Blockchain as a database—proposal for a new test for the criterion of ‘independence’ in the legal definition of a database for the purposes of copyright and the sui generis right

Abstract

Technology’s exponential growth often outpaces that of the law. The persistence of outdated legal concepts that were not drafted with new technology in mind leads to legal uncertainty. This article focuses on one example of such a friction between old law and new technology, namely the eligibility of blockchain as a‘database’ for protection under the EU Database Directive, as implemented into UK copyright law. The most problematic requirement for blockchain as a candidate is that the material inside the database be ‘independent’. This can pose a significant hurdle for blockchain to succeed as the immutability of blockchain is ensured by the ‘linked-list’ structure in between the blocks and the combinational hashing of data within the individual block. This article examines this issue and proposes a solution to this quandary: to divide the data recorded on a blockchain into ‘content’ and ‘structure’, and confine the criterion of ‘independence’ to the former. In reaching this solution, the author examines previous literature on the different types of data that can be found in databases, as well as how the concept of ‘independence’ is understood by judges and academics. This article will be of practical significance for developers of non-open source blockchain applications who wish to protect their products as a database.

1. Introduction

Blockchain technology is one of the most disruptive technologies of Industry 4.0 and its use is no longer limited to private endeavours.¹ The underlying technology is also evolving, with ever-powerful hardware,² and blockchain is increasingly marketed as a product. The latest trend is to offer blockchain-as-a-service, which outsources the use of blockchain in a manner akin to ordinary cloud services.³

Blockchain can certainly protect IP by serving as an immutable register.⁴ This debate was revived with the advent of non-fungible tokens (eg regarding whether ‘minting’ is a ‘digital artistic performance’ containing intellectual creativity).⁵ If blockchain can protect IP, could IP also protect blockchain? Indeed, if blockchains are increasingly widespread and original in their design, the creators behind these innovations would have an interest in protecting their works from unauthorized reuse.

Blockchain is commonly thought of as a method for securing the storage of data in a decentralized environment, similar to a decentralized database of verified transactions. Over time, our traditional understanding of a ‘database’ as a centralized and paper-based amalgamation of information has morphed into what is now a handy, toolbox-like digital network of data. The protection of databases as IP was significantly enhanced by Directive 96/9/EC of the European Parliament and of the Council of 11 March 1996 on the legal protection of databases (‘Database Directive’).⁶ The Database Directive is still relevant post-Brexit. The UK implemented the Database Directive through the Copyright and Rights in Databases Regulations 1997 (‘Database Regulation’). This still has effect today as ‘retained’ EU law, pursuant to section 2(1) of the European Union (EU) (Withdrawal) Act 2018 (EUWA 2018). The High Court of England and Wales, in DRSP Holdings Ltd v O’Connor (DRSP), confirmed the Database Regulation along with the relevant Court of Justice of the European Union (CJEU) and domestic authorities remain good law.⁷

The Database Directive grants protection to databases through both copyright (Chapter II of the Database Directive) and a ‘sui generis’ right (Chapter III of the Database Directive),⁸ which is a standalone right that is divorced from other regimes. Most interestingly, the Database Directive introduces a new legal definition of what constitutes a ‘database’, which applies to both regimes.

This can be found in Article 1(2) of the Database Directive and, in UK law, undersection 3A(1) of the Copyright, Designs and Patents Act 1988 (CDPA 1988).⁹

A database can be ‘in any form’ (e.g. in print or electronically),¹⁰ but must adhere to three criteria¹¹: (i) a collection of independent works, data or other material (‘independence’), (ii) arranged in a systematic or methodical way (‘systematicity’) and (iii) individually accessible by electronic or other means (‘accessibility’). Out of these, the first criterion is by far the most problematic one for blockchain technology. The immutable quality of data inside the blockchain is attributable to the interdependent nature of the structural data that forms the blocks and the chain in between these blocks.

A strict application of the ‘independence’ criterion would lead to the disqualification of blockchain as a protectable subject matter under the Database Directive. It is argued this stems from the ill-suitedness of current laws to emerging technologies rather than a reasoned and targeted exclusion. It is probable that the legislators in 1996 did not envisage the advent of blockchain technology, which emerged a decade later. As blockchain is a valuable tool in disseminating verified data and developing a competitive information market, it is crucial to examine whether and how current laws can be adapted to this new technology.

This article explores how databases that record both independent and interdependent data like blockchain can be reconciled with the ‘independence’ criterion without contradicting the spirit of the Database Directive. To that end, this article proposes a new three-step legal test for assessing whether a database that includes both types of data can still meet the definition in Article 1(2) of the Database Directive and Section 3A(1) of the CDPA 1988. The functioning of the new test rests on the categorization of the data inside a database into two categories, namely ‘contents’ and ‘structural’ data. ‘Structural’ data inside a blockchain is first and foremost interdependent to ensure the immutability of the recorded data, as well as the systematicity and the accessibility of the blockchain as a whole. By contrast, the independence of its ‘contents’ is fact-sensitive as it depends on the specific application of the blockchain. This article argues the ‘independence’ criterion should only be applied to the latter category instead of the ‘structural’ data that forms the skeleton of the blockchain and guarantees its proper functioning. To inform this discussion, this article undertakes a literature review to examine how ‘independence’ is interpreted and how it is applied to data.

Section 1 briefly outlines the main components of a blockchain that are relevant to the proposal. Section 2 discusses the relevance of the Database Directive to blockchain technology. Section 3 conducts a review of judicial and academic commentary on the ‘independence’ criterion and examines how it fails to accommodate the components of a blockchain as outlined in Section 2. Section 4 proposes a new legal test for

‘independence’ by relying on the separation of data into ‘content’ and ‘structure’ and applying the ‘independence’ criterion to the latter. Section 5 then places the content’/‘structure’ dichotomy both in the context of Article 1(2) of the Database Directive and section 3A(1) of the CDPA 1988 and of the subsistence tests for copyright and the sui generis right.

2. The Wonders of a Blockchain

The original purpose of a blockchain is to create trust between computer nodes (ie participants to the blockchain) of unknown number and identity in relation to the existence of a state of affairs in the absence of a central authority.¹² This necessity informs how the various components of a blockchain (data structure, software, transaction data, etc) may fail the ‘independence’ criterion.

2.1. How blocks are added

‘Proof-of-work’ (PoW) blockchains will be used as an example here. This does not carry the implication that the proposal cannot apply to other blockchains. The term ‘PoW’ refers to the nature of the method that is agreed between participants of the network (or nodes) as to how a block is to be added. For instance, if members of an organization decide to create a book collection and agree that a new book can only be added if the cover is red, it becomes very easy for members to decide whether a new book should be added or not (as the only criteria they can apply is that of the cover colour).

The criterion PoW employs to decide whether to add a new block is whether sufficient computational power was expanded in adding the block. Such computational power will be accepted as sufficient once a puzzle is solved, which is essentially a mathematical equation that involves finding ‘x’ in ‘2 + x = 4’. The solution to the puzzle (ie ‘x’) is called the ‘nonce’ (short for ‘number used once’). The difficulty of the puzzle, which is specified inside the block, usually increases with each block that has been added.

The way in which the conditions for finding the nonce are determined (that is how the rest of the equation is decided) depends on the contents of the block itself.¹³ By way of example, a transaction is initiated between A and B such that A transfers B 1BTC. These data are pushed onto the network of participants and gathered in a provisional block. The conditions of the equation are then formed depending on data such as the value of the transaction, the parties, the time at which the transaction was initiated and the difficulty level of the puzzle. Once the ‘x’ is found by a participant of the network, the rest of the nodes verify the solution and agree on whether the block can be added to the chain permanently. The block is then ‘timestamped’ and receives an identifier number that is unique in the entire chain (called a ‘hash reference’). This identifier can be found in what is the called the block’s header. This can be likened to a document header containing an identifier for the document. The completion of the PoW allows for nodes to establish consensus on the validity of the block. The amount of computational power needed functions as a disincentive to interfere with registered data, since this would require the redoing of that puzzle. What is worth noting is that given the equation’s conditions are dependent on the contents of the provisional block, it also means that the nonce is dependent on the contents of the block.

2.2. How blocks are linked together

A block header only containing its own identifier would be tantamount to a block independently floating inside a blockchain. There are only two solutions to this issue. First, the freshly mined block could include the identifier of the block following it in the chain. This, however, is problematic where adding information to mined blocks or predicting a future block’s identifier is quasi-impossible. The solution is, then, to have each block include the identifier of the previous one.¹⁴ In other words, Block 100 has its own hash reference and the hash reference of Block 99. This ensures that not only 99 follows 100, but also that Block 100 cannot be placed in the chain anywhere but after Block 99 (since otherwise it would be wrong for Block 100’s header to contain Block 99’s identifier).

2.3. How data are rendered immutable

The immutability of data inside and in between blocks is ensured by what is termed as a ‘cryptographic hash function’, which is an encryption method that delivers an output (called a ‘hash string’) in the form of letters and numbers. The hash function in a blockchain produces an encryption that is asymmetric, because running the output of the function back through the function does not yield the original input. Two different functions need to be used to encrypt and decrypt, respectively. This corresponds to the public/private key divide in Bitcoin transactions. Public keys used to encrypt data are accessible by everyone, whereas the corresponding private key is exclusive to the account owner (such that only the intended receiver of the encoded message can decrypt it).

These one-way encryption functions are also ‘collision resistant’¹⁵: when the inputs are different, the outputs will also be different (such that the outputs produced will not ‘collide’). These functions are extremely sensitive to changes in the input. For instance, changing a single comma in a 100-page document will yield a completely different hash string. The use of these functions is an effective way of determining whether there has been an attempt to alter data.

The sensitivity of these functions to change is what underlies the immutability of the data inside the blockchain. At a first stage, the raw transactional data (eg ‘A transfers B 1BTC’) is hashed to produce a hash string. The latter output is combined with that of other transactional data and hashed again. This process of combination and hashing is repeated until a single hash string remains.¹⁶ In other words, if there are four pieces of transaction to be recorded in a block (say Ta, Tb, Tc and Td), the hash function will be run twice (to produce Ta-b and Tc-d, and then Ta-b-c-d). The last output will be called the ‘root of the Merkle tree’, where ‘Merkle tree’ refers to the structure of the combinational hashing performed in the block. As such, the root of the Merkle tree is dependent on the contents of the transactional data. If any node attempted to modify either a transaction or an interim hash string, the root of the Merkle tree would be automatically invalidated.

The root of the Merkle tree is then included in the block header and is instrumental in determining what the conditions of the puzzle will be (ie in determining ‘2 + x = 4ʹ). This means that the nonce is also dependent on the root of the Merkle tree and the transactional data recorded within. If anyone attempted to modify any such data in the block header, the nonce and the hash reference of the block itself would be invalidated. All data are essentially ‘locked in’ through one-way, collision-resistant hash functions.

The connection between blocks is ensured by the registering of the hash reference of the previous block in the following one. The hash reference of Block 99 forms the content that goes into the defining of conditions of the puzzle in Block 100. As the hash functions are content-sensitive, the change of a single value in the transactional data in Block 99 will invalidate all subsequent hash references, from Block 100 up until the last block in the chain. Rewriting a blockchain would involve the re-hashing of all such data, which disincentivises such changes and, thus, ensures the immutability of recorded data.

2.4. The types of data inside an individual block

The above demonstrates that a blockchain is in fact a gold mine of different types of data, with each having a different function within the blockchain ecosystem. Overall, it is possible to categorize data inside a single block into two distinct types. The first type of data is application-specific data. The nature of such data depends on the purpose for which the blockchain is being used (eg transactional data for the Bitcoin blockchain to patient data for a medical application).¹⁷

The second type of data is data inherent to the blockchain structure itself, irrespective of the application. This refers to the data necessary to the functioning of the blockchain. Most of this type of data is contained in the block header alongside the block’s identifier in the form of a hash reference. Overall, the block header contains:

(1) a timestamp (denoting when the block is added to the chain), (2) a nonce (solution to the puzzle), (3) the difficulty level (of the puzzle), (4) the root of the Merkle tree of transactional data, (5) the hash reference to the previous block and (6) the version of the software used.¹⁸ The distinction drawn here between the two types of data will be useful to this article’s proposal in classifying data as either ‘content’ or ‘structure’ below.

3. The Advantages of Database Rights for Blockchain

3.1. The relevance of the Database Directive for blockchain

The Database Directive is the ideal regime for the protection of blockchain-based databases for three reasons. First, the original purpose of blockchain— creating trust over the state of a distributed ledger among untrustworthy nodes—¹⁹ matches the policy objectives underlying the Database Directive. These policy objectives can be gleaned from the Recitals to the Directive: there, databases are described as ‘a vital tool’ in the establishment of an ‘information market’ in light of the ‘exponential growth’ of information generated in today’s world, thereby leading to a need for harmonized regimes for database protection.²⁰ Protecting these vital tools is crucial where information has become a valuable ‘tradeable commodity’,²¹ and where a database’s value largely depends on how ‘comprehensive’ it is in the informational value it harbours.²²

In that regard, the power of blockchain in creating an information market that is accessible worldwide, the veracity of which is mathematically verifiable, is undeniable. Blockchain increases the trustworthy dissemination of information in a market, which is an effective way of achieving the Database Directive’s aim in promoting the free flow of information.²³ The importance of blockchain as a database was particularly emphasized by Finck, who explained that blockchain would lead to ‘a profound paradigm shift regarding data collection, sharing and processing…’.²⁴ It is unsurprising that blockchain’s use for the secure exchange of information is more widespread than its use for cryptocurrencies.²⁵

Consequently, it would only benefit jurisdictions to bring blockchain within the ambit of the Database Directive. The blockchain business is particularly profitable—but such profitability depends on the ability of its creators to protect their investment and originality from unauthorized use and distribution.²⁶ If such a protection is not granted, there risks being a market failure where the billions of dollars invested in blockchain solutions are not met with an equally important protection regime.²⁷ This risks disincentivizing blockchain creators from investing in jurisdictions where their innovations could go unprotected. This would defeat the purpose of the Database Directive in developing a strong information market in the EU and run counter to financial initiatives like Horizon Europe grants.²⁸

Second, it is arguable that a blockchain-based database can satisfy the subsistence test for copyright and/or the sui generis right in the Database Directive. To date, the most famous applications have been marketed as an opensource.²⁹ This may explain the lack of case law on this matter, but this is unlikely to last much longer. The Law Society Guidance concluded that a sui generis right could subsist in blockchain.³⁰ Most recently, Mellor J granted permission to the claimants in Wright v BTC Core (‘Wright’) to serve a claim form outside of the jurisdiction on the basis that there was a serious issue to be tried concerning, inter alia, the claim that Dr Craig Steven Wright—allegedly Satoshi Nakamoto—³¹was the owner of database rights in the Bitcoin blockchain for transactions dating up to 1 August 2017.³² Unless the dispute is settled, UK courts will be asked to answer the above question for the first time.

For copyright, the database in question, by the selection or arrangement of its contents, must be the product of an ‘author’s expression of his creative ability in an original manner by making free and creative choices’ and thus his ‘personal touch’.³³ This can be satisfied if the selection of such content is carefully made following a creative idea (for instance, a compilation of an author’s poems that evoke the theme of rebirth). Two main difficulties arise for blockchain-based databases. First, there is little room for originality where a blockchain is to record accurately the ownership of an asset like Bitcoin, since no discretion can be exercised in the selection of data.³⁴ Second, the technical elements necessary to the functioning of a blockchain can limit an author’s ability to imbue the database with a ‘personal touch’.³⁵ As operating a blockchain is a highly technical activity (eg through cryptography and data processing), its functioning is more likely to be dictated by the functionality of such tools rather than the originality of its author.³⁶

A sui generis right may provide a safer bet for blockchain creators, which only requires qualitative and/or quantitative investment in the ‘obtaining, verifying or presenting’ of the contents.³⁷ The main difficulty would be to distinguish ‘obtaining, verifying and presenting’ information from ‘creating’ information.³⁸ Although transactional data can be seen as ‘pre-existing’ in nature, the process of adding blocks to the chain does create some data, such as in the adding of inter-block hash references. However, it is at least arguable that all three activities involve some form of substantial investment. Most notably, the process of verification through PoW involves both computational power and financial investment in the form of transaction fees.³⁹

It is worth noting that the number of participants to a blockchain network is necessarily more than one and the nature of their contribution differs according to their responsibilities: investors provide the capital for developing the blockchain-based database or the assets within, users push onto the network raw transactional data, miners expand computational power in completing PoW and verifiers confirm the puzzle solution. The numerousness of the stakeholders involved in the blockchain ecosystem may create uncertainty as to who owns the IP in the blockchain.

For copyright, the author would be the person originating the intellectually creative ideas behind the selection and arrangement of database contents.⁴⁰ If blockchain is a computer-generated work, the author will be the person that undertakes the ‘arrangements necessary for the creation of the work’,⁴¹ which can be the result of a collaboration of authors.⁴² This can be more readily ascertained in private blockchains with a smaller group of creators. For the sui generis right, the database ‘maker’ is the one who ‘takes the initiative’ in performing the activities of obtaining, verifying and presenting the data and assumes the ‘risk of investing’,⁴³ which can also be the result of a collaboration.⁴⁴ This can range from the businessman investing in a blockchain project, to the programmers implementing the data processing systems and to the nodes that carry day-to-day PoW activity on the chain.⁴⁵

In practice, due to the costs of designing and running a blockchain, the most widespread type of ‘database maker’ in the current market would be large investors financing the creation of blockchain-enabled platforms Goldman Sachs is the most likely candidate for qualifying as the ‘database maker’ for the Goldman Sachs Know-Your-Customer (KYC)/Anti-Money Laundering (AML) Blockchain Application, as it would assume the risk of investing in the KYC blockchain.⁴⁶ Realistically, any rights accruing to any other database makers mandated to produce that blockchain would be assigned to Goldman Sachs.⁴⁷ The availability of database rights may be of interest to investors anxious to hedge against the risk of their investment being reproduced without necessarily going through the hassle of a patent.

3.2. Advantages of database rights compared to patents

A third reason is that the Database Directive regime harbours quite a few advantages compared to the most sought-out type of protection for blockchain,⁴⁸ namely patents—a regime which appears to have organically carved out a special category for blockchains. The European Patent Office classifies blockchains as ‘computer-enabled inventions’,⁴⁹ while the UK categorizes them as ‘processes’ or ‘methods’.⁵⁰ Similarly, the US approach is to view them as ‘method claims’.⁵¹ This is exemplified by Goldman Sachs’ recent patent filing for a blockchain application that performs KYC and AML vetting in decentralized pseudonymous transactions (herein referred to as ‘the Goldman Sachs KYC/AML Blockchain Application’).⁵² Suppose two pseudonymous parties—A and B—wish to transact. A trusted vetting third-party issues A with a ‘globally unique identifier’ (‘GUID’) that signals A’s identity has been vetted (through a YES/NO indicator or a score).⁵³ Once a piece of data (whether relating to a party’s identity or the transaction) is vetted, the third-party issues a ‘Verifiable Credential (VC)’, which is essentially proof that KYC/AML checks have been carried out for that piece of data.⁵⁴ When A wishes to engage in a transaction with B through a smart contract, A may request from the trusted third-party an indication that both B’s identity and B’s involvement in the transaction are KYC/AML-compliant. The third-party can then issue a similar VC for B after B’s identity and its involvement are vetted.⁵⁵ Where this is unavailable, the third-party can issue a VC to indicate the level of uncertainty for KYC/AML-compliance.⁵⁶ Various other embodiments are also described in the patent application, for instance, the use of a fresh smart contract to render the exchange of one set of VCs conditional upon another such exchange,⁵⁷ or the collective vetting of all parties and transactional data before proceeding with the transaction. The patent is yet to be granted by the United States Patent and Trademark Office (USPTO).

Although patents remain the prevailing form of IP protection for blockchain, there are significant practical advantages to protecting blockchains through copyright or the sui generis right (‘SGR’) regime. First, although a ‘first-to-file’ system provides the earliest filer with a head-start on competitors,⁵⁸ a filing does not necessarily result in the patent being granted. The number of patents filed may show the scale of R&D in a given country that goes towards a specific technology or the number of companies dealing with such technology. It does not necessarily evince how apt the legal regime is in protecting such inventions. Contrastingly, the copyright and SGR regimes grant the database author or maker an automatic and unregistered IP right, which is advantageous for those lacking the means for filing patents. The costs for such filings were cited as a possible reason why the vast majority of UK patent filings at the United Kingdom Intellectual Property Office in 2017 came from foreign entities.⁵⁹ As such, a database right could have already risen for Goldman Sachs as soon as the database was made and before any patent filing.

Second, the UK is much stricter in its approach to subject matter excluded from patentability. The most likely exclusion is that of ‘program for a computer’.⁶⁰ A patent is unlikely to be granted for a computer software-enabled application on a blockchain (like a smart contract) as opposed to a blockchain per se being registered as a method patent (e.g. enabling KYC/AML processes in pseudonymous transactions by linking in a vetting third-party).⁶¹ Mitsubishi Electric Corporation’s patent filing in 2022 for a smart contract that issues a bond, instead of entering virtual currency into the smart contract, with the purpose of preventing a denial of service (DOS) attack was refused as falling within the computer program exception.⁶² The ‘technical effect’ contributed by the application lay entirely in a computer program, as opposed to a system within which a computer program could operate.⁶³ This can be compared to the technical contribution that the patented Hashgraph algorithm offers, which prevents forking (ie the rise of simultaneous and alternative chains in a blockchain) in the Hedera Hashgraph blockchain through a hashgraph software.⁶⁴

Another hurdle to patentability arises with the ‘business method’ exclusion,⁶⁵ which is especially detrimental for financial applications of blockchain that improve an existing service without qualifying as a ‘technical contribution’. Mitsubishi Electric Corporation’s filing also failed as a ‘method for doing business’: the application did not lessen the likelihood of a DOS attack in a technical way (eg by strengthening the code), but rather presented an alternative to loading cryptocurrencies onto a smart contract (ie by issuing a bond).⁶⁶ The improvement was of a commercial nature, directed at the efficiency of the payment system rather than securing the terminals through which the smart contract was executed.⁶⁷

Goldman Sachs’ patent could very well be caught by the business method exception if it were filed in the UK.⁶⁸ The first claim is worded as a ‘method for certifying compliance of a transaction’ with KYC/AML and other regulations on a blockchain while preserving pseudonymity between the parties using GUIDs and VCs for transactions.⁶⁹ This claim could be interpreted more as an improvement to business methods for KYC/AML vetting in a decentralized environment than a technical advancement in the blockchain code itself.

Database rights may yet be the more appropriate form of protection where the locus of the value behind the business model in question lies in the contents of the database, and not the operation of the proposed patented process. This may be for business ideas involving large collections of data where the verification, the obtaining or the presentation of that data is costly. The method for undertaking these activities may be incidental to the business idea itself.

The main hurdle for affording protection to a blockchain as a database for such applications mainly lies in meeting the definition of ‘database’ in Article 1(2) of the Database Directive and section 3A(1) of the CDPA 1988. Mitsubishi Electric Corporation’s smart contract application may fail as a ‘computer software’,⁷⁰ but Goldman Sachs’ application arguably does fit that definition.⁷¹ The requirements of systematic or methodical arrangement and the individual accessibility of the data are met by the very purpose of the blockchain, which is to grant access to the pseudonymous and vetted identifiers. The third-party certifications issued are linked to the unique identifiers, but each unique identifier is independent from the other (as are accounts in a traditional bank). The vetting process for the identifiers can be a qualitatively and quantitatively substantial investment in the verification of the contents of the database.⁷² Beyond this, whether the database is a method for doing business or involves the operation of a computer software in its making is irrelevant.

Third, difficulties with blockchain patents also arise with infringement claims. Proving infringement may be difficult in a context where asymmetric encryption does not allow for reverse engineering.⁷³ This could obstruct the aggrieved party from pointing at a tangible act that would constitute infringement or explain how the infringing product or process was obtained. In such a situation, quantifying the probability of succeeding in infringement proceedings may also be difficult where one cannot access the behind-the-scenes features of the allegedly infringing product or process. This could constitute a disincentive where seeking disclosure and pursuing the case in court involve painstaking legal costs if settlement fails.

The above difficulties can be avoided if one resorts to the rights provided in the Database Directive. A database protected by the SGR is infringed either if a substantial part of the database is extracted or reutilized,⁷⁴ or if an unsubstantial part is extracted or reutilized repeatedly and systematically.⁷⁵ A part is ‘substantial’, if it captures a substantial part of the qualitative or quantitative investment in verifying, obtaining or presenting the contents of the database,⁷⁶ irrespective of the physical size it occupies. Demonstrating this is considerably easier than attempting to determine the inner workings of a blockchain for the purposes of a process patent. How the infringing blockchain functions is irrelevant insofar as there is proof of substantial parts that have been reutilised or extracted from the protected database.

To illustrate, extracting a part of the vetted identifiers or vetted transactions in a block inside the Goldman Sachs KYC/AML Blockchain Application will qualify as an infringement if the extraction captures that qualitative or quantitative investment (which it most likely would following the vetting process). Besides how blocks are added or how transactions are verified in the infringing blockchain, the focus of the enquiry here is whether the database maker, through such extraction or reutilization, was deprived of ‘revenue which should have enabled them to redeem the cost of that investment’.⁷⁷ Beyond this, there is no further need to determine the precise method with which it was obtained, what steps were followed in between the use of the infringed database and the infringing database, or whether there was an intermediary involved (eg another company carrying out a second vetting service onto the extracted data). Retracing the database would be as much of an infringing act as saving the database on a hard drive—what matters is that the end-product is copied. In fact, the threshold for infringement is fairly low, where even the temporary transfer of a database to another medium—eg in consulting it on a computer screen—may constitute infringement.⁷⁸

Proving infringement in copyright requires at least proof that the parts reproduced by the allegedly infringing product are still the ‘expression of the intellectual creativity of the author’.⁷⁹ The threshold for demonstrating a causal connection between the protected and the infringing work is, similarly, easier than demonstrating infringement of a process patent. Showing that the same mistake figures in both the works may be sufficient to prove copying.⁸⁰

Overall, proving infringement of database rights entail in a more holistic and functional analysis that is more straightforward to meet than comparing the oft obscure processes involved in creating and maintaining a blockchain. The main advantage of a process patent over a database right is the protection it affords to a ‘way’ of doing things, such as ‘carrying out KYC/AML in a pseudonymous decentralised network’, which a database right may not adequately capture. Whilst this is a valid concern, it does not eliminate the usefulness of database rights as a fall-back position where a patent is refused. The existence of database rights does not render a patent filing automatically redundant, but provides a more accessible form of protection.

4. The Current Test for ‘Independence’

4.1. The meaning of ‘independent’

Despite its centrality, the criterion of ‘independence’ is not defined in the Database Directive. There are only two references to it: Recital 17, which provides that ‘a recording or an audio-visual, cinematographic, literary or musical work as such’ do not meet the definition of ‘database’, and Recital 20, which provides that ‘materials necessary for the operation or consultation of certain databases such as thesaurus and indexation systems’ may be protected. No assistance can be gleaned from international treaties either, as they only focus on the protection of databases as a copyrightable subject matter without defining it as a collection of ‘independent’ data.⁸¹

CJEU case law offers some guidance. In Fixtures Marketing Ltd v Organismos prognostikon agonon Podosfairou AW (OPAP) (OPAP), the CJEU held that ‘independent materials’ meant ‘materials which are separable from one another without their informative, literary, artistic, musical or other value being affected’.⁸² In other words, data must have ‘autonomous informative value’,⁸³ being ‘mutually’ independent from one another, but also independent within the collection.⁸⁴ The CJEU in Freistaat Bayern v Verlag Esterbauer GmbH (‘Verlag Esterbauer’) held that an analogue topographic map composed of two distinct pieces of information—the ‘geographical coordinates point’ and the ‘signature’ that corresponded to the item in that coordinate point—⁸⁵ was a collection of independent materials: the extraction of geographical information such as cyclist roads from that map could be used to produce another such map, and retained sufficient and autonomous informative value.⁸⁶

Academic literature offers a similar interpretation of ‘independent’. Tzoulia echoes the importance of the relational nature of data as emphasised in OPAP: one must look at whether ‘independent material [demonstrates] autonomous informative value in relation to the rest of the database content’ (emphasis added).⁸⁷ Data within a database must be separable from other content, while retaining the same meaning within and outside the collection.⁸⁸ This relates to the distinction Jenkins makes between ‘interdependent’ and ‘interrelated’ contents: while poems in an anthology are interrelated (perhaps by their themes), chapters in a novel are interdependent as one chapter cannot be understood without another.⁸⁹ Only a database with ‘interrelated’ data would satisfy the ‘independence’ criterion. As such, elements will not be ‘independent’ where they are ‘related to each other in content [or] merged with each other in a single creation process’.⁹⁰

The CJEU’s emphasis on ‘autonomous informative value’ can be likened to Bygrave’s concept of ‘semantic’ independence,⁹¹ and Aplin’s concept of ‘conceptual’ independence.⁹² This can be tested by enquiring whether—following Leistner’s proposal—the ‘collected elements […] hold the same self-contained informational content when accessed separately as each would have when viewed together’.⁹³ Leistner cites a telephone directory as an example: each entry contains data that is ‘self-contained’, such that separating the individual entries would not render them meaningless.⁹⁴ This requires examining the place of each item within the collection, as opposed to looking at the item in isolation.⁹⁵ This is because the degree of independence of a piece of data may be contingent on the nature of the wider collection in which it finds itself, which is a fact-sensitive question. As Chalton explains, some stock exchange transactions may qualify as independent while some are interdependent, depending on ‘the nature and subject-matter of the series of transactions, the consideration given for each, the identity of the contracting parties and the inclusion of an accumulating accounting function linking and transactions together’.⁹⁶

In practice, individual items in an index can be taken out of the whole without losing their meaning, which cannot be achieved for frames within a movie.⁹⁷ As Davison explains, a film cannot be seen as a database as each frame does not have a ‘stand-alone function’, but form with other such frames an integral part of the film.⁹⁸ Video games suffer the same fate. Aplin explains that the graphical representation of the protagonist would have a different meaning outside the game as a single graphic, compared to its status as an enhanced component inside the game.⁹⁹

Vice versa, an electronic encyclopaedia does satisfy the ‘independence’ criterion, as the articles have the same meaning inside and outside the compilation.¹⁰⁰ Similarly, Beutler is correct in arguing that a collection comprising a piece by Mozart, a text about him and a picture of him would be a collection of independent materials, since they can be separately commercialized (even if they share a common theme).¹⁰¹ Other instances are less clear-cut, such as data inside a DNA sequence. On that matter, Bygrave agrees with other scholars’ commentary that the more complete the sequence is, the less independent it is (as the more of a ‘story’ the sequence tells).¹⁰² This example is significant when assessing how a string of data recorded inside different blocks in a single blockchain may affect the overall independence of the materials.

4.2. Why blockchain fails the ‘independence’ test

Despite the conclusion reached in the Law Society Guidance that ‘a chain created in [distributed ledger technology (“DLT”)] is likely to fit within the definition of a database’, a strict application of the test for what constitutes ‘independent’ data would arguably lead to a contrary outcome. The Law Society Guidance explains that the quality of an electronic coin as a ‘chain of digital signatures…would likely constitute a collection of data or other materials if nothing else’.¹⁰³ Nevertheless, the Guidance does not consider whether such a ‘chain of signatures’ constitutes a collection of independent data—despite the fact that a ‘chain’ denotes some level of dependence in the linkage of the chain. Yet, there is a risk that a blockchain fails to meet the independence criterion at two levels: the individual block and the wider chain.

4.2.1. Individual block

The wide meaning of ‘database’ can accommodate a single block in the chain just as it can include a chain of blocks.¹⁰⁴ The most evident example of self-contained data inside a block is ‘unhashed’ raw transactional data, the extraction of which is unlikely to modify its autonomous informative value for interested users, even where pseudonyms are used.¹⁰⁵ This could lead to the conclusion that blockchain qualifies as a ‘database’.¹⁰⁶ However, it is submitted that such a conclusion does not follow where the current unrefined test for ‘independence’ does not distinguish between the different types of data inside a block.

Although transactional data may show independence, the rest of the data inside that same block—which Pech ignores—fails that test. The combinational hashing up until the root of the Merkle tree is an example where the contents are merged ‘in a single creation process’ in rendering data in the block immutable. Similarly, applying Beutler’s analysis, while individual transactional data may be taken out and ‘commercialised’ individually, the combinational hashing and the data in the block header cannot be extracted without harming the collection. The puzzle in the PoW process is the most blatant example of the interdependent data, the solving of which depends on a combination of data inside that block. Data dependent on other data ‘from which it is calculated’ is an example that Derclaye provides as clearly failing the ‘independence’ criterion.¹⁰⁷ As such, the mix of independent and interdependent data creates uncertainty.¹⁰⁸

4.2.2. Chain of blocks

The independence of data throughout the chain is compromised by the nature of blockchain as a linked list of blocks. The hash reference to each previous block would constitute a ‘merger’ of the contents of the database in a ‘single creation process’ of the chain per Leistner’s definition. Following Bygrave’s analysis of a DNA sequence, the longer the chain is, the more of a ‘story’ it tells, the less independent it would be.

The above analysis echoes the judgment in Software Solutions Ltd v 365 Health and Wellbeing Ltd,¹⁰⁹ where the High Court of England and Wales held that an XML Schema (‘data formats’ part of a wider system) failed the ‘independence’ criterion as the individual components of the Schema were akin to individual words in a literary work (and thus inseparable without losing their informative value).¹¹⁰ Interestingly, the Schema was used ‘to provide a structure or format to validate and verify the contents’ of the other applications.¹¹¹ By analogy, interdependent data in the block header and the Merkle tree ensure the validation of data and the functioning of the blockchain as a whole. Considering this interdependent data is equally stored in the block as is independent transactional data, unclarity as to what type of data the ‘independence’ criterion should be applied to could impact the protection of the wider blockchain as a database.

This is a highly unsatisfactory situation since valuable databases could be disqualified on the basis that only some of their contents are interdependent. For instance, the Goldman Sachs KYC/AML Blockchain Application comprises quite a few elements that interact with each other to complete the KYC process, some of these being ‘globally unique identifiers (GUIDs)’, a VC per transaction, the transactional data itself,¹¹² and the encrypted identifying information relating to the unique identifiers that are discoverable by the regulatory agency.¹¹³ It could be argued that the vetting of one transaction is independent from another such vetting, or that the body of regulation the third party relies on is independent from the numerical amount exchanged between two parties.

However, large amounts of data in the verification process are interdependent: the GUIDs are linked to off-chain identifiers, VCs are unique to the specific transaction verified, and the vetting of transactions is dependent on public and private information gathered by the third party such as ‘entity jurisdiction, physical address [and] beneficial ownership…’.¹¹⁴ The transactions may be processed as a smart contract, the execution of which is conditional on the finalization of the vetting process.¹¹⁵ The architecture of the blockchain underpinning this application also relies on interdependent data. As the purpose of this invention is to ensure KYC/AML due diligence in a decentralized environment where no human oversight is possible (eg in high-frequency trading situations), the reliability of the recording mechanism is key for the legitimacy of the invention as a whole.¹¹⁶ The more interconnected the hash referencing is the more reliable the transactions would be. For this reason, the database could very well slip through the cracks of the database regime, not because such a result was intended by the legislators behind the Database Directive but rather as a result of the uncertainties surrounding the concept of ‘independence’. The following section will attempt to remedy these uncertainties.

5. Proposal for a New Test for ‘Independence’

The purpose of this proposal is to reframe the content and the application of the ‘independence’ criterion to ensure novel forms of database like blockchain containing both interdependent and independent data are adequately protected while adhering to the spirit of the Database Directive. This proposal is one possible way the law could modernize in adapting to unanticipated innovation while providing more clarity to the concept of ‘independence’.

Although it could be argued that the ‘independence’ criterion is a limiting notion that is meant to exclude subject matter,¹¹⁷ denying protection to all databases that may contain only some interdependent data would defeat the purpose of the Database Directive in developing a competitive information market. The vagueness of the notion of ‘independence’ fails to discriminate between the different purposes distinct sets of data within a single piece of technology may serve. These sets of data can be broadly split into two categories: data that form the ‘contents’ of the database and data that forms its ‘structure’. This proposal argues that the ‘independence’ criterion should be confined to the former.

5.1. The separation of data into ‘content’ and ‘structure’

This part discusses the justifications as to why this separation is warranted, emphasizing the wide array of databases that could be protected under the Database Directive. It also defines the respective characteristics of data in the ‘content’ and ‘structure’ categories, following the ‘referential’ nature and the ‘intrinsic’ purpose of such data. This ‘content/structure’ divide is then mapped onto blockchain technology, revealing that it is blockchain’s structural data that has non-referential characteristics with an intrinsic purpose in ensuring the functioning of the database.

5.1.1. Justifications for the separation

Current laws need to adapt to the exponential evolution of technology. The nature and complexity of databases has evolved considerably since the passing of the Database Directive. The technical features of a database should not be a barrier for it to be protectable subject matter. This accords with the Court’s holding in OPAP that the term ‘database’ should have a ‘wide scope…unencumbered by considerations of a formal, technical or material nature’.¹¹⁸ Such a tailoring of the law does not contradict the spirit of the Database Directive. Neither Article 1(2) of the Database Directive nor section 3A(1) specify to what type of data the ‘independence’ criterion is to be attached. Considering that data must also satisfy the ‘systematicity’ criterion, it is very likely that a part of the data contained in the database will be interdependent to ensure that systematicity. This type of data concerns more on the ‘structure’ of the database as opposed to its contents. Reading the ‘independence’ and ‘systematicity’ criteria together, it is more likely for the ‘independence’ criterion to be associated with ‘contents’ rather than ‘structure’.

As a solution to the uncertainties surrounding the ‘independence’ criterion, Derclaye had proposed that protection would be granted where more than 50 per cent of data are independent.¹¹⁹ Although an improvement to the status quo, this solution remains unsatisfactory as it is conceivable for an elaborate database to contain more interdependent ‘structural’ data than ‘content’. The implementation of a 50 per cent threshold should be confined to ‘contents’. Otherwise, if a database has a 50/50 divide between ‘content’ and ‘structure’, but 99 per cent of the ‘contents’ are independent, the majority of the data would be interdependent and the subject matter disqualified. This runs counter to the purpose of the Database Directive in promoting growth in the information market where such ‘structural’ data could very well contribute to the making of a powerful database.

5.1.2. Earmarking the two types of data

The simplest way of distinguishing between ‘contents’ and ‘structural’ data is associating them with the two basic elements of a database—‘data’ and ‘format’, respectively.¹²⁰ In adopting a purposive approach, Leistner explains that the difference between a database in the ‘technical’ sense (ie data regrouped in a single product) and a database in the ‘legal’ sense (conforming with the legal definition) is that the latter have data with a ‘referential’ character, ie the ability to communicate ‘selfcontained’ items of information.¹²¹ As such, entries in an electronic encyclopaedia have such referential character, whereas individual elements of a video game like sound and image are non-referential elements combined to ‘manifest’ a single product that do not go beyond the design of the product.¹²²

Leistner’s referential/non-referential divide can be likened to the CJEU’s emphasis in OPAP on ‘autonomous informative value’:¹²³ individual elements in a video game lose their autonomy in terms of the information they convey when extracted from the collection as they lack referential character. The CJEU specifies that data can also have a ‘literary, artistic, musical or other value’.¹²⁴ The expression ‘other value’ could include operational value characteristic of non-referential data. This is, however, unlikely as ‘other value’ follows on a list of specific terms that refer to value in self-contained items of information rather than within an operational and functional whole.¹²⁵ Interdependent non-referential elements would also lose their autonomous value if extracted—regardless of their type of ‘value’.

The usefulness of the referential/non-referential divide goes beyond distinguishing ‘contents’ from ‘structural’ data and may also help courts in remedying the blurred boundary between database and underlying software.¹²⁶ Davison is critical of the decision in Mars UK Ltd v Teknowledge Ltd (‘Mars’), where the Court held that the reproduction of data in an electronically erasable programmable read only memory (EEPROMs) relating to coins to differentiate between different values in a vending machine was an infringement of the right-holder’s database right.¹²⁷ He explains that this decision conflates data collected for the ‘extrinsic purpose of informing someone’ and data instrumental to the machine’s functioning.¹²⁸ He notes a similar amalgamation was made in Data Access Corporation v Powerflex Services Pty Ltd (‘Powerflex’),¹²⁹ where the Australian High Court held that a Huffman compression table was protected by copyright. Davison argues this is proof of how the Database Directive fails to separate the protection of computer programs from databases despite its intention to do so.¹³⁰

Davison’s criticism offers further clarity on the proposed division between ‘content’ and ‘structure’. While the ‘contents’ of a database have the extrinsic purpose of informing users, the ‘structural’ data have intrinsic value in ensuring the performance of the underlying technology. ‘[Extrinsic] purpose’ in Davison’s conceptualisation of the database refers to a ‘purpose unrelated to the operation’ of the underlying product (eg vending machines in Mars), which can be determined by how ‘userfriendly’ the data arrangement and accessibility is.¹³¹

In a public blockchain, a quick search does allow one to find non-referential data such as the nonce or the difficulty level of the puzzle.¹³² Yet, being able to access this type of data does not necessarily mean that it is ‘user-friendly’ as, in practice, users are more interested in seeing their transactions being recorded than what an intermediate hash in the Merkle tree reads. The ‘user-friendliness’ of this non-referential data—as is the case for the coin EEPROMs in Mars and the Huffman compression table in Powerflex—is limited to ensuring the functioning of the underlying technology.

5.1.3. Application to blockchain

The category of ‘non-referential’ data corresponds to the ‘structural’ data in the blockchain, while ‘referential’ data refer to its contents. The findings below are merged in Table 1 in  Annex 1. All the elements that fail the ‘independence’ test, such as the hash references and the nonce, are all interrelated data that do not have ‘referential’ value in communicating ‘self-contained individual items of information’, but rather serve the purpose of creating a single product (ie an immutable and cryptographically secure distributed ledger). The only ‘referential’ value the hash references have is limited to their technical feature in referencing the previous block to form a chain. Only the ‘contents’ of the blockchain (ie the raw transactional data) have informative value that transcends the product’s design.

The CJEU in Verlag Esterbauer specified that autonomous informative value must be ‘not for a typical user of the collection concerned, but for each third party interested by the extracted material’.¹³³ Following this, it would be wrong to limit the idea of ‘referential’ value to a simple category of users. This is unproblematic for the ‘contents’ of the blockchain, given that such referential data have value for all users of a blockchain. Ordinarily, the informative value of ‘structural’ data inside a blockchain would only have value vis-à-vis an ordinary user insofar as it ensures the immutability of the information contained in the transactional data. However, ‘non-referential’ data in a blockchain could, arguably, have informative value vis-à-vis miners that are undertaking PoW or any user wishing to inspect the operation of the blockchain. This is well illustrated by Drescher’s comparison of a chain of blocks to a book: the contents of a book correspond to transactional data, the content page to the Merkle tree containing the transactional data and a reference to the content page to the root of the Merkle tree.¹³⁴ Although most readers will only be interested in the contents of the book, an editor may very well find the content page informative. This is consistent with the idea that a database can have a multiplicity of purposes just as it can have a multiplicity of categories of users.¹³⁵

Nevertheless, this does not change the fact that such ‘structural’ data have ‘informative’ value only insofar as they are placed in the context of the collection from which it was extracted. By way of example, extracting the hash reference (herein referred to as ‘HRef’) of Block 100 contained in Block 101 will only yield a string of numbers and letters.¹³⁶ HRef 100 will only convey the information that Blocks 100 and 101 are linked—which is what the HRefs are functionally valuable for—when it is replaced in the context of these two blocks. At a very high level, the validity of HRef 100 would inform users of the secure functioning of the blockchain and its underlying software. Reaching this conclusion still requires comparing HRef 100 to the components of Block 100. As such, modifying the category ‘typical user’ in this context does not change the interdependent nature of ‘structural’ data.

The fact that a simple ‘decline in the informative value’ is allowed before data completely loses its independent does not change the interdependent nature of ‘structural’ data either.¹³⁷ The CJEU in Verlag Esterbauer clarified that although the value of data may be increased by being arranged within a collection, a decline in such value by isolating that data does not lead to the conclusion that such data cannot be independent from other data that happens to enhance its value within the collection.¹³⁸ This is only applicable to ‘referential’ data.

For instance, a transaction in Block 101 taken in isolation may convey the simple information that A paid B 1 BTC. Reading these data in conjunction with Block 100 may present a different narrative, for instance, that A is repaying B for a mistaken transfer of 1 BTC. Taking the Block 101 transaction in isolation may have lesser informative value in conveying what is, in fact, a rectification of a mistaken transfer, but the transaction in each block can be extracted without harming the rest of the collection. By contrast, taking HRef 100 and HRef 101 leads to the HRefs losing their meaning vis-à-vis their functional role within their respective block headers, but also also vis-à-vis the rest of the chain. This represents a total loss of informative value and not a ‘simple decline’ thereof.

There will be greater loss of informative value where data need to be read linearly (as is for works) as opposed to retaining informative value irrespective of its placement within the collection (as it should be in databases).¹³⁹ While the transactions in Blocks 100, 101 and 102 have greater value when read together as a chronology of events, the transactional data in each block retains its informative value even if a user reads Block 102 first and Block 101 last. By contrast, HRefs 100, 101 and 102 only make sense when read linearly.

Gils comes to the same conclusion that a blockchain can satisfy the ‘independence’ test, but applies it to the blocks themselves, as opposed to the individual sets of data inside the blocks: transactional data has autonomous informative value when the relevant block is ‘cut loose’ from the chain.¹⁴⁰ This approach conflates the different types of data contained inside a single block. Concluding the cut-off block itself retains autonomous informative value because it contains raw transactional data pays no heed to the rest of the data that forms the block and enables the prima facie recording of that transactional data. It also does not specify what proportion of data (if any) needs to be independent within that block for the test to be satisfied. Separating data following the ‘content/structure’ divide remedies such conflations both when viewing the block and the chain as a database.

Gils also argues that Chalton’s view of the ‘independence’ test is an alternative to that posited by the CJEU in Verlag Esterbauer: the former would require us to assess the ‘mutual (in)dependency of the blocks’ while the latter would focus on ‘the relation between the elements and the database’.¹⁴¹ With respect, the two tests can be applied complementarily. If the blocks form the database and are ‘materials’ of the database (as Gils posits), assessing their independent quality involves as much looking at its mutual independence from other blocks as its independence from the database as a whole. If a block has autonomous informative value vis-à-vis other blocks (such that the ‘cut off’ block and the rest of the blocks maintain their autonomous informative value), then a fortiori the ‘cut off’ block retains its informative value vis-à-vis the database itself. In any case, the conclusion that ‘structural’ data is lacks independence does not change whether the CJEU’s test or Chalton’s version is applied: the structural data will be mutually dependent on both the contents of the block and the chain in between blocks, such that it will never have an autonomous informative value.

Analogizing the PoW puzzle to video games illustrates how the ‘content/structure’ divide can be mapped onto blockchain. Manifesting an occurrence in a video game requires the combination of several non-referential elements—eg representing a simple jump could involve different graphics, sound effects and vibrations in the joystick. By analogy, data inside a block header are all nonreferential elements that combine to manifest the PoW puzzle. The elements in the video game and in the block header are equally involved in a ‘single process of creation’, whether it is a jump or a puzzle. The main difference between video games and blockchain is that, although non-referential elements in video games could qualify as ‘contents’, the fact that most of the data are interrelated means that video games would almost always fail the ‘independence’ test, irrespective of how their data are categorized.

Applying Davison’s notion of ‘extrinsic purpose’ to blockchain, ‘contents’ have the extrinsic purpose of informing the user of a verified state of affairs, while ‘structural’ data serves the intrinsic purpose of ensuring the functioning of the blockchain as an immutable ledger. This can be compared to the data/metadata divide in an email. Data in the block header (conveniently termed by some as the block ‘metadata’) are necessary for the functioning of the blockchain as are metadata for the functioning of the email transfer. By contrast, the informative value for the users of the email service usually lies in the contents of the email, as is the case for transactional data in the block. The ‘contents’ are usually the purpose for which the blockchain was created, whereas the ‘structure’ are data ancillary to that purpose but necessary for its functioning as an immutable ledger.

The self-contained nature of the environment in which cryptocurrencies are simultaneously created and exchanged could call into question the independence of raw transactional data. The amount of Bitcoin in a given wallet will depend on the aggregate of the values in past transactions over several blocks, since the Bitcoin blockchain operates as a ledger as opposed to a balance sheet. Although a valid concern, excluding such data would impose an unworkably high threshold for ‘independence’. All data have a starting point—a transfer of money into an empty bank account depends on the existence of another such account. The focus of the enquiry should be on whether data carry a discrete informative value. This is supported by Chalton’s comment that a collection of short stories where the characters are recurring, but the plots are self-contained, would not be excluded.¹⁴² By analogy, the recurrence of a same account or a same coin in different transactions would not offend the ‘independence’ criterion.

Gils argues that a chain of blocks recording cryptocurrency transactions cannot be ‘meaningfully taken apart’ without affecting the ‘value-bearing capacity of the security’ of the blocks being together.¹⁴³ With respect, this reasoning confuses the autonomous informative value of data as opposed to its safe recording in the blockchain. Informative value could decline if data are not properly recorded, but such a decline does not signify a total loss of independence.¹⁴⁴ The safe recording of data in a blockchain would mostly concern the strength of the accessibility and the systematicity of the data under Article 1(2) of the Database Directive and section 3A(1) of the CDPA 1988. A mutable piece of data can be independent and autonomous from others in the information it communicates without being secure.

Otherwise, all databases outside of secure blockchains would fail the test.

5.2. A new three-step test for ‘independence’

The proposed test for assessing whether a database is a collection of ‘independent materials’ involves three steps, following the categorization of data as either ‘content’ or ‘structure’. The test is illustrated through Fig. 1 in  Annex 2.

Step 1 asks whether some of the data contained in the database are interdependent. If the totality of the data is interdependent, the lack of any independence would automatically disqualify such a subject matter. If the totality of the data is independent, the subject matter may qualify as a database if the rest of Article 1(2) is satisfied. If only some of the data are interdependent, the test proceeds to Step 2, which involves examining whether the ‘contents’ of the database are independent and which is contingent on preliminarily separating the ‘contents’ from the ‘structural’ data following Table 1. If the interdependent data corresponds to ‘structural’ data and the ‘contents’ satisfy the ‘independence’ criterion, Step 3 then requires assessing whether the rest of the definition is met, namely whether there is a ‘collection…of works, data and other material’. Satisfying this leads to assessing as a last stage whether the rest of Article 1(2) is satisfied.

In the event that contents are also interdependent, two options arise under Step 2. Following Option 1, it would only be necessary for the majority of the ‘contents’ to be independent. This allows one to distinguish a table of contents from a bibliography or an index: the former fails the independence test as it contains a ‘sequential flow of numbers’ such that taking an element out would harm the table whereas the contents of the latter have self-contained meaning.¹⁴⁵ Alternatively, Option 2 would require the entirety of the contents to be ‘independent’. Importantly, Step 2 assesses the proportion of interdependent ‘contents’ against the ‘contents’ themselves, as opposed to the entirety of the data (as the proportion of dependence of ‘structural’ data should not be taken into account here). Choosing what option to follow depends on what flexibility courts are willing to accept, but this debate is outside the scope of this article as it does not concern the ‘content/structure’ divide.

5.2.1. Application of the proposal to existing examples

Questions about rendering the concept of ‘independence’ over-inclusive are promptly resolved when assessing whether subject matter traditionally excluded remains so under the new test. There are two instances of this. First, cinematographic works are explicitly excluded as databases in Recital 17 of the Database Directive. Following the proposal, a survey of the data contained within a movie quickly leads to the conclusion that the individual frames of the movie are interdependent elements and the movie as a whole fails at Step 1. The same reasoning applies to ‘audio-visual’ works in Recital 17. Second, video games are made up of non-referential elements that are combined to create a single product.¹⁴⁶ These non-referential elements are all interdependent, and, where no other data is present to proceed to Step 2, video games fail at Step 1.

5.2.2. Application of the proposal to blockchain

This part applies the proposal to blockchain pursuant to the separation of its data into ‘content’ and ‘structure’ as illustrated in Table 1. The fact that a single block contains both interdependent ‘structural’ data and possibly independent ‘content’ leads to Step 2. Whether the ‘contents’ are fully independent is fact-sensitive and application-specific. A chain of independent transactions between Bitcoin wallets is likely to be composed of independent data, whereas a different blockchain application recording a storyline over multiple blocks will be less so. What is clearer is that the data inside and in between blocks satisfy the ‘data’ or ‘other material’ criteria in the definition of ‘database’.¹⁴⁷ A similar outcome is obtained when applying the proposal to a chain of blocks as a database. The main difference is that the category of ‘structural’ data would also include hash references in between blocks as opposed to a single block header.

This proposal can also be mapped onto the Goldman Sachs KYC/AML Blockchain Application. As the proposal is a blockchain-based application using smart contracts, it necessarily follows that both a single block and a chain of blocks contain both ‘structural’ and ‘content’ data. The ‘content’ category would be populated by all data that are not related to the architecture of the platform on which the KYC/AML vetting is performed. ‘Contents’ would include GUIDs, private and public data used by the third party to vet transactions, data retained by the third-party post-vetting,¹⁴⁸ transaction details, data inside the smart contract (eg conditions precedent to transaction execution or the nature of the assets traded),¹⁴⁹ VCs and the identity of the third party. The remainder of the data allowing for the recording of the vetted identities and transactions are structural data that can be forgiven for being interdependent, as its sole intrinsic purpose is to ensure the functioning of the decentralized KYC/AML screening process.

The application of Step 2 is broadly uncontroversial, with the exception of two pieces of data. GUIDs and VCs are arguably data dependent on other materials inside the application. GUIDs are identifiers dependent on the real identity of the corresponding party off-chain and, if extracted, may lose their ‘autonomous informative value’ outside the immediate blockchain application. Although this argument has force, it ignores the fact that GUIDs retain their significance outside the collection for the vetting third party. The latter—which may also be an off-chain entity—reconciles GUIDs with independently collected data on the relevant transacting party and may use GUIDs in issuing regulatory or tax documentation when identities need reporting.¹⁵⁰ As such, GUIDs maintain their informative value whether they are used in smart contracts to proof transactions or outside the blockchain ecosystem.

The fact that GUIDs can be issued as ‘Decentralized Identifiers (DIDs)’ or ‘Non-Fungible Tokens (NFTs)’ ensures their transportability beyond the immediate application—for instance, they can figure on a public ID register.¹⁵¹ The DID or NFT may itself include a public identifier that can be decrypted by a regulatory agency outside of the blockchain system without resorting to the third party as an intermediary.¹⁵² It is irrelevant that DIDs or NFTs only harbour informative value vis-à-vis the vetting third party, since defining a specific audience for which the data retain an ‘autonomous informative value’ is not necessary.¹⁵³

A similar conclusion can be reached for VCs. VCs and transactional data are arguably interdependent, as a single VC is issued per transaction. Just as the contents of a block header are dependent on the contents of the block, the contents of a VC are dependent on the contents of the transaction (eg nature of asset, value of asset, GUIDs of parties, jurisdiction, etc). Despite this, VCs are also self-contained documents. Inside of the blockchain ecosystem, a VC enables parties to transact pseudonymously while guaranteeing KYC/AML vetting. Outside of the ecosystem, a VC works as a stamp of approval that can be demonstrated as proof of KYC/AML compliance to a regulatory agency (especially, since the transaction itself is recorded publicly upon completion).¹⁵⁴ The provision of VCs can be assimilated to credit scoring, as repeated transactions may affect a transacting party’s KYC/AML ‘scope of compliance for future transactions of the same type’.¹⁵⁵ Hence, VCs also retain their ‘autonomous informative value’ within a transaction and independently outside of the blockchain. As for Step 3, the GUIDs, VCs and all other data inside the application will satisfy the ‘collection of…works, data and other materials’ limb.

6. The Place of the ‘Content’/’Structure’ Divide in the Broader Regime

6.1. Systematicity and accessibility

Criterion (2) requires that data be ‘arranged in a systematic or methodical way’.¹⁵⁶ In a blockchain, the methodical arrangement of data is mostly ensured by the Merkle tree and the hash references between blocks,¹⁵⁷ which are ‘structural’ data. An analogy would be the ‘hierarchical’ organization of ‘non-linear’ individual pages of a website interconnected by hyperlinks: while the individual pages constitute the ‘constituent’ elements of a website (ie the ‘contents’), the hyperlinks allow for the systematic arrangement of the latter (ie ‘structural’ data).¹⁵⁸ The hyperlinks, however, would not invalidate the collection of web pages from qualifying as a database.

Criterion (3), alongside Recital 13 of the Database Directive, provides such collections that must be accessible by ‘electronic or other means’.¹⁵⁹ Such access is ensured by the hash references that enables the efficient location of data. Drescher compares these to cloakroom tickets pointing towards the location of coats.¹⁶⁰ It is also possible to employ a computer software in the form of a query language.¹⁶¹ Overall, data on a blockchain are easily accessible and displayable,¹⁶² due to its distributed architecture. The level of accessibility may depend on whether the blockchain is a public or a private one,¹⁶³ but this does not change the overall accessibility of the data once recorded to authorised nodes. These findings are summarized in Table 2 in  Annex 1.

6.2. The relevance of the distinction for the subsistence of the rights

The ‘content’/‘structure’ divide also fits well within the subsistence tests.

Copyright protects the ‘structure’ of the database irrespective of ‘contents’.¹⁶⁴ Article 3(1) of the Database Directive requires that the author’s own intellectual creation stems from the ‘selection or arrangement of their contents’.¹⁶⁵ Thus, the originality is expected to lie in the ‘structure’ as opposed to the ‘contents’. By contrast, the SGR protects the ‘contents’ from unauthorized utilization,¹⁶⁶ and investment lies in the ‘structural’ data performing these activities, irrespective of the ‘contents’ being protected.¹⁶⁷ This accords with Wei’s conception of a ‘value-added’ database with elements additional to raw data being protected by copyright and a ‘primary’ database only compiling raw data being covered by the SGR.¹⁶⁸ In a blockchain-based database, it may be harder for the structural elements to satisfy the originality requirement while the process of recording of transactions could be sufficient ‘investment’ for the SGR.¹⁶⁹

The proposal may result in ‘structural’ data lacking protection. Copyright covers all materials inside a database, as copyright infringement does not distinguish between ‘content’ and ‘structure’.¹⁷⁰ By contrast, the SGR protects the ‘contents’ from unauthorised ‘extraction and/or re-utilisation’.¹⁷¹ Thus, unoriginal ‘structural’ attributes of a database may go unprotected under both regimes. This is unproblematic for the main reason that, from a practical point of view, the unoriginal ‘structural’ data (eg hash references) is unlikely to be of interest for copying purposes. The real value of information stems from, inter alia, its immutability among nodes.

7. Conclusion

Blockchain technology is a valuable tool for the development of a reliable and controlled information market. It enables us to transact safely with potentially untrustworthy peers and to store data in a secure and tamper-proof manner. It is crucial both to promote and to protect investment in such technologies, which can be most safely achieved through the grant of IP rights. The Database Directive offers blockchain makers as an accessible and convenient way of safeguarding their investment.

However, the ‘independence’ criterion in Article 1(2) of the Database Directive and Section 3A(1) of the CDPA 1988 constitutes an initial barrier to protection. Unlike more traditional databases, blockchain’s immutability depends on highly interconnected data contained within each block and in the wider chain. This barrier results from the impreciseness of ‘independent’ as a concept, which fails to distinguish between the different types of data that serve differing purposes inside the blockchain. While the ‘contents’ of a database have the extrinsic purpose of providing the user with self-contained information, the ‘structure’, which also ensures the ‘systematicity’ and the ‘accessibility’ of data, serves the intrinsic purpose of enabling the functioning of the database. This article has proposed a new three-step test that adheres to the spirit of the Database Directive and accommodates blockchain-enabled databases by confining the application of the ‘independence’ criterion to ‘contents’.

Table 1.

Table on the characteristics of ‘content’ and ‘structure’ and its application to a PoW blockchain.^a

	Characteristics	Application to a PoW blockchain
Contents	Informative value for the user Extrinsic purpose independent from the functioning of the underlying technology Referential characteristic (communicating self-contained information)	• Transactional datab
Structural data	‘Structural’ value for the makers of the database Intrinsic purpose in ensuring the functioning of the technology Non-referential characteristic (data used in combination of other material to manifest a single product)	Block header (nonce, difficulty level, software level, timestamp, root of Merkle tree, hash reference of previous block) Merkle tree

a

This table summarizes how the proposed ‘content/structure’ divide can be applied to the component parts of a blockchain, as examined in Section 4.1.3.

b

This should satisfy the term ‘collection’ as no threshold is imposed: OPAP (n 83) para 24; OPAP Opinion (n 84) para 37.

Table 2.

Table associating the elements of blockchain with the criteria in Article 1(2) of the Database Directive and Section 3A(1) of the CDPA 1988.^a

	Individual block	Blockchain
(1) Independence	Individual ‘contents’ of a block (transactional data…)	Blocks and individual ‘contents’
(2) Systematicity	Merkle tree	Hash reference links
(3) Accessibility	Access to individual ‘contents’	Transparent and distributed architecture

a

This table outlines what component parts of a blockchain can satisfy the three limbs of Article 1(2) of the Database Directive and section 3A(1) of the CDPA 1988, as analysed in Section 5.1.

Figure 1.

Proposed three-step test to decide whether subject matter qualifies as a database when the database involves both independent and interdependent data.

Footnotes

1

See Illinois Blockchain and Distributed Ledger Task Force, ‘House Joint Resolution 25 – Final Report to the General Assembly’ (31 January 2018) Available at <www2.illinois.gov/sites/doit/Strategy/Documents/BlockchainTaskForceFinalReport020518.pdf"> (accessed 21 February 2024).

2

Reuters, ‘Intel Launches Blockchain Chip to Tap on Crypto Boom’, Reuters (Bengaluru, 11 February 2022). Available at <www.reuters.com/technology/intel-launches-blockchain-chip-tap-crypto-boom-2022-02-11/> (accessed 21 February 2024).

3

Microsoft Azure offers the possibility of creating one’s own blockchain without purchasing the underlying infrastructure (ConsenSys Quorum Blockchain Services): Microsoft Azure,

‘Blockchain’ (nd), Available at <https://azure.microsoft.com/en-us/solutions/blockchain/#overview> (accessed 21 February 2024).

4

A Rose, ‘Blockchain: Transforming the Registration of IP Rights and Strengthening the Protection of Unregistered IP Rights’, WIPO Magazine (London, July 2020). Available at <www.wipo.int/wipo_magazine_digital/en/2020/article_0002.html> (accessed 21 February 2024).

5

PÇ Aksoy and ZÖ Üner, ‘NFTs and copyright: challenges and opportunities’ (2021) 16 The Journal of Intellectual Property Law & Practice 1115, 1120–26.

6

[1996] OJ L77/20.

7

[2021] EWHC 626, Ch [186].

8

Database Directive, Recital 57.

9

Database Regulation, regs 6 and 12(1).

10

Database Directive, art 1(1).

11

ibid art 1(2); CDPA 1988, s 3A(1).

12

D Drescher, Blockchain Basics: A Non-Technical Introduction in 25 Steps (New York City, Apress 2017) 30–31. See further S Nakamoto, ‘Bitcoin: A Peer-to-Peer Electronic Cash System’ (2009). Available at <www.ussc.gov/sites/default/files/pdf/training/annual-national-training-seminar/2018/Emerging_Tech_Bitcoin_Crypto.pdf> (accessed 21 February 2024).

13

The explanation on how the puzzle functions is sourced from: Drescher (n 12) 138–40.

14

ibid 119.

15

A Narayanan et al, Bitcoin and Cryptocurrency Technologies: a Comprehensive Introduction (Princeton, Princeton University Press 2016) 2–3.

16

ibid 12–13.

17

A Haleem et al, ‘Blockchain Technology Applications in Healthcare: An Overview’ (2021) 2 IJIN 130.

18

Drescher (n 12) 139–40.

19

M Orcutt, ‘How Secure is Blockchain Really?’ (MIT Technology Review, 25 April 2018). Available at <www.technologyreview.com/2018/04/25/143246/how-secure-is-blockchain-really/> (accessed 21 February 2024).

20

Database Directive, Recitals 9–10 and 12; Case C-46/02 Fixtures Marketing Ltd v Oy Veikkaus Ab [2004] ECLI:EU:C:2004:694, para 33. Granting database protection post-Brexit would still provide the UK with a competitive advantage in protecting UK-based database makers from exploitation abroad: J Lederberg, ‘Database Protection in the European Union and the United States: The European Database Directive as an Optimum Global Model?’ (2001) 1 Inverse Piecewise Quadratic 100, 125–26.

21

S Beutler, ‘The Protection of Multimedia Products through the European Community’s Directive on the Legal Protection of Databases’ (1996) 7 Entertainment Law Review 317, 319.

22

GM Dutfield and U Suthersanen, ‘The Innovation Dilemma; Intellectual Property and the Historical Legacy of Cumulative Creativity’ (2004) 4 Inverse Piecewise Quadratic 379, 411.

23

The EU had announced their plan to invest further in blockchain technology in 2021: Aoife White, ‘EU Targets Blockchain, Data in “Make or Break” Tech Investment’, Bloomberg (London, 15 September 2021). Available at <www.bloomberg.com/news/articles/2021-09-15/eu-targets-blockchain-data-in-make-or-break-tech-investment> (accessed 21 February 2024).

24

M Finck, Blockchain Regulation and Governance in Europe (Cambridge, CUP 2018) 6. See also M Campbell-Verduyn, ‘Introduction: What are Blockchains and how are they Relevant to Governance in the Global Political Economy?’ in M Campbell-Verduyn (ed), Bitcoin and Beyond: Cryptocurrencies, Blockchains and Global Governance (Routledge 2018) 1: ‘…this digital database grants you nearly real-time overviews of peer-to-peer exchange…’ (emphasis added).

25

Deloitte, ‘Blockchain Use Cases’ (Deloitte’s 2021 Global Blockchain Survey, 2021) 18 Available at <www2.deloitte.com/content/dam/insights/articles/US144337_Blockchain-survey/DI_Blockchainsurvey.pdf> (accessed 21 February 2024).

26

E Tzoulia, ‘The Blockchain Ecosystem in the Light of Intellectual Property Law’ (2022) 13 Journal of Intellectual Property, Information Technology and Electronic Commerce Law 290, 293.

27

See Team Blockdata, ‘Top 100 Public Companies Investing in Blockchain & Crypto Companies’, Blockdata (New York, 16 August 2022). Available at <www.blockdata.tech/blog/general/top-100-public-companiespubliccompanies-investing-in-blockchain-and-crypto-companies> (accessed 9 November 02023) 21 February 2024.

28

For instance, the EU Commission has recently awarded €5 M as a prize for projects submitted to the EIC Prize on Blockchains for Social Good initiative: Commission, ‘The Commission’s European Innovation Council Awards €5 Million to Blockchain Solutions for Social Innovations’, DIGIBYTE (Brussels, 30 June 2020). Available at <https://digital-strategy.ec.europa.eu/en/news/commissionseuropean-innovation-council-awards-eu5-million-blockchain-solutions-social-innovations> (accessed 21 February 2024).

29

Bitcoin, ‘Bitcoin is an Innovative Payment Network and a New Kind of Money’ (nd), Available at <https://bitcoin.org/en/> (accessed 21 February 2024).

30

R Burbidge et al, ‘Section 10: Intellectual Property’ (Blockchain: Legal & Regulatory Guidance 2nd edn, The Law Society 2022) (‘Law Society Guidance’) 166. Available at <www.lawsociety.org.uk/topics/research/blockchain-legal-and-regulatory-guidance-second-edition> (accessed 21 February 2024).

31

The issue concerning Satoshi Nakamoto’s real identity is the subject of separate proceedings, the hearing of which started in the UK High Court in early 2024: Michael Cross, ‘“Satoshi Identity” Hearing Draws a Worldwide Audience’, The Law Society Gazette (London, 5 February 2024). Available at <www.lawgazette.co.uk/news/satoshi-identity-hearing-draws-a-worldwide-audience/5 118 649.article> (accessed 21 February 2024).

32

[2023] EWHC 222, Ch; [2023] ECDR 9 [14], [21]. Mellor J also refused permission to serve the claim outside the jurisdiction on the basis that there was no serious issue to be tried as to whether ‘Bitcoin File Format’ in its XML form could be copyrighted. At [62], Mellor J explained that the Bitcoin File Format was not ‘expressed or fixed anywhere’ for it to be copyrightable. This was overturned on appeal, while the other matters were unchallenged: Wright v BTC Core [2023] EWCA Civ 868, CA. The question of fixation of the Bitcoin File Format is outside the scope of this article.

33

Database Directive, art 3(1); CDPA 1988, s 3A(2); Case C-604/10 Football Dataco Ltd v Yahoo! UK Ltd [2012] ECLI:EU:C:2012:115 (‘Football Dataco’), para 42.

34

S Pech, ‘Who Owns the Blockchain? How Copyright Law Allows Rights Holders to Control Blockchains’ (2021) 16 Journal of Business & Technology Law 59, 71.

35

Football Dataco (n 33) paras 38–39.

36

Pech (n 34) 71; T Gils, ‘Blockchain and Law: An Analysis of Blockchain Technology under the Database Directive 96/9’ (Master’s thesis, Brussels Faculty of Law 2016–2017), 31–32. Available at <https:// papers.ssrn.com/sol3/papers.cfm? abstract_id = 3 737 360> (accessed 21 February 2024).

37

Database Directive, art 7(1); Database Regulation, reg 13(1).

38

See Case C-203/02 The British Horseracing Board v William Hill Organisation [2004] ECLI:EU:C:2004:695 (‘William Hill’), para 38.

39

Pech (n 34) 73; Law Society Guidance (n 30) 166.

40

Database Directive, art 4(1); CDPA 1988, s 9(1).

41

ibid art 4(2).

42

ibid art 4(2); CDPA 1988, s 10(1).

43

ibid Recital 41; Database Regulation, reg 14(1).

44

Database Regulation, reg 14(5).

45

See Pech (n 34) 76–78 and Gils (n 36) 49–50.

46

Database Regulation, reg 14(1).

47

Under CDPA 1988, s 11(2) or Database Regulation, reg 14(2).

48

Major technology companies are reported to hold patents for blockchain-enabled inventions, such as Microsoft, Alipay, IBM and Toyota: B Zemp, ‘The Crucial Role of Patents and Intellectual Property in the Blockchain Industry’, Forbes (Medellín, 17 August 2023). Available at <www.forbes.com/sites/forbesbooksauthors/2023/08/17/the-crucial-role-of-patents-and-intellectual-property-in-the-blockchainindustry/> (accessed 21 February 2024).

49

European Patent Office, ‘Guideline 3.9.3: Cases where the Invention is Realised in a Distributed Computing Environment’ (2023). Available at  <www.epo.org/en/legal/guidelines-epc/2023/f_iv_3_9_3.html> (accessed 21 February 2024).

50

For instance, the claims in a UK patent, relating to a system for securing assets such as computer software using blockchain technology, use the expression ‘method’: S Savannah and CS Wright, ‘A Method and System for Securing Computer Software using a Distributed Hash Table and a Blockchain’ (UK Patent GB 2 561 465 B, 8 December 2021). Available at <www.ipo.gov.uk/p-find-publicationgetPDF.pdf?PatentNo=GB2561465&DocType=B&JournalNumber=6916> (accessed 21 February 2024).

51

J Carlin and D Sharrott, ‘Protecting Blockchain and Cryptocurrency Tech’, Intellectual Property Magazine (June 2018) 39. Available at <www.venable.com/-/media/038_039ipm_june_2018feat.pdf> (accessed 21 February 2024).

52

P Burchard et al, ‘Pseudonymous Transactions on Blockchains Compliant with Know Your Customer Regulations and Reporting Requirements’ (US Patent Application Publication No US/0117344 A1, August 2023) 1. Available at <https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/ 20 230 117 344> (accessed 21 February 2024). This application is just one example of a wave of blockchain technology-based patents that Goldman Sachs has recently filed, ranging from fractional banking using tokens to settlements with ‘SETLcoins’: Ledger Insights, ‘Goldman’s Patent Aims to bring Fractional Banking to Crypto, DLT’, Ledger Insights (17 August 2023). Available at <www.ledgerinsights.com/goldman-sachs-crypto-patent-fractional-banking/> (accessed 21 February 2024).

53

Goldman Sachs KYC/AML Blockchain Application ibid [0007].

54

ibid [0008].

55

ibid [0026].

56

ibid.

57

ibid [0027].

58

ibid [0030]. The ‘first-to-file’ rule is implemented through the priority date of the invention, which is the first date of filing: Patent Act 1977, s 5(1).

59

J Fusi, ‘Is the UK Patent Framework Helping or Hindering Home-grown Innovation?’ Aalbun (nd). Available at <www.aalbun.com/blog/is-the-uk-patent-framework-helping-or-hindering-home-grown-innovation> (accessed 21 February 2024).

60

Patent Act 1977, s 1(2)(c).

61

Some theorize that UK patents that are granted for blockchain-enabled technologies relate to ‘a physical system or input’, for instance, a blockchain working with biometric data: D Kramer and D Morgan, ‘Blockchain Patents – Comparing the Approach of Patent Offices’, Mathys & Squire Insights (6 October 0202). Available at <www.mathys-squire.com/insights-and-events/news/blockchain-patents-comparing-theapproach-of-patent-offices/> (accessed 21 February 2024). This can also be seen in UK Patent GB2549085A granted in 2016 for a ‘blockchain state reliability determination’, which transmits an executable code from a base station onto a targeted device (say from a network to a phone) in a secure way: D Joshua, D Gery and Z Mohammad, ‘Blockchain State Reliability Determination’ (UK Patent Application GB2549085A, 11 October 2017) 3. Available at <https://patentimages.storage.googleapis.com/7a/75/de/e38cb029d7e2c4/GB2549085A.pdf> (accessed 21 February 2024).

62

Patent Decision BL O/1113/22 (16 December 2022). Available at <www.ipo.gov.uk/p-challenge-decision-results/o111322.pdf> (accessed 21 February 2024).

63

ibid [38]–[40]. See Re AT&T Knowledge Ventures LP [2009] EWHC 343 (Pat) and Gemstar-TV Guide International Inc v Virgin Media Ltd [2010] RPC 10 for the ‘signposts’ to determine whether a computer software contains a sufficient ‘technical contribution’ for patentability. For a similar application of s 1(2)(c) of the Patents Act 1977 for a blockchain-related filing, see Patent Decision O/710/21 (28 September 2021). Available at <www.ipo.gov.uk/p-challenge-decision-results/o71021.pdf> (accessed 21 February 2024).

64

LC Baird III, ‘Methods and Apparatus for a Distributed Database within a Network’ (US Patent No US9646029B1, 9 May 2017). Available at <https://patentimages.storage.googleapis.com/8a/48/5d/a30fe18bd06301/US9646029.pdf> (accessed 21 February 2024). See further L Baird, M Harmon and P Madsen, ‘Hedera: A Public Hashgraph Network & Governing Council: The Trust Layer of the Internet’ (Hedera Hashgraph Whitepaper v.2.1., 15 August 2020). Available at <https://hedera.com/hh_whitepaper_v2.1–20 200 815.pdf> (accessed 21 February 2024).

65

Patent Act 1977, s 1(2)(c).

66

Patent Decision BL O/1113/22 (n 63) [48]–[50].

67

ibid [51].

68

The USA still allows for business method patents, following the decision in Alice Corp v CLS Bank International (2014) 573 US 208 and pending the Patent Eligibility Restoration Act of 2023: US

Patent and Trademark Office, ‘Software/Business Method Patents in the US and China – Comparisons and Strategies’ (September 2023). Available at <www.uspto.gov/sites/default/files/documents/Patents-in-the-US-and-China.pdf> (accessed 21 February 2024).

69

Goldman Sachs KYC/AML Blockchain Application (n 52) [0043–1].

70

See Database Directive, Recital 23. It could be separately protected as a literary work under s 3(1)(b) of the CDPA 1988.

71

For the purposes of this argument, it is assumed that the jurisdictional requirements for the blockchain to be granted a UK database right under reg 18 of the Database Regulation are met.

72

Database Directive, art 7(1); Database Regulation, reg 13(1).

73

European Patent Office, ‘Talking about a new revolution: blockchain’ (Conference Report, The Hague, 2018) 10. Available at <https://link.epo.org/web/Talking_about_a_new_revolution_blockchain_conference_report_en.pdf> (accessed 21 February 2024).

74

Database Directive, art 7(1); Database reg 16(1).

75

ibid art 7(5); Database reg 16(2).

76

Football Dataco Ltd and Others v Stan James plc and Others, Sportradar GmbH and Another [2013] EWCA Civ 27 [84] (‘Sportradar’).

77

William Hill (n 38) para 51.

78

DRSP (n 7) [231].

79

Case C-406/10 SAS Institute Inc v World Programming Ltd [2012] ECLI:EU:C:2012:259, para 65.

80

Sportradar (n 76) [80].

81

See Berne Convention for the Protection of Literary and Artistic Works (Paris Act of 24 July 1971, as Amended on 28 September 1979) WIPO TRT/BERNE/001, Article 2(5); WIPO Copyright Treaty (signed 20 December 1996, effective 6 March 2002) WIPO TRT/WCT/001, art 5.

82

Case C-444/02 Fixtures Marketing Ltd v Organismos prognostikon agonon Podosfairou AW [2004] ECLI:EU:C:2004:697 (‘OPAP’), paras 29, 32; affirmed in Case C-762/19 ‘CV-Online Latvia’ SIA v ‘Melons’ SIA [2021] ECLI:EU:C:2021:22, Opinion of AG Szpunar (‘CV Online Opinion’), para 18.

83

OPAP ibid para 33. See also OPAP ibid, Opinion of AG Stix-Hackl (‘OPAP Opinion’), para 39.

84

OPAP Opinion (n 82) para 39.

85

Case C-490/14 Freistaat Bayern v Verlag Esterbauer GmbH [2015] ECLI:EU:C:2015:735 (‘Verlag Esterbauer’), para 18.

86

ibid para 29.

87

Tzoulia (n 26) 295.

88

See E Derclaye, ‘Do Sections 3 and 3A of the CDPA Violate the Database Directive? A Closer Look at the Definition of a Database in the UK and its Compatibility with European law’ (2002) 24 European Intellectual Property Review 466, 469.

89

J Jenkins, ‘Database Rights’ Subsistence: Under Starter’s Orders’ 2006 7 The Journal of Intellectual Property Law & Practice 467, 470.

90

M Leistner, ‘Legal Protection for the Database Maker: Initial Experience from a German Point of View’ (2002) 33 India International Centre 439, 443.

91

LA Bygrave, ‘The Data Difficulty in Database Protection’ (2013) 35 European Intellectual Property Review 25, 29.

92

T Aplin, Copyright Law in the Digital Society: The Challenges of Multimedia (Oxford, Hart Publishing 2005) 47.

93

M Leistner, ‘The Legal Protection of Telephone Directories Relating to the New Database Makers Right’ (2000) 31 India International Centre 950, 956–7.

94

ibid.

95

S Chalton, ‘The Copyright and Rights in Databases Regulations 1997: Some Outstanding Issues on the Implementation of the Database Directive’ (1998) 20 European Intellectual Property Review 178, 179.

96

ibid 178–9.

97

Database Directive, Recitals 17 and 20. See OPAP (n 82) para 29 referring to Database Directive, Recital 17.

98

MJ Davison, The Legal Protection of Databases (Cambridge, CUP 2003) 72.

99

Aplin (n 92) 48.

100

ibid.

101

Beutler (n 21) 323–4.

102

Bygrave (n 91) 29.

103

Law Society Guidance (n 30) 165, which mentions ‘DLT’ but does include blockchain as a subset at 28–32.

104

OPAP Opinion (n 82) para 33. In practice, this approach can be helpful in analysing ‘enclaves’ of blockchains that are offered as a service by providers like Quorum Blockchain: TK Sharma, ‘What is Quorum Blockchain?’ (Blockchain Council, nd) . Available at <www.blockchain-council.org/blockchain/what-isquorum-how-is-it-different-from-other-blockchain/> (accessed 21 February 2024).

105

Pech (n 34) 70.

106

ibid.

107

Derclaye (n 88) 469.

108

Chalton (n 95) 179.

109

[2021] EWHC 237 (IPEC); [2021] FSR 25.

110

ibid para 136.

111

ibid para 79.

112

Goldman Sachs KYC/AML Blockchain Application (n 52) [0008].

113

ibid [0010].

114

ibid [0017].

115

ibid [0030].

116

ibid [0003], [0023] and [0031].

117

Beutler (n 21) 321.

118

OPAP (n 82) para 20.

119

E Derclaye, ‘What is a Database? A Critical Analysis of the Definition of a Database in the European Database Directive and Suggestions for an International Definition’ (2002) 5 The Journal of World Intellectual Property 981, 988.

120

Lederberg (n 20) 103.

121

M Leistner, ‘Legal Protection for the Database Maker: Initial Experience from a German Point of View’ (2002) 33 India International Centre 439, 444–445. This is similar to what Bygrave (n 91) at 25 terms ‘informational function’.

122

ibid. Beutler (n 21) 323 terms this a ‘digital representation of a single work’ of independently accessible but dependently interrelated elements.

123

OPAP (n 82) para 29.

124

ibid para 29.

125

This follows the idea behind the concept of ‘ejusdem generis’ in statutory interpretation. See ‘ejusdem generis’ Jowitt’s Dictionary of English Law (5th edn, London, Sweet & Maxwell 2019).

126

Perhaps this is due to the non-implementation in UK law of Article 1(3) of the Database Directive, although UK courts would still have been under an obligation to comply with EU law pre-Brexit: Derclaye (n 88) 473.

127

[1999] All ER (D) 600, QB; Davison (n 98) 71.

128

Davison (n 98) 71.

129

[1999] HCA 49.

130

Davison (n 98) 75; Database Directive, Recital 23.

131

Davison (n 98) 74.

132

Such data can be easily retrieved with a quick search with providers like Btcscan.org (<https://btcscan.org) or Blockchain.com. Available at <www.blockchain.com/explorer?view=btc> (accessed 21 February 2024).

133

Verlag Esterbauer (n 85) para 27.

134

Drescher (n 12) 119.

135

Verlag Esterbauer (n 85) paras 25–26.

136

In the Bitcoin blockchain, HRef 100 would correspond to ‘000000007bc154e0fa7ea32218a72fe2c1bb9f86cf8c9ebf9a715ed27fdb229a’: Blockchain.com, ‘Bitcoin Block 100ʹ (nd). Available at <www.blockchain.com/explorer/blocks/btc/100> (accessed 21 February 2024).

137

Verlag Esterbauer (n 85) para 24.

138

ibid para 23.

139

Derclaye (n 119) 986.

140

Gils (n 36) 23.

141

ibid 25.

142

Chalton (n 95) 178.

143

Gils (n 36) 25.

144

See Verlag Esterbauer (n 85) para 24

145

Derclaye (n 88) 469; Database Directive, Recital 20.

146

Leistner (n 121) 445.

147

Database Directive, art 1(2) and Recital 17; CDPA 1988, s 3A(1).

148

Goldman Sachs KYC/AML Blockchain Application (n 52) [0024].

149

ibid [0025] and [0027].

150

ibid [0028].

151

ibid [0016] and [0024].

152

ibid [0028].

153

Verlag Esterbauer (n 85) para 27.

154

Goldman Sachs KYC/AML Blockchain Application (n 52) [0032].

155

ibid.

156

Database Directive, art 1(2); CDPA 1988, s 3A(1)(a).

157

Law Society Guidance (n 30) 165.

158

M Leistner, ‘Creating Cyberspace — Intellectual Property Law and Competition Law Protection of the Web Designer’ (2003) 34 IIC 132, 139.

159

Database Directive, art 1(2); CDPA 1988, s 3A(1)(b). See also OPAP (n 82) paras 30–31 and CV Online Opinion (n 82) para 18.

160

Drescher (n 12) 83–84.

161

DeHive, ‘How to Solve the Blockchain Data Query Issue and What The Graph Service Has to Say’ Coinmarketcap (May 2021). Available at <https://coinmarketcap.com/alexandria/article/how-to-solve-the-blockchain-data-query-issue-and-what-the-graph-service-has-to-say> (accessed 21 February 2024).

162

For instance, it is possible to look up a transaction, an account address or a specific block on the Bitcoin blockchain via: Blockchain.com, ‘Bitcoin’ (nd). Available at <www.blockchain.com/explorer> (accessed 21 February 2024).

163

See J Polge, ‘Permissioned Blockchain Frameworks in the Industry: A Comparison’ (2021) 7 Information and Communications Technology Express 229.

164

Database Directive, Recital 15; Football Dataco (n 33) para 41.

165

CDPA 1988, s 3A(2).

166

Database Directive, Recitals 41–42 and Recital 58.

167

Database Directive, arts 7(1) and 7(4); Database Regulation, regs 13(1)–(2); British Horseracing Board Limited v William Hill Organisation Ltd [2005] EWCA Civ 863; [2001] RPC 31, HC [30].

168

G Wei, ‘Telephone Directories and Databases: The Policy at the Helm of Copyright Law and a Tale of Two Cities’ (2004) 3 International Pharmaceutical Quality 316, 346–7.

169

See further Law Society Guidance (n 30) 166–7.

170

CDPA 1988, ss 16–27.

171

Database Directive, arts 7(1) and 7(5); Database Regulation, reg 16.

Author notes

Ilsu Erdem Ari has obtained a BA (Hons) in Law at the University of Cambridge (Trinity College) and an LLM at the London School of Economics and Political Science. She recently completed her Bar Vocational Studies at City Law School, which she represented in the Oxford IP Moot 2023 and was a quarterfinalist. She has experience as a Copyright Consultant for the Cambridge Journal of Law, Politics and Art. She has a keen interest in the interaction of IP law with new technologies and has completed the Computer Science for Lawyers course of Harvard Law School Executive Education.