions attached to those crimes, even if fewer people are hurt. Acts of rape, homicide, incest, but also robbery, burglary or vandalism are usually positioned toward this spatial location. At the center of the prism is where social deviations and diversions would best be placed; those include for example public drunkenness and crimes of juveniles. There is little harm inflicted and those acts are generally accompanied by a lack of consensus about their criminalisation.

Through the center to the very lower half of the prism can be positioned conducts where the obscurity of their criminal character is more and more apparent. The harm may be relatively hidden, there is a lack of consensus (one may even talk about conflict) over the criminal definition, the seriousness of the society’s perception will differ among the various interests groups and the reflection in punishment is much less severe (ie: fines). The context is therefore particularly relevant for those crimes.

At the final level are those crimes that are so obscure as to rarely be punished and many consider those acts as not deserving their criminal quality. Sexism, for instance, would probably fall under this category.

Now that you have a better understanding of how to integrate the dimensions of the prism, you can try to apply it to different acts on which you wonder whether the criminal label is legitimate or not. As you will see, behaviours such as sexual harassment, marital rape, domestic violence, and so on, are interesting examples of crimes that will be located at different points of the prism depending on the context (some consider those as normal while others are largely outraged and agree upon the wrongfulness).

One last word should be given to the fact that crimes of the powerless (street crimes) are mostly positioned on the upper half of the prism while crimes of the powerful are located on the lower half of it. Noticeably, holding powerful positions in society influence the degree of the prism’s dimensions!

The Urgenda case, a delicate balancing of interests.

On the 20th of December, a Supreme Court ruling will be held in the Netherlands on the controversial case “Stitching Urgenda v. De Staat der Nederlanden”.

The dispute faces Urgenda “Urgent Agenda”, a foundation concerned with the development of plans and measures to prevent climate change, against the State of the Netherland. The Foundation demands the State to further its ambitions to reduce the emission of greenhouse gas to 25 percent compared to 1990. The foundation initially sued the State in the District Court of The Hague in 2015 then the State appealed in the Appeal Court of The Hague in 2018. Both times, the courts have supported the climate activist foundation.

The particularity of this case lies in the fact that the Court is ordering the State to achieve a specific goal, it is giving the State an order. This peculiarity might arguably be on the edge of breaching the “Trias Politica” (separation of power between the legislative, executive and judiciary). The court usually holds the State accountable by means of a judgment basically concluding, “You acted illegally, you committed a tort, now pay damages". This post will particularly focus on the crucial argument that will be brought by the State in its last chance to over-rule the previously held court decisions, namely the “Trias Politica”.

The Trias Politica underlines a crucial need to keep a separation of power in a democratic society. The rationale behind the separation of powers is to prevent concentration and therefore an abuse of power. The separation of power in question in the case at hand, possibly concerns the judiciary over-stepping the legislative. The legislative power, according to the Dutch constitution is composed of the government and the States General (Parliament).

Back in 2003, the Supreme Court raised the Trias Politica question in the Waterpakt case. There, the court ruled that even if the state acted unlawfully, Dutch law precludes the court from ordering the State to enact legislation in order to rectify the situation. After all, laws are, pursuant to art. 81 of the Dutch Constitution, adopted by the government and the States General (Parliament), who together form the legislative power. The question as to whether, when and in what form of law, the law will be enacted must be answered based on political decision-making taking into consideration all the interests involved. The balancing of interest must be made by a democratically elected body. This ruling is an influential precedent to the issue at hand.

In the Urgenda case, in paragraph 30, the State brought this argument back on the table. It argued that there is no need to reduce by 25% to 40% greenhouse gases by the end of 2020 per se. It thereby meant that it is to the discretion of the State to choose what it considers the most appropriate for the needs of its electors, taking into consideration all the interests involved, such as those of industry, finance, energy supply, care, education, and defense. It is a "political question". It would be acting in opposition to the Trias Politica for the Judge to impose such an obligation on the State, especially since it can only be achieved through engaging in a legislative procedure, meaning that the judgment materially means an order to legislation.

However, the Court argues that Urgenda's claim is not intended to enact legislation and that the State retains complete freedom to determine how and in what form it will comply with the order demanded, thereby loopholing the conditions previously established in the Waterpakt case. It further develops that if it were to be judged that execution of the imposed order is only possible through the adoption of legislation, the order does not prescribe in any way what content this legislation should have. For this reason, there is no "warrant for legislation". The court goes on explaining that the State has not substantiated sufficiently well why the execution of the imposed order is only possible through the adoption of legislation. Indeed, whilst Urgenda has proposed different solutions for achieving their demand not requiring any legislative measures, the State insufficiently refuted their suggestion.

The last point raised by the court as a justification for enforcing the demand of Urgenda, arises from the relatively newly established Article 93 of the Dutch Constitution (hereinafter, GW). This article sets that some international law provisions can be used directly in a Dutch court. This particularity of the Dutch constitution implies that Urgenda can make effective use of its rights derived from the European Convention of Human Rights (Hereinafter, ECHR). Therefore, the Court of Appeal is required to apply directly effective provisions of treaties to which the Netherlands is a party, such as Articles 2 and 8 of the ECHR. Such provisions form part of the Dutch legal sphere and they even take precedence over Dutch laws that deviate from it.

The implications of The Hague Court of Appeal’s judgment might be perceived as unprecedented and ground-breaking for Dutch Constitutional Law. It will have a major impact on the judiciary-legislative interplay. On the 20th of December, a crucial decision will lie in the hands of the Dutch Supreme Court. The dilemma simply put, boils down to a trade-off between a crucial factor in a functioning democracy and the need to take drastic measures for the sake of climate emergency.

This balancing of interest raises some discussion. It has been pointed out by academics that the collective action right exercised by Urgenda Foundation, might be a new way for citizens to have their voices heard beyond checking some boxes during election time. On the other hand, what if the Supreme Court upholds the Court of Appeal’s judgment, will this lead to a floodgate of future judgments? Will it lead to an abuse of power by the Court?