On 12 September, the Federation of European Publishers, in collaboration with the Cooperation of Greek Book Publisher Associations, organised an Aldus event “Technology, law and economics: the rationale”, as part of the FEP Summer Meeting in Athens. The theme of the session was defined having in mind how, while traditionally publisher associations needed to understand the impact of copyright legislation on the economics of their members, today this becomes very hard without also understanding the evolution of technology. Technological skills are in fact now essential complements to the legal and economic ones.
Phaedon Kidoniatis, publisher delegate of the Cooperation of Greek Publisher Associations to FEP, and Rudy Vanschoonbeek, President of FEP, welcomed the participants. Mr Vanschoonbeek explained that the event programme had been drafted to help publisher representatives put into perspective the technologies affecting the present and future of publishing, with a view to finding out especially how emerging technologies can facilitate the work of publishers.
Giulia Marangoni, an expert at the International Affairs and R&D office of the Italian Publishers Association, made a presentation on “Technologies for rights data management”. Mrs Marangoni introduced the Association, which was celebrating 150 years of activity, highlighting the role of its dedicated R&D group, working at innovation in publishing in areas such as standards, identifiers and metadata for the book trade; technologies for rights data management; and accessibility of digital publications. This department followed the development of the Copyright in the DSM Directive both from a normative and a technological perspective, driven by the view that technological innovation for rights management and rights information management (two related but distinct notions) was crucial for the successful implementation of the directive.
Several provisions of the directive indeed call for developing innovative technological solutions to manage rights information. Many articles of primary interest for book publishers entail the design of innovative answers for managing information on rights and licenses:
- The exception for illustration for teaching (art.5) allows Member States to decide that licenses prevail over the exception – provided those licences are easily available on the market and suitable to users’ need This requirement calls for tools to enhance the discoverability and access to licensing information, by linking rights data to digital content. In this area, publishers have already started investing in digital solutions that offer more flexible, granular licenses, tailored on users’ needs (for example, academic content platforms for universities).
- The exception for text and data mining of lawfully accessible works (art. 4) provides that rightholders can choose to reserve the rights to that use in an appropriate manner, including by machine-readable means. This provision requires the implementation of user-friendly, affordable tools supporting rightholders (also in the case of user-generated content) in the expression of rights information in a standard format. The adoption of open international standards for rights expression is essential to ensure that all rightholders can reserve their rights.
- Articles 8 to 11 regulate the use of out-of-commerce works by cultural institutions under licenses granted by relevant collective management organisations and provide for the creation of an out-of-commerce works portal by the EUIPO. Technology solutions (based on the experience of the ARROW project, for instance) could support cultural heritage institutions interested in digitising and making available their collections of works no longer in commerce in the identification of the rights status of the works.
- Article 15 establishes for press publishers a specific right for the digital use of their press publications by information society service providers (such as Google News). In this framework, licensing information can be expressed in a standard, machine-readable format, to streamline interactions with news aggregators. This creates also opportunities for the deployment of blockchain technologies for automated IPR transactions and micro-payments
- Article 17 (known as the “value gap” provision) establishes a regime of liability of online content sharing services for copyrighted content uploaded by their users and requires them to obtain a license from rightholders, if the latter are willing to offer one. Best efforts are required to prevent the availability of unauthorised content on platforms. Technology solutions for content identification (standard identifiers, content recognition), communication of rights ownership data and licensing information using standard, machine-readable formats can be very helpful in these scenarios. Once again, blockchain for automated rights transactions and micro-payments could be deployed to facilitate the necessary interactions; also, artificial intelligence applications (such as deep learning solutions) can support the identification of copyrighted content.
Mrs Marangoni briefly recounted the experience of the Italian Publishers Association in designing technological solutions for rights and rights information management; this includes work in a series of projects in the last decade: ARROW (Accessible Registries of Rights Information and Orphan Works towards Europeana), LCC (Linked Content Coalition), The Copyright Hub, RDI (Rights Data Integration) and ARDITO (Access to Rights Data via Identification Technology Optimisation).
The latter project aimed at providing simple tools and services to support creators and SMEs in the creative content sector, to find new business ideas through monetising the re-use of their content in the digital world. ARDITO tools are designed to facilitate access to rights and licensing information for creative content in a digital environment (books and ebooks, images and audiovisual products). This is achieved through standard persistent identifiers (DOI, ISBN), content identification technologies (watermarks, fingerprinting, visual recognition systems) and resolution technologies enabling access to rights information from the point of discovery of the content asset itself. Thanks to such tools, users – including machines – can access rights and licensing information linked to digital content anytime, anywhere.
Mrs Marangoni made a demonstration of the functioning of several of these tools. The first demo showed how, thanks to a plug-in for browser developed by The Copyright Hub, a user can get permission to use an image found online, by being presented a digital rightholder statement; the system is based on fingerprinting – it works also machine to machine. The second, in the text-based content domain, illustrated how the system can identify a work from an excerpt and provide a user with the relevant rights information; such information can also be linked to an ISBN, a parallel service offered by the Italian Publishers Association. Finally, a use case for collective management organisations proved how these tools can be integrated in an RRO environment, to complement licences they offer.
The key principles implemented by ARDITO, as Mrs Marangoni explained, are: rights information management is an independent function from rights management, a distinction that is essential to preserve open competition in the rights supply chain; it should remain distributed, and be based on standards for identification, communication, and description of the components of an IP-rights transaction; solutions should be neutral to business models (commercial vs. not-for-profit, direct vs. collective licenses, subscriptions vs. pay per use, and so on) and to content types. Future developments of this R&D stream include exploring the use of artificial intelligence to enhance content recognition technologies and intelligent agents to enable automated IPR transactions, and the use of blockchain technologies to enable micro-licensing and micro-payments.
Efstratios Tzoannos, Chief Engineer and Product Manager at ATC (Athens Technology Center), addressed the topic “Blockchain and the Future of Licensing – Hype or Reality?”, outlining opportunities and practical experiences related to how blockchain is disrupting media and entertainment and how it can support innovation in copyright management.
Mr Tzoannos started with an explanation of the basics of blockchain, which is defined as a Distributed Ledger of Transactions (DLT), shared between peers, without a single controlling entity, so that the tampering of records is not possible (in principle), users have the ability to monitor transactions execution in real time and various degrees of anonymity, depending on tech selection, can be achieved. Blockchain technology has been around since 2008 but became popular only recently. It can be exemplified in comparison with the traditional banking system: in it, every transaction goes via a bank, which has the ledger; a DLT removes the need for central authority, since everyone has a copy of the ledger and updates their records, there is a validation process that ensures a transaction is allowed and fine and encryption ensures it cannot be tampered with.
Blockchain’s core features are: disintermediation – no intermediaries are needed, and options enabled include anonymity or know your customer processes; immutability – stored transactions do not change, which permits to implement tracking & security procedures; and Smart Contracts – enabled by the facilitation of negotiations and execution of transactions. Smart Contracts can be thus illustrated: an option contract is written as code into a blockchain; the contract is part of the public blockchain; the parties involved in the contract are anonymous; the contract executes itself when conditions are met; regulators can use blockchain to keep an eye on contracts.
Blockchain applications can come in different flavours: public, private, hybrid. Examples include Ethereum, Quorum, Ripple, R3 Corda, IBM, Hyperledger, Multichain, IOTA. Many myths have arisen around blockchain, which Mr Tzoannos exposed as not grounded in reality: blockchain is Bitcoin, used for criminal activity; there is only one blockchain; blockchain will replace banks; blockchain can be used for anything and everything; blockchain is always open and free; blockchain records can never be hacked; smart contracts are the same as legal contracts.
Mr Tzoannos then illustrated the possible uses of blockchain for copyright management. The technology’s characteristics that can be applied in this field are distributed ledgers, tokenisation and digital scarcity, smart contracts, decentralisation and disintermediation.
A distributed ledger provides for transparency: (everyone with access can see a full transaction history), security (only consistent entries can be approved for being added to the DL), ownership tracking (ledgers can also be used to keep track of tokens belonging to specific accounts) and integrity (DLTs can ensure that the transactions are consistent over time, and tokens are not spent twice). For example, if tokens represent rights, and wallet holders represent rightholders, DLTs may host public copyright registries, which record—in a transparent manner—the ownership, distribution, use and remuneration of works. Another example: a collective management organisation can set up a private distributed database to facilitate the interorganisational identification of works and payment of royalties.
The notion of tokenisation implies that any kind of information may be expressed as a token, using a cryptographic signature. Examples include: a copy of a protected work; a record of rights information management for protected content; information such as licensing terms (static/general or dynamic/conditional); remuneration tokens for the use of a work.
An example of smart contract can be as follows: a user pays for getting digital access to a creative work; the creator is automatically remunerated. Smart contracts should allow to automate a large volume of similar transactions, lower transaction cost, enact faster transactions, standardise licensing terms and conditions across uses and different jurisdictions, address and resolve jurisdictional conflicts and organise the distribution of revenue.
Regarding decentralisation and disintermediation, the idea to completely remove middlemen raises doubts whether it is hype or reality. It is thought by some to be applicable to publishers and music labels, CMOs, but also online platforms, including those that host user-uploaded content.
Mr Tzoannos described the current environment in the creative sector as one in which content creators are separated from content consumers by a series of distribution channels, a situation characterised by smaller cuts of revenue getting to creators, who have no control over pricing, distribution and advertising and no access to audience, and need to deal with time consuming contracts and payments. He proposed his view of a creative industry redesigned: a creator-centric model for blockchain, with no intermediaries, based on P2P content sharing and using smart contracts, in which creators could define pricing models, monetise content fast and easy, track transactions and access audiences.
Mr Tzoannos finally illustrated the Bloomen project, a project on copyright management and blockchain co-funded by the EU for 2017-2020 of whose consortium ATC is a member. The consortium has developed use cases in several domains: to protect copyright and enhance access for web TV, an audiovisual content acquisition, copyright management and sales solution for TV streaming services; to improve efficiency and transparency in the music sector, a global music information database for rights management and claims that engages all relevant interested parties; to identify rights, ownership and compensation with regard to news media content, an acquisition and management tool for news picture content that aims to improve visual journalism and collaboration with external picture contributors. Mr Tzoannos also provided a practical example related to the photo marketplace.
Colin Lovrinovic, Managing Director of Gould Finch, spoke about “Understanding the current and future impact of AI on the publishing industry”. Mr Lovrinovic, after introducing his company, provided a brief definition of AI, explaining that machines performing tasks perceived as intelligent are the subject of what is typically called Artificial Intelligence. AI fields, among others, contain knowledge-based systems, robotics, computer vision, NLP and machine learning (the ability of a machine to learn, usually from data, especially in large amounts). In the last decade, he explained, machine learning saw technological breakthroughs and widespread use in commercial applications.
More commonly, Mr Lovrinovic noted, AI is defined as “Whatever we want to achieve in 10 years”. The potential of AI – he argued – is virtually limitless (a huge economic windfall is expected), as AI can perform certain task faster (e.g. OCR/digitizing books), cheaper (e.g. speech-to-text for blind people), better (e.g. complex recommendation engines), more reliably (e.g. automated payment/reporting or finding typos) and more relentlessly (e.g. 24/7 customer service chatbots) than humans can. It can even do some things humans can’t do at all.
Certainly, there are limitations in reality, things AI cannot do, things that only humans can. AI can achieve very high levels of accuracy, but costs and complexity increase exponentially with it: 80% accuracy is easy to achieve, 100% is very hard. In setting parameters, users need to ask themselves how many false predictions they can accept, and which impact they have: reading a boring book if AI fails in recommending readings, as opposed to being run over if AI fails in driving a car – for example.
The best application of AI, argued Mr Lovrinovic, is to well-defined problems with clear input that would take humans too long because of repetitive steps. A good example would be predicting which book someone might like to read next; a bad one, writing the next bestseller.
Mr Lovrinovic presented briefly the results of a study conducted by his company on the role that AI currently plays in the publishing industry (the results are available as a White Paper). Currently, only 25% of publishers surveyed work with AI in their companies; future investments in the field will likely happen in marketing and distribution (this is where AI can be the most beneficial in the short term, followed by (supportive functions of) editorial, but there also seems to potential in the optimisation of administration, production and press.
Mr Lovrinovic then outlined several use cases of AI in publishing: automated pricing and dynamic advertising; super-local news creation; a bookshelf recommendation engine; automatic translations (with DeepL); text-to-speech for audiobooks. However, despite the potential benefits and the many possible use cases, AI adoption is still limited.
The biggest challenges for the implementation of AI solutions are the financial investments required and the uncertainty about the return on such investments. There are also issues related to the necessary skills and competencies. Other challenges stem from the huge importance of mindset and cultural factors within companies, as well as from ethical questions and the need to strike the right balance of trust vs. fear.
Nevertheless, the survey revealed that a strong majority of publishers believe that AI will have a strong impact on their industry. Still, companies are hesitant to invest accordingly, for the reasons highlighted. Correlation reveals that companies who have experimented with AI seem to be happy with the results.
To sum things up, Mr Lovrinovic stated that surely AI will become a critical success factor, but not a magic wand. Publishers have understood this, but so far only few plan on taking the necessary steps. Those who have started are happy with the results and ROI, and sceptics need to lose their fear and understand the technology. Machines won’t write your books any time soon; publishers should not substitute your creatives with machines, but instead strengthen their core business by optimising everything else. The main advice for publishers is then basically to get started.