3.1 Data Governance Foundations
3.1.1 Data Governance and the Protection of Personal Data
Data is taking a central role in many day-to-day processes. In connecting data, ensuring interoperability is often the hardest part as the merging and connecting of databases takes a lot of curation time, as was stated by Mercè Crosas in an interview with the BDVe.Footnote 8 Therefore, it is important that data practices are arranged solidly by doing good data governance to avoid interoperability problems. In addition, data is an indispensable raw material for developing AI, and this requires a sound data infrastructure (High-Level Expert Group on Artificial Intelligence, 201) and better models on data governance. In a recent panel held during the BDV PPP Summit in June 2019 in Riga,Footnote 9 a researcher from the DigiTransScope project – a project in which an empirical deep-drive is made into current data governance modelsFootnote 10 – gave a definition of the concept of data governance, as follows: ‘the kind of decisions made over data, who is able to make such decisions and therefore to influence the way data is accessed, controlled, used and benefited from’.Footnote 11 This definition covers a broad spectrum of stakeholders with varying interests in a big data landscape. More research is needed to find insights on the decision-making power of the different stakeholders involved so that a good balance is found between fostering economic growth and putting data to the service of public good. Concepts such as data commons (Sharon and Lucivero 2019) and data trusts have been emerging recently. Any kind of guidance should take all of these elements into account. It is important that all stakeholders are involved in the process of developing guidance, as otherwise the emergence and development of a true data economy are hampered.
In a data landscape, many different interests and stakeholders are involved. The challenging part about regulating data is the continuous conceptual flux, by which we mean that the changing meaning and social and cultural value of data is not easily captured in time or place. Yet, one can set conditions and boundaries that can aim to steer this conceptual flux and value of data for a longer foreseeable timeframe. One of the most notable regulations passed recently is the General Data Protection Regulation (hereafter referred to as GDPR). With this regulation, and accompanying implementation acts in several member states, the protection of personal data is now firmly anchored within the EU. However, the distinction between personal and non-personal data has proven to be challenging to make in practice, even more so when dealing with combined datasets that are used in big data analytics. It has also recently been argued that the broad notion of personal data is not sustainable; with rapid technological developments (such as smart environments and datafication), almost all information is likely to relate to a person in purpose or in effect. This will render the GDPR a law that tries to cover an overly broad scope and it will therefore potentially lose power and relevance (Purtova 2018). In this vein, there is a need to continue developing notions and concepts around personal data and the types of data use.
For most big data analytics, privacy harm is not necessarily aimed at the individual but occurs as a result of the analytics itself because it happens on a large scale. EU regulation currently lacks in providing legal remedies for the unforeseen implications of big data analytics, as the current regime protects input data and leaves inferred dataFootnote 12 out of its scope. This creates a loophole in the GDPR with respect to inferred data. As stated by the e-SIDES project recently,Footnote 13 a number of these loopholes can be addressed by court cases. The question remains as to whether and to what extent the GDPR is the suitable frame to curb such harms.
Despite many efforts to guide data workers through the meaning and bases of the GDPR and related data regulations such as the e-Privacy Regulation, such frameworks are often regarded by companies and governments as a hindrance to the uptake of innovation.Footnote 14 For instance, one of the projects within the BDV PPP found that privacy concerns prevent the deployment, operation and wider use of consumer data. This is because skills and knowledge on how to implement the requirements of data regulations are often still lacking within companies. The rapidly changing legal landscape and the consequences of potential non-compliance are therefore barriers to them in adopting big data processes. Companies have trouble making the distinction between personal and non-personal data and who owns which data. This was also reflected in a recent policy brief by TransformingTransport, which looked into many data-driven companies in the transport sector.Footnote 15 Additionally, these same companies experience trouble defining the purpose of processing beforehand, as within a big data context the purpose of processing reveals itself after processing. Mapping of data flows onto purposes of the data-driven service in development presents difficulties, especially when having to understand which regulation ‘fits’ on different parts in the data lifecycle. On the other hand, sector-specific policies or best practices for sensitive personal data are perceived as assets by professionals because these give them more legal certainty, where they face big risks if they do not comply. In this sense, privacy and data protection can also be seen as an asset by companies. We feel that there is a need for governance models and best practices to show that the currently perceived dichotomy between privacy and utility is a false one (van Lieshout and Emmert 2018). Additionally, it is also important to raise awareness among companies in which scenarios concerning big data and AI are useful, and in which scenarios they are not.Footnote 16 One of the main challenges for law- and policymakers is to balance rights and establish boundaries while at the same time maximising utility (Timan and Mann 2019).
3.1.2 Coding Compliance: The Role of Privacy-Preserving Technologies in Large-Scale Analytics
One of the more formal/technical and currently also legally principled ways forward is to build in data protection from the start, via so-called privacy-by-design approaches (see, among many others, Cavoukian 2009 and Hoepman 2018). In addition to organisational measures, such as proper risk assessments and data access and storage policies, technical measures can make sure the ‘human error’ element in the risk assessment is covered.Footnote 17 Sometimes referred to as privacy-preserving technologies (PPTs), such technologies can help to bridge the gaps between the objectives of big data and privacy. Currently, many effective privacy-preserving technologies exist, although they are not being implemented and deployed to their full extent. PPTs are barely integrated into big data solutions, and the gap of deployment in practice is wide. The reasons for this are of a societal, legal, economic and technical nature. The uptake of privacy-preserving technologies is, however, necessary to ensure that valuable data is available for its intended purpose. In this way data is protected and can be exploited at the same time, dissolving the dichotomy of utility and privacy. To ensure this is achieved, PPTs need to be integrated throughout the entire data architecture and value chain, both vertically and horizontally. A cultural shift is needed to ensure the uptake of PPTs, as the current societal demand to protect privacy is relatively low. Raising awareness and education will be key in doing so. It is important that PPTs are not provided as an add-on but rather are incorporated into the product. There is wide agreement that the strongest parties have the biggest responsibilities concerning protecting privacy and the uptake of PPTs, as was also confirmed by the e-SIDES project (2018).Footnote 18
Another point of discussion has been the anonymisation and pseudonymisation of personal data. It has also been argued that companies will be able to retain their competitive advantage due to the loophole of pseudonymised data, which allows for unfettered exploitation as long as the requirements of the GDPR are met.Footnote 19 Anonymised data needs to be fully non-identifiable and therefore risks becoming poor in the information they contain. Also, anonymisation and pseudonymisation techniques may serve as mechanisms to release data controllers/processors from certain data protection obligations related to breach-related obligations. Recent work done by the LeMO project found that anonymisation and pseudonymisation may be used as a means to comply with certain data protection rules, for instance with the accountability principle, measures that ensure the security of processing, purpose limitation and storage limitation. Pseudonymisation and anonymisation techniques can serve as a means to comply with the GDPR,Footnote 20 but at the same time, too far-reaching anonymisation of data can limit the predictability of big data analytics (Kerr 2012). However, as long as the individual remains identifiable, the GDPR remains applicable. It has been argued that, because of this, companies will be able to retain their competitive advantage by being able to unlimitedly exploit data as long as it is pseudonymised or anonymised.
3.1.3 Non-personal Data (FFoD)
In 2019, Regulation 2018/1807 on the free flow of non-personal data (FFoD) came into force, which applies to non-personal data and allows for its storage and processing throughout the EU territory without unjustified restrictions. Its objective is to ensure the free flow of data across borders, data availability for regulatory control and encouragement of the development of codes of conduct for cloud services. The FFoD is expected to eliminate the restrictions on cross-border data flows and their impacts on business, reduce costs for companies, increase competition (LeMO 2018),Footnote 21 increase the pace of innovation and improve scalability, thereby achieving economies of scale. This is all supposed to create more innovation, thereby benefiting the uptake of big data, in which the flow of non-personal data will remain of continuing importance in addition to having solid data infrastructures. For instance, the GAIA-X Project addresses how open data plays a role in creating a data infrastructure for Europe.Footnote 22 Other more developed initiatives include European Industrial Data SpacesFootnote 23 or the MOBI network for opening up and sharing data around blockchains.Footnote 24
The FFoD is the complementary piece of legislation to the GDPR as it applies to non-personal data. However, this distinction between the two regimes based on these concepts of personal and non-personal data is highly debated. The distinction is not easy to make in practice as datasets are likely to be mixed and consist of both personal and non-personal data. This is especially the case for big data datasets, as it is often not possible to determine which part of the set contains personal or non-personal data. This will result in it being impossible to apply each regulation to the relevant part of the dataset (LeMO 2018). In addition, as mentioned in the previous sections, these concepts are broad and subject to the dynamic nature of contextual adaptation. Whether data has economic value is not dependent on its legal classification. Hence, when facing opaque datasets, there is the risk of strategic firms on the basis of this legal classification, and they are likely to exploit the regulatory rivalry between the FFoD and the GDPR. The limitation of the FFoD to non-personal data is likely to be counterproductive to innovation, as personal data has high innovation potential as well (Graef et al. 2018). There is also further guidance needed where it concerns parallel/subsequent application of the GDPR and the FFoD, or where the two regimes undermine each other (Graef et al. 2018). Regardless of whether data is personal or non-personal, it is of major importance that it is secured. Hence, the following section addresses the EU regime on the security of data (Fig.1).
3.1.4 Security of Data
The Cybersecurity Act (Regulation (EU) 2019/881) was adopted to set up a certification framework to ensure a common cybersecurity approach throughout the EU. The aim of this regulation is to improve the security standards of digital products and services throughout the European internal market. These schemes are currently voluntary and aimed at protecting data against accidental or unauthorised storage, processing, access, disclosure, destruction, loss or alteration. The EC will decide by 2034 whether the schemes will become mandatory.
The NIS Directive (Directive (EU) 2016/1148) puts forward security measures for networks and information systems to achieve a common level of cybersecurity throughout the European Union to improve the functioning of the internal market. The security requirements that the NIS Directive puts forward are of both a technical and organisational nature for operators of essential services and digital service providers. If a network or information system contains personal data, then the GDPR is most likely to prevail in case of conflict between the two regimes. It has been argued that the regimes of the GDPR and the NIS Directive have to be regarded as complementary (Markopoulou et al. 2019). Cyberattacks are becoming more complex at a very high pace (Kettani and Wainwright 2019Footnote 25). The nature of the state of play is constantly evolving, which makes it more difficult to defend against attacks. Also, it has been predicted that data analytics will be used for mitigating threats but also for developing threats (Kettani and Wainwright 2019). The companies that can offer enough cybersecurity are non-European, and the number of solutions is very limited (ECSO 2017). Due to the characteristics of the digital world, geographical boundaries are disappearing, and a report by the WRR (the Dutch Scientific CouncilFootnote 26) called for attention to cybersecurity at an EU level.
Some of the characteristics of cybersecurity make tackling this challenge especially difficult; fast-paced evolvement, lack of boundaries, the fact that infrastructures are owned by private parties and the dependence of society on these architectures are recurring issues (ECSO 2017Footnote 27). Currently, cyber-strategies of SMEs mainly focus on the detection of cyber risks, but these strategies should shift towards threat prevention (Bushby 2019). Just like data and robotics, AI faces all of the possible cyberthreats, and every day threats are only further evolving. Cybersecurity will also play a key role in ensuring technical robustness, resiliency and dependability. AI can be used for sophisticated automated attacks and at the same time also to provide automated protection from attacks. It is important that cybersecurity is integrated into the design of a system from the beginning so that attacks are prevented.
This section has discussed the EU regime on the security of both personal and non-personal data. Cybersecurity attacks are continually evolving and pose challenges for those involved in a data ecosystem. Keeping different types of data secure is one aspect, but successfully establishing rights upon data is another. The next section addresses the interaction between data and intellectual property rights and data ownership.
3.1.5 Intellectual Property
Due to the fact that many different players are involved in the big data lifecycle, many will try to claim rights in (part of) the datasets to protect their investment. This can be done by means of intellectual property rights. If the exercise of such a right is not done for the right reasons, this can stifle the uptake of big data and innovation. This also holds true for the cases in which an intellectual property right does not exist yet is enforced by an actor that is economically strong.
3.1.6 Public Sector Information and the Database Directive
In January 2019, an agreement was reached on the revised Public Sector Information Directive (PSI Directive). Once implemented, it will be called the Open Data and Public Sector Information Directive. The revised rules still need to be formally adopted at the time of publication of this deliverable. Public bodies hold huge amounts of data that are currently unexploited. The access and re-use of raw data that public bodies collect are valuable for the uptake of digital innovation services and better policymaking. The aim of the PSI Directive is to get rid of the barriers that currently prevent this by reducing the market entry barriers, increasing the availability of data, minimising the risk of excessive first-mover advantages and increasing the opportunities for businesses.Footnote 28 This will contribute to the growth of the EU economy and the uptake of AI. The PSI Directive imposes a right to re-use data, obliges public bodies to charge the marginal cost for the data (with a limited number of exceptions), stimulates the uptake of APIs, extends the scope to data held by public undertakings, poses rules on exclusive agreements and refers to a machine-readable format when making the data available. Although open data licences are stimulated by the PSI, they can still vary widely between member states. Another challenging aspect is the commercial interests of public bodies in order to prevent distortions of competition in the relevant market. Some of the challenges that the use of public sector information faces are related to standardisation and interoperability, ensuring sufficient data quality and timely data publication, and a need for more real-time access to dynamic data. In addition, the licences to use the data can still vary, as member states are not obliged to use the standard formats. Another challenge that the PSI Directive faces is its interaction with the GDPR, either because it prevents disclosure of large parts of PSI datasets or because it creates compliance issues. The GDPR is not applicable to anonymous data. In practice, however, it is very hard for data to be truly rendered anonymous, and it cannot be excluded that data from a public dataset, combined with data from third-party sources, (indirectly) allows for identification of individuals. The interaction between the GDPR and the PSI Directive is also difficult with respect to public datasets that hold personal data, especially because of the principle of purpose limitation and the principles of data minimisation (LeMO 2018). Another challenge is the relationship of the PSI Directive with the Database Directive (DbD), as public sector bodies can prevent or restrict the re-use of the content of a database by invoking its sui generis database right. How the terms ‘prevent’ and ‘restrict’ are to be interpreted is not clear yet. Exercise of these rights bears the risk of hindering innovation. Where it concerns data portability requirements, the interaction between the DbD, PSI Directive and the GDPR is not clear either (Graef et al. 2018).
In 2018, the Database Directive (hereafter: DbD) was evaluated for the second time. The DbD protects databases by means of copyright or by means of the substantial investment that was made to create it, the sui generis right. The outcome of the evaluation was that the DbD is still relevant due to its harmonising effect. The sui generis right does not apply to machine-generated data, IT devices, big data and AI. At the time of the evaluation, a reformation of the DbD to keep pace with these developments was considered too early and disproportionate. Throughout its evaluation, one of the challenges was measuring its actual regulatory effects.
3.1.7 Copyright Reform
As part of the Digital Single Market Strategy, the EU is revising the rules on copyright to make sure that they are fit for the digital age. In 2019, the Council of Europe gave its green light to the new Copyright Directive (European Parliament, 2019). The aim is to ensure a good balance between copyright and the relevant public body objectives, such as education, research innovation and the needs of persons with disabilities. It also includes two new exceptions for Text and Data Mining (TDM), which allows for TDM for the purpose of scientific researchFootnote 29 and the opt-out clause of Article 4 New Copyright Directive. This exception will be of special importance to the uptake of AI. In a big data context, it is difficult to obtain authorisation from the copyright holder of individual data. When a work is protected by copyright, the authorisation of the rights holder is necessary in order to use the work. In a big data context, this would mean that for every individual piece of data, the authorisation needs to be obtained from the rights holder. Also, not all data in a big data context is likely to meet the originality threshold for copyright protection, though this does not exclude the data from enjoying protection under copyright. This creates uncertainties on which data is protected and which data is not, and whether a work enjoys copyright protection can only be confirmed afterwards by a court as copyright does not provide a registration system. The copyright regime is not fully harmonised throughout the EU, and a separate assessment is required on whether copyright protection is provided. This bears the potential of having a chilling effect on the uptake of EU-wide big data protection. Regarding AI-generated works of patents, it is still unclear whether, and if so to whom, the rights will be allocated. The multi-stakeholder aspect plays a role here as well, and the allocation of rights is difficult.
The manner in which intellectual property rights on data will be exercised will have a significant impact on the uptake of big data and innovation in general. This will all be shaped by the interaction between the PSI Directive, the GDPR and the new Copyright Directive. These are all instruments to establish security on data in the form of a right, as this is currently lacking.
3.1.8 Data Ownership
There is no particular framework to regulate the ownership of data. Currently, the only means to establish ownership in data or protection of data is through the provisions of the GDPR, the DbD and the Trade Secrets Protection Directive, or by contracts through contract law. Whether there should be an ownership right in data has been widely debated in recent years, as this current framework does not sufficiently or adequately respond to the needs of all the actors involved in the data value cycle. At the same time, there is consensus that a data ownership right is not desirable, as granting data ownership rights is considered to create an over-protective regime with increased data fragmentation and high transaction costsFootnote 30. The difficulty of assigning ownership to data lies in the nature of data, because it is neither tangible nor intangible, it is limitless and non-rivalrous, and its meaning and value are not static. Data has a lifecycle of its own with many stakeholders involved. This also implies that no stakeholder will hold exclusive ownership rights over the data. The lack of a clear regulatory regime creates high levels of legal uncertainty. Ownership is currently mainly captured by contractual arrangements. This situation is far from ideal, as it creates lock-in effects and power asymmetries between parties, and is non-enforceable against third parties. However, the fact that there is no legal form of ownership does not prevent a de facto form of ownership from arising either. The rise of data bargaining markets illustrates this. The de facto ownership of data does not produce an allocation that maximises social welfare. This results in market failures, strategic behaviour by firms and high transaction costs. There is a need for policies and regulations that treat ‘data as a commodity’. This requires new architectures, technologies and concepts that allow sellers and buyers of data to link and give appropriate value, context, quality and usage to data in a sense that ensures ownership and privacy where necessary.Footnote 31 In the next section, we will elaborate how this plays out in the data economy.
3.1.9 Data Economy
The digital economy is characterised by extreme returns based on scale and network effects, network externalities and the role of data in developing new and innovative services. As a result, the digital economy has strong economies of scope with large incumbent players who are difficult to dislodge. In order to realise the European Digital Single Market, we need the conditions that allow for the realisation thereof. Moreover, AI and the IoT are dependent on data; the uptake of both will be dependent on the data framework.Footnote 32
There have been many developments in the field of competition law that are of importance for the regulation of big data. The legal principles of competition law stem from a time when the digital economy did not even exist yet. It has been widely debated whether the current concepts of competition law policy are sufficient tools to regulate emerging technologies or whether new tools are needed. Currently, there is still a lot of legal uncertainty concerning the practical implementation of competition law related to the data economy due to its lack of precedents. The concepts of, among others, the consumer welfare standard, the market definition and the manner in which market power is measured need to be adapted or refined in order to keep up with the digital economy (European Commission, Report - Competition policy for the Digital Era,Footnote 33). The question of whether big tech must be broken up was often asked in competition policy debates. Facebook is currently under investigation by the US Federal Trade Commission for potentially harming competition, and Federal Trade Commission Chairman Joe Simons has stated in an interview with Bloomberg that he is prepared to undo past mergers if this is deemed necessary to restore competition. However, there are no precedents on breaking up big tech firms, and knowledge on how to do this if considered desirable is currently lacking.Footnote 34 The aim of some of the projects that are a part of the BDVA (Zillner et al. 2017) is to make sure that we as an EU landscape become stronger through data sharing, not by aiming to create another company that becomes too powerful to fail (e.g. GAFAM). The overall aim of DataBenchFootnote 35 is to investigate the current big data benchmarking tools and projects currently in operation and to identify the main gaps and provide metrics to compare the outcomes that result from those tools. The most relevant objective mentioned by many of the BDVA-related projects is to build a consensus and reach out to key industrial communities. In doing so, the project can ensure that the activity of benchmarking of big data activities is related to the actual needs and problems within different industries. Due to rules imposed by the GDPR, the new copyright rules on content monitoring and potential rules on terrorist content monitoring,Footnote 36 and realising the complexity of tasks and costs that all such regulations introduce, for the moment only large international technology companies are equipped to take up these tasks efficiently. As of this moment, there is no established consensus on how to make regulation balanced, meaning accessible and enforceable.
Over the last couple of years, several competition authorities have been active with competition law in enforcement regarding big tech. For instance, the EC has started a formal investigation into Amazon as to whether they are using sales data (which becomes available as a result of using the platform) to compete unfairly.Footnote 37 In addition, several national competition authorities have taken action to tackle market failures causing privacy issues by using instruments of competition law.Footnote 38 For example, on 7 February 2019, the German Bundeskartellamt accused Facebook of abusing its dominant position (Art. 102 TFEU) by using exploitative terms and conditions for their services. The exploitative abuse consisted of using personal data which was obtained in breach of the principles of EU data protection law. The Bundeskartellamt used the standards of EU data protection law as a qualitative parameter to examine whether Facebook had abused its dominant position. The European Data Protection Board (EDPB) also stated that where a significant merger is assessed in the technology sector, longer-term implications of the protection of economic interests, data protection and consumer rights have to be taken into account. The interaction between competition law and the GDPR is unclear, and it seems like we are experiencing a merger of the regimes, to a certain extent.
It has been considered that if substantive principles of data protection and consumer law are integrated into competition law analysis, the ability of competition authorities to tackle new forms of commercial conduct will be strengthened. If a more consistent approach in the application and enforcement of the regimes is pursued, novel rules will only be necessary where actual legal gaps occur (Graef et al. 2018). It is also been argued that, even though there are shared similarities between the regimes of competition law, consumer protection law and data protection law because they all aim to protect the welfare of individuals, competition law is not the most suitable instrument to tackle these market failures (Ohlhausen and Okuliar 2015; Manne and Sperry 2015) because each regime pursues different objectives (Wiedemann and Botta 2019). Currently, the struggle of National Competition Authorities in tackling the market failures in the digital economy creates uncertainties about how the different regimes (of competition and data protection) interact, and this creates legal uncertainty for firms.
Even though competition authorities have been prominent players in the regulation of data, the lack of precedent creates much uncertainty for companies. The next section will discuss how data sharing and access, interoperability and standards play a role in this.
3.1.11 Data Sharing and Accessibility
Data is a key resource for economic growth and societal progress, but its full potential cannot be reaped when it remains analysed in silos (EC COM/2017/09). More industries are becoming digitised and will be more reliant on data as an input factor. There is a need for a structure within the data market that allows for more collaboration between parties with respect to data. Data access, interoperability and portability are of major importance to foster this desired collaboration. In this respect, data integrity and standardisation are reoccurring issues. Accessibility and re-use of data are becoming more common in several industries, and sector-specific interpretations of the concept could have spill-over effects across the data economy. There is a need for governance and regulation to support collaborative practices. Currently, data flows are captured by data-sharing agreements.
The complexity of data flows, due to the number of involved actors and the different sources and algorithms used, makes these issues complicated for the parties involved. The terms in data-sharing agreements are often rather restrictive in the sense that only limited access is provided. This is not ideal, as restriction in one part of the value chain can have an effect on other parts of the data cycle. Access to data is mainly restricted because of commercial considerations. An interviewee suggested that the main reason that full data access is restricted is that it allows the holder of the entire dataset to control its position on the relevant market, not because of the potential value that lies in the dataset.Footnote 39 Parties are often not aware of the importance of having full access to the data that their assets produce, resulting in the acceptance of unfavourable contractual clauses. The interviewee also suggested, however, that the real value creation does not lie in the data itself, but in the manner in which it is processed, for instance by combining and matchmaking datasets. In addition, there is a lack of certainty regarding liability issues in data-sharing agreements. Data-sharing obligations are currently being adopted in certain sectors and industries, for instance in the transport sector (LeMO 2018Footnote 40), though due to the absence of a comprehensive legal framework, these still face numerous limitations. In some cases, the imposition of a data-sharing obligation might not be necessary as data plays a different role in different market sectors. It is worthwhile to monitor how the conditions imposed by the PSI Directive on re-use and access for public sector bodies play out in practice to see whether this could also provide a solution in the private sector (LeMO 2018).
The right to data portability of Article 20 GDPR (RtDP) is a mechanism that can facilitate the sharing and re-use of data, but regarding its scope and meaning, many areas are still unresolved. For instance, a data transfer may be required by the data subject where this is considered ‘technically feasible’, though what circumstances are considered to be ‘technically feasible’ by the legislator are not clear. In addition, there is no clarity on whether the RtDP also applies to real-time streams, as it was mainly envisaged in a static setting. There is also a strong need to consider the relationship between the right to data portability and IP rights, as it is not clear to what extent companies are able to invoke their IP rights on datasets that hold data about data subjects.Footnote 41 The interpretation of these concepts will make a big difference with respect to competition, as the right to data portability is the main means for data subjects to assay the counter-offers of the competitors for the services they use without the risk of losing their data. However, if competition law has to enforce the implementation and enforcement of interoperability standards that ensure portability, it will be overburdened in the long run.
The sharing and re-use of data require that effective standards are set across the relevant industry. Currently, the standardisation process is left to the market, but the efficient standards are still lacking, and this slows down data flows. Setting efficient standards will smoothen the process of data sharing and therefore also encourage it. Each market has its own dynamics, so the significance of data and data access will also be market dependent. In the standardisation process, it needs to be taken into account that a standard in one market might not work in another. Guidance on the creation of standards is needed to provide more legal certainty, because if this process is left to the market alone, this can result in market failures or standards that raise rivals’ costs. The role of experts in the standardisation process is crucial, as a deep understanding of the technology will lead to better standards. In addition, due to the multidisciplinary nature of many emerging technologies, the regulator should not address the issue through silos of law but have a holistic approach and work in regulatory teams consisting of regulatory experts that have knowledge of the fields relevant in setting the standard.Footnote 42
Data access, interoperability, sharing and standards are important enabling factors for the data economy. The manner in which the data economy will be shaped will have an impact on commerce, consumers and their online privacy. The next section discusses these three points.
3.1.12 Consumers, e-Commerce and e-Privacy
In January 2018, the Payment Services Directive (PSD2) became applicable. This Directive was expected to make electronic payments cheaper, easier and safer. On 11 April 2018, the EC adopted the ‘New Deal for Consumers’ package. This proposal provides for more transparency in online marketplaces and extends the protection of consumers in respect of digital services, as they do not pay with money but with their personal data. The new geo-blocking regulation that entered into force will prohibit the automatic redirecting and blocking of access, the imposition of different general conditions to goods and services, and payment transactions based on consumer nationality. Furthermore, the EU has been working on the revision of the Civil Procedure Code regulation on consumer protection (Regulation (EC) 2017/2394), which entered into force on 17 January 2020. The new rules for VAT for the online sale of goods and services will enter into force in 2021. The Digital Services Act is a piece of legislation which is planned to tear up the 20-year-old e-Commerce Directive; it also targets Internet Service Providers (ISPs) and cloud services. It is likely to contain rules on transparency for political advertising and force big tech platforms to subject their algorithms to regulatory scrutiny (Khan and Murgia 2019). In the Communication on online platforms (Communication 2016 288Footnote 43), the EC formulated principles for online platforms. These are about creating a level playing field, responsible behaviour that protects core values, transparency and fairness for maintaining user trust, and safeguarding innovation and open and non-discriminatory markets within a data-driven economy. Following this Communication, on 12 June 2020, the Regulation on platform-to-business relations (Regulation (EU) 2019/1150) was adjusted and is now applicable. The objective is to ensure a fair, predictable, sustainable and trusted online business environment within the internal market. Due to the scale and effects of platforms, this measure is taken at EU level instead of member state level. It applies to online intermediation services, business users and corporate website users, and it applies as soon as the business user or the corporate website user has an establishment within the EU. It sets requirements for the terms and conditions, imposes transparency requirements and offers redress opportunities.
The European Data Protection Supervisor has stressed the urgency for new e-privacy laws (Zanfir-Fortuna 2018), and since the publication of the previous deliverable in 2017, the e-Privacy Directive has been under review. Several governments and institutions have expressed their opinion on its current new draft. For example, the German government has stated that they do not support the current draft version as it does not achieve the objective of guaranteeing a higher level of protection than the GDPR,Footnote 44 and the Dutch Data Protection Authority has stated that cookie walls do not comply with EU data protection laws.Footnote 45 Furthermore, in October 2019, the Court of Justice of the European Union (CJEU) gave its decision in the Planet49 case (C-673/17, ECLI:EU:C:2019:801) and stated that the consent which a website user must give for the storage of and access to cookies is not valid when this consent is given by means of a pre-ticked checkbox. In addition, information that the service provider gives to the user must include the duration of the operation of cookies and whether or not third parties may have access to these cookies. This judgement will have a significant impact on the field of e-privacy and on big data in general as well, as a lot of the data that ‘forms part of big data’ was gathered and processed on the basis of pre-clicked consent-box cookies. Thus, this judgement will change how data should be processed from now on.Footnote 46 In extension thereof, the case Orange Romania (C-61/19) is currently pending at the CJEU for a preliminary ruling on what conditions must be fulfilled in order for consent to be freely given.