Originally published on the University of East Anglia Information Society Policy Blog, October 2018
On the 12th of September, the updated wording for the Proposed Directive on Copyright in the Digital Single Market was approved by the European Parliament. This follows the vote rejecting the Parliament’s mandate on the 5th July, sending the Proposed Copyright Directive back to the drafting stage. Two measures within the proposal have attracted the most attention: the misleadingly-titled ‘link tax’ in Article 11, which passed with 58% of the vote, and the commonly-named ‘upload filters’ provision in Article 13, which passed with 55%. I will analyse each of these articles in turn before concluding that that the vision the EP has for the future of the internet is unlucky for everyone concerned and explain what happens next.
Voices heard
This Directive is one of the most debated pieces of legislation on intellectual property to come out of Europe in recent times, and should be seen within the context of international legislation aimed at implementing territorial borders to the web in order to apply national laws to internet users (such as the GDPR in the EU, and net neutrality in the US). Those in favour of the proposals argue broadly that the law currently fails to sufficiently prevent copyright infringement online, and that there is a ‘value gap’ between those who invest in the generation of creative works and those who exploit them online without paying. Artists such as Paul McCartney believe the Directive is necessary to “assure a sustainable future for the music ecosystem and its creators, fans and digital music services alike”. This statement is laudable, as is the intention behind it to protect the income of creative individuals in a time when the vast majority are suffering from low returns on their efforts. Few would disagree that artists should be remunerated for infringing reuses of their works online. The argument surrounds the practicality of the suggested methods for doing this – especially in regards to upload filters.
It has also been argued that rather than increasing intervention into online reuses of copyright works (as this legislation seeks to do), artists should work with online users and use other non-legislative methods such as licensing to control their work (as Wyclef Jean stated earlier this month: “Don’t tear down the building, be the landlord”). There are a variety of options available now to artists who wish to permit specific reuses, such as the GNU General Public License from the Free Software Foundation specifically for software, or the user-friendly selection applicable to all types of creative works available through Creative Commons (CC). There is a need for more licensing options like these and an improvement in how they link with platforms and content. A strong example to build on could be the relationship between Wikipedia and Creative Commons, which asks users to choose from the list of CC licenses when uploading media.
The Directive aims at protecting the rights of copyright holders, to protect incentives to create. However, this myopic fixation on the profit-maximisation incentive fails to understand that much user-generated creation in the digital era is utility-maximising rather than profit-maximising – done for love, not money. Much of the furore regarding this Directive has focused on this mostly non-commercial type of creation, such as parodies, memes and fanworks. These works can have important social welfare benefits, promoting discussion and debate and the spread of news and information, especially among young adults who have historically been disengaged from standard forms of news consumption.
Article 11: the miscalled ‘Link Tax’
Article 11 introduces a ‘neighbouring’ right for press publishers. It does not protect original creative thought but rather the economic investment made in that work (similar to the protection given to sound recordings, films, broadcasts and typographical arrangements in the UK). This grants an additional type of copyright protection to publishers if the headline of the news article appears in the web address for the story (standard procedure for many news sites). For more information, please see Prof. Lionel Bently’s recent study.
This is important as increasing amounts of internet users now consume news articles via aggregators such as apps, social media and search engines rather than through the purchase of newspapers, leading to issues of scarcity of attention. Therefore, proponents of Article 11 argue that this will protect the investment in the underlying story – it will ensure that any service that wishes to refer to or provide a link to a news story will have to pay a fee or a licence to do so. It is hoped that the fee will be passed on to the news publisher and through them to the journalist/writer (demonstrated by the additional Article 11(4)(a) addition, stating that “Member States shall ensure that authors receive an appropriate share of the additional revenues press publishers receive” for the use of their works. Without a stated definition of ‘appropriate share’ or a mechanism for revenue sharing, it is unclear how this will happen).
However, this has been referred to as a ‘link tax’ by those who argue that it will follow the negative experiences of similar legislation in Germany and Spain. In Germany, the legislation (Leistungschutzgesetz) was used not to prevent work being taken, but to force larger platforms like Google to both use the works and pay a fee to do so. In Spain, a similar law (Article 32(2) of their Copyright Act) showed that small publishing companies are harmed most by the legal uncertainty and threats of enforcement. Research has shown that many of these smaller publishers have subsequently closed, and it has been seen to act as a deterrent to innovation in those markets. This has obvious anti-competitive effects on the market for news publishers as well as the digital economy as a whole. It is alo feared will also restrict access to information and could lead to more groups desiring similar forms of protection. It has been argued that a better legal solution would be to extend Article 5 of the Enforcement Directive 2004/48/EC to allow publishers to enforce copyright over the works they publish.
There have been some small improvements to this article as it has worked its way through the European political system. The original wording of Article 11(4) called for this right to extend for 20 years from publication, but the accepted amendment decreased this to 5 years, and importantly added wording stating that this provision does not apply retroactively. Further, the new subsection 2(a) states that protection shall not extend to hyperlinks that only have individual words attached. However, this does not go far enough to protect the rights of users to have free access to information and expression. In today’s world of ‘fake news’, we should surely be tearing down barriers to entry to the news market in order to broaden the variety of views given in the mainstream media, and as such this is a troublesome subsection of the legislation.
Article 13: Upload Filters
The content and context of Article 13 was discussed in detail by my colleague Dr Sabine Jacques earlier this year. This blog will focus on two main issues with the legislation as it stands: issues in relation to the content recognition technology that it seeks to impose on all online content sharing service providers; and the logistical difficulties that will arise for users attempting to claim that their work benefits from a limitation or exception (L&E) to copyright.
By passing Article 13, the EP demonstrates a belief that there is an efficient, effective tool already in existence that could be used to identify copyrighted works within other works as they are uploaded to the internet. However, there are relatively few current iterations of this technology – for example YouTube’s Content ID, Audible Magic, Digimarc and ACR Cloud. This immediately means that any smaller sites will be at an economic disadvantage, as they will either need to devote resources to develop their own technology before the enforcement of this legislation (and pay for an expensive, powerful server to run it), or pay to license existing technology. The liability for errors in the system will also hit smaller firms disproportionately as fines under Article 13 of the Enforcement Directive 2004/48/EC are calculated on the harm to the value of the work being shared, not on the ability of the firm to pay (although they should be proportionate under Article 3(2)).
Content recognition software currently works through the creation and maintenance of huge databases of copyrighted material and the application of scanning technology to check whether any part of a suggested new work matches an entry on the database (hence the need for strong servers). However, even Content ID operating on YouTube, the industry standard for this filtering software, has been proved not to work correctly – either failing to identify approximately 1/3 of the sound recordings contained within it, or throwing up false positives.
As well as being insufficiently functional from a technological standpoint, this filtering technology has also been proved unable to handle the complex and contextually-driven issues surrounding copyright L&Es, which much of the non-commercial creation discussed at the start of this blog (memes, fanworks and parodies) rely upon. These L&Es are laid out in Article 5 of the Infosoc Directive which sets out a list that member states can choose to implement (the UK has chosen to implement several of these into Articles 29-30 of the CDPA 1988, for example). The variety of L&Es between member states and the correlated incompatibility of indefinite general monitoring of UGC was discussed in C-360/10 Sabam v Netlog. As Sabine argued in her previous blog post, it is hard to see how the requirement for the use of content recognition software differs from ‘general monitoring’. It is also hard to see how the technology would function without capturing certain types of personal data (such as usernames and locations) that could also be incompatible with users’ privacy rights (the readers will remember the introduction of the GDPR earlier this year).
Given the large fines possible to be levied in cases of copyright infringement, it is likely that service providers will block the upload of any work that matches the content management software. This puts the onus on users to appeal to the service provider if they believe their work is protected by an L&E, presumably through the complaints procedure referred to within Article 13(2). However, as anyone who has ever used an online complaints procedure knows, the response rates and efficiency can vary dramatically! It is feared that this could create a chilling effect on freedom of speech.
Conclusion
The issue of legislative failure to prevent harmful reuses of copyright-protected materials is one that requires legislative attention. However, the Proposed Directive as it stands has several important issues that mean it will not create the ‘Digital Single Market’ it desires. Most of these issues relate to incompatibility with other European legislation, but they also highlight a fundamental misunderstanding regarding creation in the digital era. It is ironic, perhaps that this misunderstanding is demonstrated most strongly by Article 13, which will arguably prove unlucky for all internet users.
This is not the end of the road. This Autumn there will be a Trilogue between the EU President (Austria), the Parliament and the Commission that could (although it is unlikely) lead to further changes. If it is accepted, there will be a final plenary vote in the Spring (currently anticipated for January 2019). While it is rare that legislation is rejected at this stage, there is precedent (for example, the ACTA in 2012). ACTA was rejected for being overly vague and open to misunderstanding, which could overreach and harm citizen’s rights. An analogy could definitely be drawn to the issues stated above and elsewhere regarding this proposed Directive.
In the words of MEP Julia Reda, #SaveYourInternet, it is time to #FixCopyright
Dr Ruth Flaherty 