In the summer of 2023, songwriter and producer Julia Karlsson muses on the scope of digital technology and artificial intelligence and the possibility of music generated by AI. Three centuries earlier, German composer Johann Kuhnau (1660–1722) busied himself with the development of a combinatory logic for music. The plan was to publish a treatise on composition wherein a mathematical ars combinatoria would aid the composer with the invention (as it was called then) of basic melodic materials. It failed. However, the idea turned into a parlour game towards the end of the century that was played with dice; it allowed the players to randomly compose little ditties over a cup of coffee or a glass of wine, while having a discussion, perhaps about the rational principles of the fine art of music, in line with the prevailing enlightenment zeitgeist (Rose 2019, 43–45; Ratner 1979). This is not what is being discussed today. Behind Karlsson’s pondering one senses the issues of Big Data and AI. Add music to the mix, and we’re talking about Big Money. In this chapter we will alter the theme slightly. Rather than inquiring into the nature of musical creativity and how it may accord with combinatorial algorithms (dice-thrown, computer-based, or other types), we will focus on the end product, the music.

More specifically, we want to explore how techniques used for musical identification may shape the ways in which people—composers, producers and listeners—hear sounds as music. For more than three centuries, manual and automated techniques used for identifying music have played a central role in efforts to safeguard rights to music and to define what counts as music in the first place. In other words, musical identification techniques are not just tools for managing music rights and finances but are also vehicles for establishing and safeguarding culturally specific ideas and boundaries concerning music itself.

To test and explore this thesis, we focus on three moments in history that each reveal unique standards for recognizing and defining music. These are introduced with examples of legal issues that arose in 1817, 2002 and 2017. We focus on legal conflicts at each of the moments in time concerning the authenticity and originality of musical works, because such periods of conflict provide fertile grounds for examining how dominant ideas regarding creativity and authenticity correlate with methods that are used to identify sounds and music. Central to our investigation is the realization that distinguishing one piece of music from another is key to the protection of copyright and that the principles that underlie such distinctions define what a composer does, what a producer produces and what listeners listen to. In other words, music identification techniques both shape and delimit musical possibilities.

Our analysis shows that prior to the 2000s, a series of different technologies played a central role in efforts to identify original musical works: technologies such as musical notation and the printing press (in 1817), and music recording and playback devices (in 2002). However, these technologies only played an assistive and supportive role, as the ultimate decision regarding what counts—and what doesn’t count—as an authentic and original musical work worthy of legal protection was made with the help of a human ear. After the turn of the twenty-first century, however, we can observe a paradigmatic shift as the task of identifying music (not just representing it to the decision makers) is increasingly outsourced to machines. When algorithmic systems are increasingly used to recognize and compare musical works, we can see how the role of the musical listener (expert or not) in identifying music is diminished, benefitting software systems and their ways of tracing and tracking musical identities.

In other words, we assume, as our point of departure, that techniques used for musical identification shape the ways sounds are heard as music—but we want to ask how that happens. We know that copyright is a modern historical phenomenon, and we know that both “music” and “musical work” are Western concepts—but our inquiry aims to highlight the contingencies behind these principles. Our historical perspective is thus intended to be a critical examination of what may appear to be given, non-historical, perhaps even natural preconditions. This means not only that we reject any technological determinism but also that we regard principles which define what a composer does, what a producer produces and what listeners listen to—that is, music—as always being susceptible to re-conception and technological change. In the end, that is what artistic freedom and authenticity are and should be about.

1817: Notation Manuscripts, the Printing Press and Reading as Musical Identification Techniques

In 1817, the heirs of Swedish composer and poet laureate Carl Michael Bellman (1740–1795) sued a publisher, Olof Åhlström (1756–1835), for reprinting a collection of Bellman’s songs. The heirs, Bellman’s widow and son, claimed they had inherited the ownership of the songs and that only they had the right to publish. In the end, they lost the case. Åhlström succeeded in convincing the court that he had rightfully acquired the manuscripts of the lyrics directly from Bellman: Åhlström had even written down the melodies himself, that is, produced an original manuscript—which gave him the right to copy it. Moreover, he already had a royal privilege (an exclusive permission granted by the king) to print the collection in question, the plates of which he had kept since its first edition.

This anecdote is significant for several reasons. Bellman’s heirs referred to the Swedish Freedom of the Press Act of 1810. According to this constitutional law, “each and every writing is the author’s or its legal rights holder’s property”. That is, the author was the owner of the physical manuscript until he or she sold it. The author owned the notational manuscript, just like one can own and sell a physical object such as a bicycle or a diamond ring (commercial public performances of music had not yet become a widespread issue in Europe and were not regulated in the way that print was). Åhlström, in addition to rightfully owning Bellman’s manuscripts, which gave him the right to copy them, also had a royal privilege to make musical prints. Such privileges had been the norm in Europe for centuries. Now they were successively being replaced by egalitarian laws. Although this court case illustrates legal practice at the turn of the nineteenth century, Åhlström’s royal privilege was a remnant of a feudal tradition. It was not renewed when it expired in 1823 (Byström 1931; see also Volgsten 2012, 146–147).

It was also possible for composers to acquire these privileges. This was not done primarily for economic benefits; the main reason was to control the correctness of the printing of their own works.Footnote 1 Since its invention, musical notation (which began around the first millennium AD, long before Gutenberg had invented the printing press technique) had mainly been used to notate music after the fact, that is, after the music had been performed. But during the fifteenth century, notation increasingly became a tool for composing music before it was performed. What was written down, which we call the score today, became known as an opus, a work—writing which increasingly replaced oral transmission, assigning a definite and traceable identity to the composed music (Volgsten 2012; von Loesch 1998; Wegman 1996). This written manuscript could be multiplied by printing copies of it. Acquiring a royal privilege for the printing of the music was not, however, copyright in any modern sense of the word. The composer was not credited for creating anything. It was thought that only God could create in the sense of making something out of nothing. The composer’s job was instead likened to that of an artisan, such as a clock maker or a lapidarist. And just like for the artisan, the quality of the crafted product functioned as a recommendation for future commissions. Gaining control of the printing and reprinting processes was thus a way for the composer to finally identify and thereby control the quality of the end product, the printed work, which would ensure future commissions to compose (Rose 2019, 81–83).

Correctness and credit were the issues at stake. Piracy was not the issue for the composers (it was the printers’ business). But neither were imitation and similarity, issues which today would be regarded as plagiarism. Imitation and reuse could even be considered reverent homages.Footnote 2 A famous example is that of composer and theorist Johann Mattheson, who in the 1730s tells us that Georg Friedrich Händel has used one of Mattheson’s arias “almost note for note” in his opera Agrippina. Rather than accusing Händel of plagiarism or theft, Mattheson expounds an economic metaphor that was new at the time. “Borrowing” from others, Mattheson says, is acceptable if you pay back “with interest”, if you further elaborate the composition or the parts of a composition that you have borrowed—and credit the lender by name (Mattheson 1999, 131). But substandard revisions or adaptations are not welcome, since they are detrimental to the precursor’s reputation (Rose 2019, 116–118).

In his 1739 treatise Der vollkommene Capellmeister (The Accomplished Music Director), Mattheson gives us a clue about the craftmanship behind the music that a composer may wish to control. In accordance with the rhetorical principles of the time, a performance of a piece of music is preceded by three stages: inventio, elaboratio and dispositio (Mattheson 1999, 122). Simply put, the first concerns the “invention” of thematic material, the second the elaboration of the themes according to different compositional principles, and the third relates to their disposition in accordance with the rules of established genres. Borrowing, fairly or foully, could therefore concern any or all of these stages. The reason why borrowing thematic ideas was not regarded as straightforward theft is that inventio was not regarded as creation in any divine or metaphysical sense. Rather than referring to creating something out of nothing, inventio referred to an inventory, or a common stock of ideas that was available to everyone, a locus communis (Mattheson 1999, 123). The crux of the inventio was the capability of the composer to find the right ideas at the right time, a capability that not everyone was gifted with. Despite this, techniques were developed for memory retrieval, as well as for formulaic variations of themes, of which the ars combinatoria, mentioned in the introduction, was but one (Rose 2019, 36–38).

What these examples tell us, then, is that in situations where the source of income was not the making of copies (such as print), plagiarism was not necessarily a serious problem and could even be seen as a compliment or sign of respect, given that the originator is duly credited. More particularly, the examples show how the control of print became an act of identification, of specifying what exactly had been done by the composer (and no one else) in the masterful act of composition—the art of inventing, elaborating and deciding the disposition of musical material. A third point to be made is that when composers’ rights first became regulated by egalitarian laws (rather than feudal privileges), the rights concerned the ownership of an original physical manuscript. However, the notion of owning the copyright to a non-physical, immaterial “work” was yet to come.

2002: From Reading to Listening and the Impact of Sound Recordings and Playback Machines on Musical Identification

Nearly 200 years after the Bellman case, as the twentieth century was drawing to a close, a case of music plagiarism reached the Supreme Court in Stockholm. This was a rare occasion. Most cases of plagiarism were solved outside court, often following a routine whereby the Swedish composers’ collecting society, STIM, summoned an assessment committee to give a “guiding statement” on the issue. In this case, however, the record company EMI chose to sue one of its Swedish competitors Regatta. EMI considered that the song “Tala om var du ska resa” (“Tell Me Where You’re Going/Traveller in My Songs”), recorded by a group called Landslaget (The National Team), had been plagiarized by a group called Drängarna (The Farmhands) via their recording of the song “Vill du bli min fru” (“Will You Be My Wife”).

The case only concerned a limited part of the two songs, a short instrumental interlude of eight measures which in both cases is repeated throughout as a kind of refrain. Both melody lines consist of four phrases. The first and third phrases are the same in both melodies—the third is a repeat of the first—while the second and fourth phrases differ. In other words, the two songs have the same melodic content—same sequence of intervals—in only four out of eight measures.

Regatta pleaded in its defence that their melodic line was inspired by a traditional Swedish folk song, “Oxdansen”, and should therefore be considered in the public domain (lots of songs in Scandinavian and Anglo-American folk and popular music are based on similar melodic material). In addition, they pleaded that consideration should be given to the formal functions of both rhythm and harmony, since harmonic identity determines the different musical function and character of the two melodies’ otherwise identical interval sequences.

In the end, the Supreme Court followed EMI’s line and the case was judged in their favour. EMI’s lawyer convinced the court of their claim by, among other things, having the two melodies played simultaneously and unaccompanied, whereby they were perceived as similar enough to be judged to be identical, leading the court to the conclusion that the latter melody was a plagiarism of the former (Edlund 2007, 116).

This case shows important changes that had occurred since the Bellman vs Åhlström case in 1817. By 2002, when the case was heard, copyright had become a statutory right and had replaced the property right to a written manuscript. As is still the case, what is protected by law is the musical “work”. But changes are also significant in musical practice, as is evident when we compare the situation with Mattheson’s treatise. Whereas for Mattheson, a Capellmeister was a performing musician as much as a composer, the two “functions” were severed in the nineteenth century: a composer could act as a performing musician, but a musician was not a composer. To secure a composer’s right to remuneration, the composing of a “work” of music was defined in contrast to performing, as a unique act, whereas performing was (and still is) repeatable. According to the romantic and idealistic metaphysics of the time, and in line with the needs of copyright jurisdiction, composing was defined as creation. Whereas for Mattheson, masterful composition was subject to a universal ideal to be pursued by each and every composer, in the nineteenth century the universal gives way to the original.

Another practical issue that resulted in intense theoretical debate in the nineteenth century was the wish to control both the printing of musical works and the performances of them (Kawohl 2002). The property right to a physical manuscript was designed to control print, not performances. As ticket sales to public concerts increased, performances were a source of income that could not be neglected. But according to what principle could the two—print and performance (the one an object, the other an activity)—be regarded as of a single kind? The answer given by music theorists and legal scholars of the time was that the musical work created by the composer is an abstract entity identifiable through its individual—unique and original—form (cf. Marx 1997; Kohler 1892). Print and performance (and subsequently sound recording) are no more than material expressions of this idealistic form. This was the philosophical foundation upon which nineteenth- and early twentieth-century copyright law came to be articulated, on the European continent as well as in Scandinavia (Strömholm 1970).

The separation of composer and performer and the assignment of the abstract form as the common identifying denominator of print and performance both rely on this concept, according to which a musical work is an intangible, non-physical entity that exists beyond space and time, independently of being realized in sounding performance (Goehr 1992; Volgsten 2015). This concept of the musical work is at the centre of national copyright laws as well of the international treaty dated 1886 that goes under the name of the Berne Convention (even today’s TRIPS agreement of the WTO assumes the “work” concept by referring to the Berne Convention in its initial paragraphs). The Berne Convention comprised sound recordings at the outset of the twentieth century. Thus, copyright metaphysics (if one may label it as such) influenced the way gramophone records came to be viewed. In the 1930s, the International Federation of the Phonographic Industry (IFPI) argued that sound recordings should be viewed as complex generic works of music (Zucconi 2022, 169–170). Although the attitude towards recordings may have caught on, the IFPI lobbyists failed to secure copyright, achieving in its place a so-called neighbouring right to recordings (Fleischer 2015). The resulting strategy of the music industry was to persuade composers and creators of musical works to sign away their rights to the record companies.

This was the situation for literally thousands of creators, composers and song writers during the twentieth century, and it still is at the time of writing. This is also the reason why a major record company such as EMI may sue another record company for plagiarism, as in the Swedish case mentioned earlier. The company controls the copyright. The case is also telling since it reveals a tendency whereby the market for sound recordings in the twentieth century changed the identifying practice of the previous century. While the defendant’s lawyer referred to formal functions of both rhythm and harmony, which in this case apparently showed similarities to a traditional melody rather than to the plaintiff’s song, the prosecutor appealed to a so-called overall assessment, whereby the jury could hear for themselves that the melodies were similar enough (Rosén 2004, 14; it is not clear, however, to what extent actual recordings were used as sounding evidence in court). In other words, the nineteenth century principles of formal identity were in the end overridden by an appeal to a subjective judgement of sounding music. Although this turn from formal analysis through the reading of notational scores to subjective listening had not been discussed in Swedish copyright doctrine (Scandinavian doctrine had busied itself instead with disposing of the idealist concept of the musical work; see Strömholm 1970), it had—just like the principle of “the reasonable man”—been the norm in the United States since the composer of popular songs Cole Porter was (unsuccessfully) sued for plagiarism in the mid-1940s (Balganesh 2016; Manta 2012).

Although one may get a sense that copyright laws are designed to facilitate court decisions rather than to secure income for composers (a legal engineering, one might say, paraphrasing Popper’s notion of social engineering), perhaps the most significant practical conclusion is drawn in a handbook on judicial matters, written in Swedish for Swedish music creators and published at about the same time that the EMI vs Regatta trials were taking place. It was formulated as a warning to its readers (summarized here in translation): “If you are in the least bit unsure whether you have been too inspired by another work or borrowed from already existing works, then try to settle this before publication, so that there is no risk of accusations of plagiarism. Such cases can be both painful and costly for those involved” (Stannow et al. 1999, 42). The warning would turn out to be even more urgent as the identifying techniques changed once more after the turn of the millennium.

2017: Digital Music, Audio Fingerprints and Algorithmic Identification Techniques

Well into the new millennium, in 2017, concert pianist Valentina Lisitsa received a message from YouTube saying that her performance of Johann Sebastian Bach’s Partita no. 2 had been removed from its website. Lisitsa was informed that 90 seconds of her video were an infringement of the copyright owned by Sony Music. A similar message reached classical pianist James Rhodes in 2018, informing him that one of his tutorial videos on how to play Bach had been removed from YouTube, since it contained 47 seconds of music that was also said to belong to Sony Music. Likewise, a test by music professor Ulrich Kaiser to see what music could be uploaded to YouTube generated similar messages concerning music by Schubert, Wagner, Beethoven and several other composers—all deceased long ago and whose music rests safely in the public domain (that is, the rights to the music belongs to no one and everyone, see Bottum 2018). The situation not only concerns performers but also creators of music. And whereas it may seem that the most obvious group to be affected by this is creators of mashups (Brøvig-Hanssen and Jones 2021), any active composer, producer or songwriter may be affected by the scope of the new technology.

Although not court cases, these are examples of how present-day digitized identification technology “create[s] an additional layer of private adjudication in which initial determinations about […] copyright are made before the publication of the work” (Urban et al. 2017, 59). Whereas the identification of music after 1817 primarily involved an act of reading musical notations, which by 2002 had given way to listening, 2017 signalled a shift to algorithmic ways of identifying what does and does not count as music worth protecting and safeguarding. Largely developed in response to the gigantic amount of content that is uploaded online every day, algorithmic music identification techniques introduce a solution to the problem of manually scanning the internet to search for copyright abuse.

The technology at the heart of the YouTube copyright disputes mentioned earlier is the system known as Content ID, a device used for automatic content filtering and sound analysis, which by a single brushstroke has made traditional identifying procedures such as listening or formal analysis obsolete. Content ID belongs to a set of software technologies known as audio fingerprint technologies, which automate the detection of digital audio files that violate copyrights online. They thereby unburden rights holders from the task of manually scanning online content in search of evidence of copyright infringement. Instead, algorithmic techniques such as Content ID silently scan and filter user-uploaded files in the background of platforms, only making themselves known when suspect content is identified. Paired with systems of so-called notice and takedown, audio fingerprinting techniques have become a powerful and efficient tool that allows (selected) rights holders to automatically block, monetize or track how their copyright-protected content circulates on the platform (Eriksson 2019).

On a technical level, audio fingerprinting techniques combine the skills of reading and listening to music to determine its originality, although these tasks are no longer performed by human eyes or ears but rather by complex combinations of algorithms. The process of making and comparing fingerprints involves converting digital music files into visual representations and mapping/compressing the internal characteristics of sound files in elaborate ways. The algorithms that perform such tasks are directly modelled on human hearing and represent a form of perceptual coding (Sterne 2012) which mimics human auditory perception to the extent that Content ID can detect heavily remixed and modified sound content in similar ways to a human listener. Regarding music specifically, Content ID identifies content by isolating key sound characteristics in copyright-protected audio files that rights holders want to track across the platform. Content ID saves and compresses such sound characteristics into smaller data files—that is, fingerprints—that are stored in a database. This database is then used as a source against which other videos on YouTube can be matched and compared. Currently, all YouTube videos (with or without music content) are filtered through Content ID, which immediately notifies selected rights holders of suspect cases of copyright abuse. Elsewhere, companies such as AudibleMagic are applying similar techniques for detecting music on the radio, in streaming and in TV broadcasts. Taken together, these techniques provide radically new ways of making music automatically identifiable based on its auditory features, such as melody, rhythm or pitch. As musical identifications are transformed into mathematical procedures, we also witness a shift where the appreciation of music becomes a matter of numerical calculations and statistical evaluations of similarity (Ernst 2016).

Importantly, the audio fingerprinting techniques that started to emerge around 2017 would not be as powerful and influential if they were not combined with notice-and-take-down systems, that is, systems whose acknowledged rights owners are immediately notified of suspect copyright infringements and then offered a series of choices regarding how to act on that information vis-à-vis the person who uploaded the content. On YouTube, the options offered to such rights claimants are as follows: (1) immediately blocking the content from being shown online, (2) allowing the content to be uploaded but getting access to information regarding how it is seen/used, and (3) allowing the content to be uploaded but getting full control of its advertisement royalties. Claimants associated with major record labels and collecting societies can choose one of these options on YouTube as a default setting so that takedown demands or ad-royalties claims can automatically be sent out in response to all detected cases of suspected copyright abuse. By 2018, for instance, YouTube had sent out more than one thousand messages wherein Sony Music claimed it owned the copyright to music by Johann Sebastian Bach (Bottum 2018), likely as a result of automation as opposed to human analysis.

Unsurprisingly, this automation of copyright enforcements implies that a significant amount of power is located in the hands of those who have their music fingerprinted first. A key aspect of how content fingerprinting techniques uphold divisions between “authentic works” and “illicit copies” can be found in the database where fingerprints are stored. Thus, it is this database that constitutes “the benchmark for originality” in the online music domain (Edwards 2018, 68), since any new upload that bears similarities to the content in a fingerprint database runs the risk of being identified as an illegitimate copy. This makes the question of who is allowed access to YouTube’s fingerprint reference database—and by way of precisely which decision—a highly political issue (Eriksson 2023). Whereas major labels are granted the advantage of having full access to Content ID, independent labels or artists are often given access to a simplified version of the system, called Copyright Match, which includes the same system for identifying fingerprints but does not give rights holders the right to automate takedown or monetization requests. Unlike major labels, independent labels/artists who want access to Copyright Match are also forced to upload their music onto YouTube and make it fully available to the public on the platform before it can be fingerprinted and thus be qualified as a unique musical work worth protecting. In other words, YouTube’s reference database for musical fingerprints is a corporate asset that YouTube/Google retains full strategic control over, including maintaining the power to decide precisely who is allowed to access the database and who is not and on what conditions.

In 2018, Google reported that Content ID was hosting more than 80 million fingerprints of copyright-protected content and filtering roughly 600,000 hours of YouTube videos per day. Google also claimed that 98 per cent of the copyright infringement complaints made to YouTube had been channelled through Content ID and that 97 per cent of those copyright complaints were the result of purely automated processes (meaning that no human being had been directly involved in detecting or issuing the complaints; Google 2018, 24). By conducting hidden forms of sound analysis and identification at scale, Content ID has thus succeeded in automating most of the copyright disputes that occur on YouTube. In doing so, it has also opened up vast amounts of video content on YouTube—videos that were previously incomprehensible due to their volume—to copyright control. In short, the logic applied by YouTube Content ID is that music is what is stored in its fingerprint reference database. And as this system for identifying musical works is sifting through all content that enters one of the world’s largest websites—day and night, round the clock—it is also having a significant impact on how music is defined.

Discussion: On Documents, Recordings, Digital Content and Manual/Automated Protection Techniques

This overview has explored techniques used to identify music at three historical moments (signified by the years 1817, 2002 and 2017). We have highlighted significant changes in legal practice and the historical relativism of the norms and values on which institutionalized control mechanisms have relied, including changes in the techniques used for identifying music and the reasons for doing so.

In 1817, the identification of music relied on notation, and since no copyright existed that could conceive of a right to both print and performance, issues concerned property rights. The right to make copies, written or printed, belonged to the owner of the original manuscript. By 2002, copyright laws had long since been established, but during a century characterized by the development of recording technology, the identification of music increasingly relied on the analysis of sounding music. In both cases, it was the human sensory system that was directly used to execute the final mode of control, that is, deciding whether a set of notations comprised illegal copies and determining whether two sound recordings sounded similar enough for one to be considered an inauthentic replica. With algorithmic techniques for the identification of music starting to be used in about 2017, however, these tasks become outsourced to machines scanning immense amounts of online content.

In these historical examples, principles of identification in music have served different purposes. By emphasizing the role of eighteenth-century composers’ privileges (rather than printers’ privileges, or the system of privileges as such), we have seen how controlling the musical print served to protect the authentic work from distorting errors. This was a measure taken by the composer to ensure that the printed score disclosed only those notes decided by the composer’s aesthetic judgement. This sifting of the chaff from the wheat implied an identification of the composition to be published in the composer’s name. In other words, the identification technique and the compositional technique coincided, and the authenticity claims of the former did not tamper with the artistic freedom of the composer. This practice changed as reference to the work came to serve a new purpose, that of subsuming print and performance under the same identifying principle (the abstract form of the individual musical work). This is where modern musical copyright enters the scene. However, at the same time as sound recordings outcompeted sheet music in sales, courts increasingly appealed to a “reasonable listener” to ensure a safe zone around the copyrighted music. At the time of writing, no songs that are similar to existing ones are allowed, and perceived similarity is enough to claim infringement (disproving plagiarism becomes the responsibility of the accused). Close enough is also the principle for fingerprint identifications of digital technology.

By being highly attentive to the qualities of sound—and carefully drawing boundaries between good and bad, legal and illegal musical expressions—music identification techniques, ranging from visual notation comparisons to auditory sound analysis to algorithmic fingerprinting scans, have all helped to produce particular musical rights regimes: regimes centred around music in the form of manuscripts, sound recordings and digital binary information. Each of these regimes has also opened up new ways of tracing and tracking copyright abuse. But there is at present an unjust division that benefits major stakeholders in the industry by giving them opportunities to block, monetize and track how copyright-protected content circulates—what was described above as “private adjudication” (Urban et al. 2017, 59) and which could even be described as a modern form of rights privilege at the expense of democratic jurisdiction (cf. the notion of refeudualization in Habermas 1989). A perhaps more worrying tendency that emerges as these historical moments are juxtaposed is how identification and control increasingly come to imply that a criminalizing attitude is being taken towards music creators. Identification techniques are no longer applied to secure aesthetically satisfying performances, which used to be the aim of the composer’s control of print. Instead, today’s digital fingerprinting techniques strike against any music that may sound similar—in a process that can be described as turning from identifying pirate copies to making accusations of plagiarism. Anyone trying to write a song who plagiarizes until a personal style is found—as mentioned by Andreas Carlsson in the introductory quote—will be constantly at risk of being accused of theft. And once the accusation comes, and after a notice-and-take-down message is received, there is no local, regional or even national court to appeal to for a just and public trial, while music becomes the crop of a minefield of forfeit.

But maybe Andreas Carlsson has spotted the crack that will eventually let the light in when he somewhat prophetically says that “[m]usic has a strong vibe which makes it more powerful a communication device than any religion, any political slogan, any ad. A three-minute pop song […] will make people around the world move in an odd manner. It is such a strong force that you need to have respect for it. As long as we control it and make sure that it is used for pursuing good aims, music is the best thing we have on earth, I think, and with digital assistance we can make this fantastic thing accessible to everyone”.