Computer Scientist’s Review of the EPRS “InfoSoc” Directive Review

Dissecting a European Parliamentary Research Service report

Table of contents

Introduction

1. Responsibility of Online Intermediaries
2. User-Generated Content
3. The “Value Gap”
4. Freedom of Hyperlinking
5. Copyright Exceptions
6. Text and Data Mining
7. Digital Restrictions Management
8. Geoblocking and VPNs
9. Copyright Enforcement
10. Further remarks on the EPRS summary
11. Beyond Minor Tweaks

Introduction

Last year, the European Parliamentary Research Service (EPRS) published a review of the so-called “InfoSoc” Directive (2001/29/EC).

Review of the EU copyright Framework: European Implementation Assessment - Think Tank

The review includes a separate impact assessment and 3 briefing papers written by external organizations (Annex I-IV) which provide an overall balanced overview of key achievements and problems of the directive. There are a few mistakes which I will explain in greater detail but the annexes themselves are certainly worth reading.

However, the same cannot be said about the 32-page summary written by EPRS itself. The EPRS summary contains subtle errors, omits crucial context on some topics and section 7 contains downright ludicrous recommendations. Therefore I recommend skipping the summary altogether and reading the full review starting on page 41. (All page numbers in this article referring to the review correspond to PDF navigation. If you need to look up the page in paper version of the review, subtract 2.)

For the sake of clarity, I should define some terminology which is often used incorrectly:

• Copyrighted content is any subject matter eligible for copyright, including works of amateur authors, no matter how obscure and marginal.
• Free culture content is copyrighted content which is available under any free culture licence, e.g. the less restrictive Creative Commons licences. This does not necessarily mean that the content is available free of charge.
• Proprietary content is copyrighted content that is not free culture content.
• Commercial content is copyrighted content which the copyright holder wishes to commercially exploit. This does not mean that the content is necessarily proprietary.

Before we begin discussing key points of the review, here is a short overview of what you will find in the annexes:

• Annex I is the full analysis of the directive by CEPS. It identifies the most important problems of the directive but does not propose any specific solutions. Reading it is highly recommended because it explains the details of each problem.
• Annex II is a briefing paper by Milieu Ltd. It focuses on the benefits and drawbacks of formal variants of copyright reform, such as whether to only merge existing directives into a single document, pass an additional directive with new provisions or rewrite existing legislation from scratch. It makes no recommendations how specific policy issues should be solved. Annex II will be of interest mainly to the European Commission.
• Annex III is a briefing paper by CEPS. It focuses on solving fragmentation of digital market and better aligning copyright exceptions with new online uses of content. Reading it is highly recommended.
• Annex IV is a briefing paper by VVA. It focuses on copyright enforcement and fair remuneration of authors. There are some questionable statements in the part about copyright enforcement which I will address below but otherwise it is worth reading.

Now let’s finally take a look at what the review says. I will group the key points by topic and comment on all parts of the review together. I will also comment on how how some of these topics were addressed in the new copyright directive proposal. In the final section, I will add some further ideas that were beyond the scope of the “InfoSoc” Directive review.

1. Responsibility of Online Intermediaries

Responsibility (or liability) of online intermediaries is mentioned many times throughout the entire review. If you read only the EPRS summary, you might come under the impression that the review recommends enacting some sort of compulsory proactive anti-piracy measures that should be implemented by online services. However, that is not the case. This topic is discussed in greater detail in Annex I and Annex IV. In both cases, uncertainty regarding responsibility of online intermediaries relates only to court injunctions based on Article 8(3) of the “InfoSoc” Directive or Article 11 of the IPRED Directive. More specifically, courts avoid granting such injunctions because they are unable to assess proportionality of proposed measures.

Annex I, Part II contains six sections which need to be highlighted. Sections 1.6.1 (page 99) through 1.6.3 provide an overview of CJEU case law regarding injunctions against online intermediaries. Sections 3.2 (page 150) through 3.4 then add an overview of what the case law means in practical terms and what are the current enforcement trends in copyright industry. The case law explicitly prohibits injunctions in the form of permanent content filtering on the grounds that it would be excessively costly and it would entail a high risk of overblocking.

Annex IV, Chapter 4 explicitly says in section 1.2 (page 344) that this uncertainty can be solved through legislative action. However, the EPRS summary completely missed that the possibility of solving this issue through “soft law” approach is also implied at the end of the preceding section. (Responsibility of online intermediaries is listed in section 5.5.2 of EPRS summary, page 23, even though the same issue is also listed under a different name in section 5.5.1.)

The new copyright directive proposal does nothing to solve this issue. Instead, Article 13 of the proposal creates even more uncertainty where previously was none.

The proposal explicitly mandates content filtering measures that were rejected by CJEU due to incompatibility with Article 15 of the E-commerce Directive. The full impact assessment of the proposal cites existence of commercial content recognition services for audiovisual content as evidence that such measures are proportionate. However, the obligation to implement such measures depends only on a vaguely defined “large” amount of copyrighted content hosted by the online service, no matter whether any of that content actually infringes copyright or not.

As a result, online services that have nothing to do with audiovisual content will have to dedicate a significant amount of resources in order to preemptively solve a problem that may not even exist in their respective field at all.

(Note: The proposal very likely uses the word “copyrighted” to mean “commercial and proprietary”, which is a common misconception repeated even in the “InfoSoc” Directive review. All content that is eligible for copyright is copyrighted by definition, including user-generated content and other amateur or independent production.)

And last but not least, Article 13 of the proposal lacks even the most basic safeguards against abuse. The current wording of the proposal allows privileged rightholders with access into the content filtering system to block anything they want, no matter whether they hold copyright on the work in question or not. If the censorship of non-infringing content was intentional, the proposal does not allow service providers to override the block. And yes, even the current “notice and takedown” system was abused for deliberate censorship in the past. Which brings us to the next key topic.

2. User-Generated Content

Annex I provides a well-written explanation of what is “user-generated content” in Box 1 on page 107. User-generated content is mentioned multiple times throughout the review but it is not given as much attention as it really deserves. The review focuses mainly on its relation with copyright exceptions and recommends that copyright legislation should do more to encourage its creation. However, legal uncertainty is not the biggest problem that creators of user-generated content have to face.

The biggest problem for them is overblocking through the “notice and takedown” system on massive scale.

To show the extent of the problem, I will use takedown statistics for Google search. Although its scope is much larger than just user-generated content, it is representative of the problem. According to the most recent reports, Google receives takedown requests for 24 million links per week for its web search engine. That adds up to 1.25 billion links per year. Google removes about 97.5% of links requested for takedown. However, a study done earlier this year on a sample of takedown requests found that over 28% of requested links may be non-infringing. 4.2% of links requested for takedown pointed to content that was completely unrelated to the copyrighted work in question. Some takedown requests were based on subject matter that wasn’t eligible for copyright in the first place. Large number of links requested for takedown were very likely protected by copyright exceptions. And some of the requested links never even existed. If 28% does not seem too high, it translates to roughly 350 million links to legitimate content taken down in a year. And the number of links requested for takedown has been doubling every year for quite some time.

Just to illustrate that rightholders don’t bother with even the most basic sanity checks before sending a new copyright notice, last year, an anti-piracy outfit working for Columbia Pictures sent a takedown request listing 10 video links to video sharing platform Vimeo. The takedown request was allegedly targeting pirate copies of the new movie Pixels. Seven of the videos requested for takedown were completely unrelated original content. Another video was an award-winning short film that the new feature-length movie was actually based on. And the remaining two videos were their own official trailers. The only thing all of those videos had in common was the word “pixels” in their title.

In short, the main problem for creators of user-generated content is not lack of legal certainty.

The main problem is that copyright exceptions and even the most basic right to publish your own original work is effectively meaningless in the online environment. Article 14 of the E-commerce directive (and similar legislation in United States) forces online platforms to blindly trust self-proclaimed rightholders and err on the side of overblocking. EU legislation needs to recognize that user-generated content has equal cultural value as commercial works published by the copyright industry. Amateur creators of today are the future of our culture, whatever their business model may become in the end. Their right to create must not become collateral damage in an endless war on copyright infringement.

3. The “Value Gap”

This topic is not mentioned in the review by the title above but there are a few short and vague mentions that “rightholders must engage with powerful online intermediaries on weaker negotiating footing” in Annex I and the EPRS summary. There is no further explanations of who these “powerful online intermediaries” are or why the rightholders must engage with them. But this topic is used as a justification for the worst parts of the new copyright directive proposal so I need to address it anyway. The rest of this section is based on Impact Assessment on the modernisation of EU copyright rules, Part 1, section 5.2.

First, a short overview of the most general types of online intermediary business models. You should be familiar with the first two types.

• Business-to-business (B2B): This category represents mainly stock image and stock video archives and other online services that supply commercial content to commercial users. Rightholders usually work closely with these intermediaries and don’t complain about them much.
• Business-to-consumers (B2C): This category represents services which provide consumers with access to content supplied primarily by the copyright industry itself, such as iTunes, Amazon, Spotify, Netflix, etc.
• Consumers-to-consumers (C2C): This category represents online services which provide access to content supplied primarily by amateur or independent creators, such as YouTube, Vimeo, Wikipedia, WordPress, Instagram, etc. Note that calling users of platforms in this category “consumers” is misleading. Many of those users are artists and creators in every sense of the word. Peer-to-peer might be more accurate description but it would also be confusing because the term already refers to unrelated technology. So I will refer to this category by the acronym “C2C” for the sake of consistency with established terminology. Hopefully, the acronym will not carry the misleading connotations of the word “consumer”.

Some online platforms have features from more than one category. For example, Amazon is primarily a B2C platform but it also has some C2C features, mainly for self-publishing of books. YouTube is primarily a C2C platform but it also has B2C partnerships with the copyright industry.

Section 5.2 of the Impact Assessment makes it clear that those “powerful online intermediaries” are only C2C platforms. In essence, the copyright industry is complaining that C2C platforms which focus on audiovisual content refuse to make a major change in direction towards B2C business models.

If some C2C platform does not want to host content published by the copyright industry, it has no obligation to buy a licence for it, third-party copyright infringement on the platform notwithstanding. Of course, rightholders are free to ask for anything they want but they are not entitled to automatically get it.

4. Freedom of Hyperlinking

Annex I, Part II, section 1.2.3 (page 75) is possibly the most absurd section of the analysis written by CEPS. It explicitly calls for making all hyperlinks subject to authorization from copyright holder of the target work, under the exclusive right of communication to the public. Such position is also hypocritical, given that Annex I and Annex III (both written by CEPS) together contain 52 unique clickable hyperlinks and the whole “InfoSoc” Directive review contains a total of 162 clickable hyperlinks. (Yes, I have really counted them. Not manually, of course.) I have serious doubts that the authors have asked anybody for permission to include those links, contrary to what they preach.

A clickable hyperlink is just a slightly more convenient version of traditional bibliographic reference.

Nothing more, nothing less. Just like saying that “you can find some particular piece of information in the book Swarmwise by Rick Falkvinge” is not an act of communicating said book to the public, this hyperlink by itself is not an act of communicating that book to the public either. The only act of communicating the book to the public is done by the server which hosts it and provides a copy upon request, that is, after you click the link. This is what the hyperlink really looks like to your computer:

<a href=”http://falkvinge.net/files/2013/04/Swarmwise-2013-by-Rick-Falkvinge-v1.1-2013Sep01.pdf”>this hyperlink</a>

As you can see, the hyperlink itself does not contain any piece of copyrighted material. It consists of only the address where the book itself can be found and some short description that will be turned into a clickable hyperlink by your web browser. If the book gets removed from that address in the future, the hyperlink will stop working. Here are a few handy diagrams clearly showing when is communication to the public taking place. Red arrow means it is an act of communication to the public, green arrow means it is not. First, a simple clickable hyperlink like the one above:

Now let’s look at the slightly less intuitive case of resources embedded into a web page from external sources (such resources might be images, videos, sounds files or entire web pages):

The reason why embedding does not constitute an act of communicating the embedded resource to the public is that access to the resource can be blocked at the third-party server which actually hosts the resource. If the third-party server is a legitimate source of the copyrighted content in question, it can use many simple verification techniques to prevent embedding into unauthorized websites. If the third-party server itself infringes copyright by hosting the content in question, then extending copyright liability to other websites which merely embed it creates massive legal uncertainty and undue burden for anyone who wants to use legitimate public resources such as open data (copyrighted under the Database Directive, 96/9/EC), public domain works or free culture artwork and media. Note that according to Annex I, Part II, section 2.2.2 (page 122) of the review, it takes up to six months to even identify the relevant rightholders to seek authorization from. And all that additional legal uncertainty does not help address the main problem in any meaningful way either, since the copyright-infringing content will remain accessible to the public. Here is a diagram of blocking unauthorized embedding at the source:

The last diagram is the most complicated situation: The server in the middle is relaying copyrighted data (as a proxy) from third-party server to the user. The difference between the two situations is that on the left, the server operator did not configure the server to relay predetermined proprietary copyrighted content. The server merely processes any data from user-provided address and it was the user who provided a link to third-party copyrighted content. For example, this is how Google Translate operates.

On the right, the server operator has intentionally configured the server to relay third-party content that is known to be both copyrighted and proprietary. For example, imagine a website that converts Netflix or Spotify streams into downloadable files and publicly advertises that functionality.

5. Copyright Exceptions

One of the leading causes of internal market fragmentation identified by the “InfoSoc” Directive review is the optional nature of certain copyright exceptions. Annex I, Part II contains two sections dedicated to this issue. The explanation of cross-border impact of copyright exceptions in section 1.2.5 (page 79) in particular needs to be highlighted. Section 2.3 (page 129) then continues with detailed overview of the most important exceptions and their current implementation in member states. Annex III then proposes several options how to address this issue.

The new copyright directive proposal introduces three new mandatory exceptions covering only two out of seven key areas identified by the review where harmonization is still sorely lacking. I will comment on the exception for text and data mining (Article 3) in the next section.

In short, the exception is inadequate and grants absurd new powers to rightholders.

• The exception for cross-border teaching activities (Article 4) is deliberately written to be nearly useless in practice.
• Amending the existing exception for teaching and scientific research (Article 5(3)(a) of the “InfoSoc” Directive) to allow online and cross-border uses would have been better.
• Only exception for preservation of cultural heritage (Article 5 of the proposal) is actually good.

6. Text and Data Mining

Text and data mining (TDM) is a method for automated extraction of facts from large amounts of data which may or may not be copyrighted.

Annex I, Part II, section 1.5.3 (page 95) correctly notes that it is unclear whether TDM falls into the scope of exception for transient and incidental copies (Article 5(1) of the “InfoSoc” Directive). The opening text of Chapter 5 in Annex II then notes that other countries regard TDM as lawful use of copyrighted content which doesn’t require special permission from rightholders.

The legal uncertainty in EU therefore creates clear disadvantage for our researchers.

Mining the same data the hard way, for example by reading thousands of articles yourself (or, as is common practice on universities, with help from dozens or hundreds of students), does not require any special permission from rightholders. There is therefore no valid reason why doing the same thing using a computer should be treated any differently. Making TDM subject to a separate permission from rightholders is like charging extra for a box of nails if you intend to bash them into something using a hammer instead of your bare hands.

Here I would like to quote the Commission’s justification for leaving TDM partially restricted, from Part 1, section 4.3.3 of Impact Assessment on the modernisation of EU copyright rules (page 117, emphasis mine):

Corporate research users

Differently from the other options, because of the broader scope of application of the exception, Option 4 would specifically benefit researchers in commercial companies as they would no longer need a specific licence to mine content to which they have lawful access to. However, corporate users, notably life-science companies, benefit today from a functioning licensing market for TDM of scientific publications and they have not requested any intervention at EU level. TDM is often licensed to these users as part of a wider licensing agreement with rightholders including several uses and services that go well beyond what the exception would allow them to get for TDM purposes (notably in terms of formats, structured data, getting direct feeds into their own databases etc.). Therefore corporate users are likely to continue to purchase value added services from content owners. This option also entails a risk that publishers may increase the subscription fees for commercial users to compensate for the loss of TDM related revenues (this is more likely to happen with corporate users than with universities because of the different purchasing power, at least as regards larger commercial operators).

Rightholders

Impact on TDM licensing market and the revenues thereof: This option would have a more significant negative impact on rightholders. As a consequence of the broad scope of the exception, STM publishers would no longer be able to licence TDM for scientific research purposes to commercial players, which represent an essential market for them, notably in areas such as life science and pharmaceutical. Industry estimates the value of the commercial TDM market (in Europe) to be worth more than 56 million euros by 2019. Two major STM publishers alone currently have 302 existing TDM licences with life science companies, which is a significant figure given the characteristics of the market. […] Similarly to the other legislative options, this option would in principle not remove rightholders’ ability to generate revenues from selling access to their content. However, deals between STM publishers and corporate users usually include TDM as part of comprehensive agreements covering a whole series of usage rights and added value services mentioned above. The introduction of an exception would lower the value of these agreements, since TDM rights as such can no longer be subject to licence. Rightholders may try to compensate the value lost as a consequence of the legislative intervention by raising licences fees for access and other uses/value added services. However, it is not clear whether and to what extent they would manage to do so. […]

The smoke screen of supposedly unclear and unpredictable effects on value of agreements obscures merely a huge non-issue. In reality, the virtual TDM licence fee would simply move to the price of specialized subscription service for TDM and the total price of the whole bundle would stay the same. You do not need copyright to charge for access to your online service. Since those contracts cover a much larger bundle of licences and services, neither side will be interested in renegotiating the whole contract because of such a minor change. If there will be any changes at all, it will be mainly adjustments of legal terminology. For comparison with the expected future size of TDM market (€56 million by 2019), the global market for English-language scientific literature was worth $10 billion as of 2013. Also note that Annex 11A of the Impact Assessment estimates the size of TDM market to be only €2 million as of 2015.

The Commission shows remarkable caution not to cause essentially cosmetic changes to a grand total of 302 licence agreements between scientific publishers and the pharmaceutical industry. Contrast that with its aggressive disregard for legal certainty concerning online services which we have discussed at the beginning of this article.

But the worst part of the TDM exception proposal is that it gives rightholders the legal authority to tamper with network infrastructure of research institutions (Article 3(3) of the new copyright directive proposal). The only thing more ridiculous than such power is the justification given in Part 1, section 4.3.3 of the Impact Assessment (page 115):

Impact on the protection of content: Today rightholders may impose on users, through TDM licences clauses, technical conditions which they consider important, among other things, to prevent unauthorised uses of the content being mined and protect the technical stability of their databases. As described in Annex 11B, these measures are, for example, APIs for automated downloading, access limited to a determined range of IP addresses or other user authentication measures, limits on the speed or number of downloads, etc. In view of the current limited self-standing economic value of TDM in licences with universities/public research organisations (see above), the possibility for STM publishers to impose such conditions is often invoked as a key reason for them to retain the ability to licence TDM.

There is only one slight problem: Implementing the access control technologies listed above on the user side (e.g. on the academic network of a university) makes no technical sense. Those technologies are all currently implemented on rightholders’ own central servers and therefore require no legal backing, whether through contract or by law. Annex 11B mentioned above lists three additional technologies which also make sense only on rightholders’ own servers. Citing them as evidence that rightholders must be entitled by law to tamper with computer networks of their customers shows that the Commission is embarrassingly ignorant of the most basic technical knowledge. But more importantly, let’s assume for a moment that rightholders really do impose some security measures (not listed above) through licence agreements. If that were true, why would they tie them specifically to the TDM clause? Research institutions get access to the same amount of content whether their contract includes a TDM clause or not. Requiring security measures through contract makes sense only as part of the core licence to access the content at all.

And last but not least, any active involvement of rightholders in the process of text and data mining creates major legal barriers for mining any content other than scientific journals from large publishers. Input data for mining can be anything, images, sound, video, software, public social media posts, websites, literature, anything.

The TDM exception must work even for a collection of copyrighted content from millions of different rightholders.

7. Digital Restrictions Management

Our next topic is what is called “Technological Protection Measures” (TPM) throughout the review. But I will refer to it as “Digital Restrictions Management” (DRM) because this term is widely used by the IT community and the general public. And frankly, it is a much more accurate descriptions of what these technologies are supposed to do.

The core idea of Digital Restrictions Management would be best described as technically sophisticated wishful thinking. The whole concept is fundamentally flawed from the outset because it pursues two mutually exclusive goals: Grant full access to the authorized user while simultaneously preventing the very same user from having any access at all. All media except software are limited to the same basic model of DRM: Encrypt the content and hide the decryption key “under the doormat” bundled with the content. The player/reader software will then extract the decryption key and decrypt the content. Similarly, anybody who knows how to look “under the doormat” can simply decrypt the content and convert it into any unencrypted format, effectively stripping it of all DRM once and for all. What the law says about the act is mostly irrelevant because there is simply no way to enforce it at all. It is also interesting to note that in one particular case, anti-circumvention provisions of copyright law created a free speech controversy and freedom of speech prevailed in the end.

Software, on the other hand, can use much more aggressive DRM techniques. While that makes DRM circumvention and stripping significantly harder (but still not impossible), it also creates even greater problems for interoperability, preservation for future generations, computer security and sometimes effectively results in vendor lock-in and planned obsolescence (when DRM periodically requires renewal of authorization from vendor’s central servers which may be shut down later).

In any case, DRM serves mainly as a reminder to paying customers that they could have saved themselves a lot of headaches just by pirating the same content in the first place:

Annex I, Part II of the review correctly reports in section 1.3 (page 83) that DRM has failed its main objective of enforcing copyright and briefly notes in the opening text of section 3 (page 144) that it is instead being used for anti-competitive purposes. However, the review fails to convey the extent to which DRM has become a major consumer protection issue. DRM is increasingly being used by manufacturers to infringe property rights to physical products that have long since been sold to their customers. Those products currently include cars, farming equipment, coffee machines, printers and the list will most certainly grow in the future. In addition, anti-circumvention provisions of copyright law are sometimes used to harass security researchers and crucial security research is either not reported or not done at all in the first place as a result. Criminals looking for software bugs to exploit, on the other hand, are not subject to any of those limitations.

8. Geoblocking and VPNs

Virtual Private Networks represent another topic where the “InfoSoc” Directive review fails as a whole. Box 2 in Annex I (page 123) provides a brief explanation of the basic idea behind the technology in its first paragraph but I would say it is too short for its intended audience. I recommend reading the Wikipedia article instead. The other two paragraphs then discuss commercial anonymization services without making a clear distinction between VPN the technology and VPN the commercial service. The part about “socks” in the second paragraph in particular demonstrates that the writer lacks even the most basic knowledge of VPNs and related technologies. SOCKS (which stands for “Socket Secure”) is a protocol for connecting to a proxy server. VPNs and proxies are two different technologies designed for different tasks, with different features and limitations. But in general, both may be used to hide your public IP address.

The third paragraph then questions whether commercial anonymization services have any significant legitimate use. Here are a few examples of legitimate uses:

• Protection of whistleblowers and dissidents living in repressive regimes.
• Circumvention of censorship in repressive regimes.
• Partial protection against user tracking on the web (IP address anonymization is necessary but not sufficient, it must be used in combination with other privacy technologies).
• Protection against eavesdropping when you connect through public Wi-Fi or other potentially malicious networks (assuming the VPN service itself can be trusted).
• Protection against violations of network neutrality (or to detect that your Internet access provider is breaking the law by violating network neutrality).

Annex III, section 4.2.1.3 (page 306) then reiterates the doubts from Annex I about legitimacy of VPNs while, ironically, citing a BBC article about Chinese censorship. But at least it adds a key observation that rightholders can solve the problem of geoblocking circumvention simply by offering legal access to their content.

But the mistakes in Annex I and Annex III aren’t all that bad. The real “gem” is section 7.3.1 of the EPRS summary (page 36) which proposes crippling VPNs by imposing DRM on them that would block access to blacklisted addresses. That is insane. Never mind that such a plan is completely delusional because there are dozens of different VPN protocols with hundreds of existing independent implementations, many of them open-source, which are not going to just disappear. More importantly, VPN as a technology is crucial part of network infrastructure that hundreds of thousands of businesses in Europe rely upon every day. Such a blacklist would give malicious hackers opportunity for crippling denial-of-service attacks against core European Internet infrastructure.

9. Copyright Enforcement

I would like to begin this topic by pointing to a study of access to foreign media in North Korea. In short, even a regime as tyrannical as North Korea is failing to prevent illegal sharing of content, even though the general public has no access to the Internet, or any other computer network for that matter. In their case, it is a matter of censorship rather than copyright, but that is irrelevant. The only practical difference between copyright and censorship is the existence of one legal distribution channel. Enforcement methods against illegal sharing are exactly the same for both. Given that even the threat of North Korean gulag does not deter people from sharing culture illegally, it is time to ask ourselves whether copyright law isn’t trying to solve the wrong problem in the first place.

Chapter 2, section 2.1 of Annex II (page 237) and Chapter 3, section 1 of Annex IV (page 337) both describe copyright as a fundamental right. Here is the relevant part of The Universal Declaration of Human Rights adopted by United Nations:

Article 27.

(1) Everyone has the right freely to participate in the cultural life of the community, to enjoy the arts and to share in scientific advancement and its benefits.

(2) Everyone has the right to the protection of the moral and material interests resulting from any scientific, literary or artistic production of which he is the author.

Here is the relevant part of Charter of Fundamental Rights of the European Union:

Article 17: Right to property

1. Everyone has the right to own, use, dispose of and bequeath his or her lawfully acquired possessions. No one may be deprived of his or her possessions, except in the public interest and in the cases and under the conditions provided for by law, subject to fair compensation being paid in good time for their loss. The use of property may be regulated by law in so far as is necessary for the general interest.

2. Intellectual property shall be protected.

Here is the relevant part of Charter of Fundamental Rights and Freedoms which is part of the Constitution of my home EU member state, Czech Republic:

Article 34

(1) The rights to the fruits of one’s creative intellectual activity shall be protected by law.

(2) The right of access to cultural wealth is guaranteed under the conditions set by law.

As you can see, all three documents define the fundamental right to intellectual property very broadly and do not specify any particular form of protection. Even the UN special rapporteur in the field of cultural rights herself has explained in a speech to the EP Legal Affairs Committee that copyright as such is not a fundamental right, but only one of many possible variants of codifying the underlying fundamental right into concrete law. A variant which is increasingly at odds with true fundamental rights. At the end of this article, I will propose an alternative system which is better suited for the digital age while still conforming to all three definitions of intellectual property quoted above.

One of the enforcement mechanisms which is currently gaining popularity in the EU and other countries is the so-called “follow the money” approach, where private companies such as advertising and payment providers are supposed to block payments on a voluntary basis to services accused of copyright infringement. The Electronic Frontier Foundation calls such voluntary measures “shadow regulation”. Essentially, it represents outsourcing of law enforcement to private entities which, unlike government institutions, are not bound by the principle of proportionality, due process or other inconvenient legal safeguards. Annex I, Part II, section 3.5 (page 153) describes this approach in greater detail and also mentions two main problems: coordinated action against allegedly infringing service could violate competition law and wrongly accused legitimate services may have no way to defend themselves. Annex IV, Chapter 4, section 1.1 (page 340) also repeats that lack of accountability is a major problem.

This approach has large potential for abuse and there need to be clear rules for accountability and remedies when legitimate services get wrongly targeted.

10. Further remarks on the EPRS summary

Here I will point out some interesting parts of the EPRS summary which do not fit any of the topics discussed above.

Section 3

• Section 3 (page 11) compares the European copyright industry with its American counterpart and points out that revenues of the American copyright industry are significantly higher despite much smaller domestic market. The revenue figures used in that comparison date back to 2012. If we make a naive per-capita comparison, dividing the total revenue by the size of the respective domestic market, American revenues turn out to be about three times higher than European. EPRS makes no further comments on the difference in revenue but we can safely interpret most of it as the true cost of market fragmentation. The main competitive advantages of American copyright industry are the unified mostly homogeneous domestic market and consequently better access to global markets, including EU member states. Piracy affects both industries more or less equally and translations represent a major expense only in the book industry.
• Section 3 also notes that digital revenues of European copyright industry represent only a small fraction of its total revenues. Section 7.2.2 (page 36) then interprets this finding as a chicken and egg problem — presenting two different explanations that either low investment in digital technologies is a result of outdated legal framework, or that EU copyright industry is “more traditionally structured” and therefore modernisation of copyright should be delayed in order to give it more time to adapt. The reality is that the copyright industry has been fighting against transition to digital technologies every step of the way and any legislative measure intended to stabilize old crumbling business models will be used by the industry as an excuse to delay the transition even further. Section 7.2.2 also says that (paraphrased) “geoblocking cannot be eliminated because of piracy” which is ridiculous. The only link between the two is that lack of legal availability creates strong incentives for piracy. The opening paragraph of Section 7.3 on the same page then blames poor digital revenue on piracy.
• Let’s take a look at the legal options of accessing commercial content online back in 2012: iTunes was available in all 27 EU member states but in 12 of them only for a year. Renting of movies was added between 2008 and 2011 (depending on state) and TV shows are available in only 3 member states even today. Spotify (first launched in 2009) was available in only 11 EU member states, in 3 of them for less than a year. Vevo (launched in December 2009) is still available today in only 8 EU member states but some content is available globally through YouTube partnership. Netflix was available in only 6 EU member states for less than a year. Other services, including the whole e-book market, were too marginal to be worth mentioning here. In short, if something was not available on iTunes, most of the EU had no legal way to access it online. If there is no way to pay for specific content in the first place, blaming low revenues on piracy is ridiculous.

Section 4

• Section 4 (page 12) cites a study by Ernst & Young about the European copyright industry. But detailed employment numbers in section 4.2 (page 13) show that employment in book industry and visual arts industry is inflated by taxpayer-funded jobs which contribute little (if any) revenue to the industry total. Public and academic libraries account for over 40% of all book industry jobs. The visual arts industry numbers are less inflated, museums and heritage institutions account for only 9% of its jobs. Libraries, museums and other heritage institutions would continue to exists even if the corresponding domestic industry disappeared, therefore including public institutions in industry employment statistics is questionable.

Section 5.1.1.

• Section 5.1.1. (page 15) mentions the closing down of Napster as a victory for legislation criminalizing circumvention of DRM. This is factually incorrect on several accounts. First, Napster was a file-sharing service and the lawsuit had nothing to do with DRM circumvention. Second, Napster settled the lawsuit in 2001 and spent the next year developing a music subscription service. Unfortunately, the company went bankrupt before the new service could be launched.

Section 7.3.2

• Section 7.3.2 (page 38) makes a surprisingly insightful observation that piracy may also have positive impact on sales and the effect should be studied further.

11. Beyond Minor Tweaks

Our final topic falls beyond the scope of the original “InfoSoc” Directive review. We should start with some historical background of the current copyright legislation. Foundations of copyright were created in England with the Licensing of the Press Act 1662 which was a government censorship legislation. The law expired in 1695 and after much lobbying by publishers (not authors), it was redesigned into the first copyright law in 1710 as Statute of Anne which did not include the original censorship mandate. Current global copyright system is based on the Berne Convention for the Protection of Literary and Artistic Works adopted in 1886, only 9 years after the invention of phonograph and a few years before the first practical motion picture camera was developed.

Copyright was designed to resemble ownership of physical property mainly because the concept was familiar to everybody and the distinction between a physical book and the text written inside it was probably not readily apparent back then.

Until the second half of the 20th century, copyright was an obscure industry regulation which had no impact on consumers and amateur creators. That has changed drastically with the advent of digital technologies.

Trade is based on scarcity. Trade of physical goods emerges naturally because they are inherently scarce. Intangible goods like literature or film, on the other hand, require large initial investment after which they can be replicated quickly and cheaply by many independent suppliers. Copyright tries to solve this lack of scarcity by creating artificial legal barriers between copyrighted goods and potential users. The basic idea seems to be that the higher the artificial barriers, the better the bargaining position for authors. That was partly correct in the past when setting up mass manufacturing of copies was still somewhat expensive (although publishers benefited from the resulting lack of competition a lot more than authors did).

However, in the age of Internet, this model breaks apart.

Copyright used to regulate only the relationship between the author and the publisher. Most other parts of the cultural business chain between the publisher and the end-user were just transporting finished physical copies from point A to point B which does not require any licence. But in the digital world, copyright now regulates the whole business chain all the way down to and including the end-user. If you want to sell music CDs, you can simply order a truckload of CDs from catalog and you will get a shipment and invoice a few days later. Quick, simple, no copyright lawyers involved. If you want to sell music on the Internet, it will take you about 6 months just to identify and contact the relevant rightholders and another 2 years to negotiate a licence agreement (see Annex I, Part II, section 2.2.2, page 122 of the “InfoSoc” Directive review). The cost of this unnecessary copyright red tape runs into hundreds of thousands of euros per company per year and authors get none of that money.

A system where every use of copyrighted content requires explicit permission from the rightholder is no longer fit for purpose because it does not scale as far and wide as it needs to.

The scalability problem is only going to get progressively worse because we are still at the beginning of the digital revolution. Keeping the old system will cost us not just astronomical amounts of money, but first and foremost lost opportunities for authors, innovators and the society as a whole. We need a new system which actively facilitates mutually beneficial trade between authors and users with as few speed bumps as possible.

Well then, what if we based copyright on a legal right to remuneration instead?

Current system of copyright is based on the right of arbitrary exclusion. Enforcement of exclusion and lengthy negotiations of permissions generate most of the costs associated with current system but without providing any significant benefits compared to the much more simple remuneration right.

Ensuring remuneration for the author is the whole point of economic rights in copyright law.

It is the only legitimate interest derived from economic rights. The ability to arbitrarily exclude somebody from using a legally published copyrighted work, on the other hand, is not in itself a legitimate interest due to its inherently anti-competitive nature when applied to intangible goods.

• Is this idea compatible with cultural and economic fundamental rights? Yes, it is much more compatible than the current system.
• Is it compatible with Berne Convention? No.
• Will it cause massive upheaval in the copyright industry? Absolutely, large part of the industry exists only to deal with artificial barriers which would be eliminated by the reform. The EU legislature must come to terms with the fact that current structure of the copyright industry is not set in stone and economic progress is not possible without elimination of obsolete businesses.
• Will it harm authors? It’s hard to predict with certainty but most likely not. Any negative impact on existing revenue streams due to increased competition should be more than compensated by income from new markets and new business models which the copyright industry today refuses to exploit. Making the role of rightholder mostly passive while still ensuring remuneration will allow the digital market to scale better and reach wider audiences.

You should also note that copyright law has been slowly replacing the right of exclusion with right to remuneration in some peripheral parts of the market for quite a while. Compulsory collective management of copyright is the best example of such shift. Although rightholders still retain the right to explicitly prohibit use of their work even under compulsory collective management, they rarely use it. But collecting societies could be optional in the new model. Users don’t need any intermediary to redistribute remuneration anymore. Public copyright registry in combination with modern electronic payment methods will be good enough.

Full transition to the remuneration-only model of copyright will not be easy.

It will require renegotiation of several international treaties and long transitional period just to prevent overnight collapse of an entire industry. But the concept could be tried on a limited scale for example by allowing commercial use of sufficiently old orphan works under statutory licence (but without the requirement of prior diligent search which would be too burdensome for smaller commercial users). The statutory licence fee would be held in escrow until the rightholder is found. Another option is EU-wide system for voluntary registration of copyright which would include a blanket licence that requires only remuneration as specified by the author.