Mashup music as expression displaced and expression foregone

Introduction: mashup music as expression displaced and expression foregone

Music sampling is an established music appropriation practice with roots in dub, rap, hip-hop and electronic dance music (Demers, 2006, pp. 98–110; Vaidhyanathan, 2003, pp. 134–138) and more broadly in popular music (Schuster et al., 2019).¹ As a vehicle for expression, sampling enables creators to ‘articulate a personalised “aural” history—an archive of sounds that can be employed to express specific musical and political statements’ (Rodgers, 2003, pp. 313, 315). As samples often feature copyrighted music, sampling is also tethered to copyright law.²

This article focuses on mashup music, a form of sampling expression combining samples from two or more recognisable and popular music recordings into a new whole (Brøvig-Hanssen, 2016; Hui, 2018; Sinnreich, 2010). While mashups also exist in music and video forms, this article focuses on the musical variety. It explains why mashups are a type of ‘expression foregone’. This is inspired by the term coined by law scholars Kylie Pappalardo, Patricia Aufderheide and their colleagues—‘imagination foregone’—to describe creators with a ‘risk-averse understanding of copyright law’ who self-censor their work, by abandoning creations-in-progress, changing creations to avoid copyright issues or not creating at all (Pappalardo et al., 2017, p. 38). I use the slightly revised term ‘expression foregone’ to make an explicit link to the expressions that mashup producers forgo, noting this link is implicit in the term coined by Pappalardo et al. I also give attention to how platform regulation—including both content identification and content moderation by platforms and on platforms—silences and otherwise alters mashups without the consent of mashup producers.

By using copyrighted musical recordings, mashup music attracts copyright regulation by copyright rights holders and operators of online platforms. Platforms increasingly mediate online musical expressions, extending and distorting the reach of copyright law. Although mashups predate platforms such as YouTube and SoundCloud, most mashups are now shared on and via such platforms where content identification systems—operated in-house for platforms, and third-party on platforms for rights holders—detect when copyright material is used. Content identification ensures that online music and music sampling are increasingly entangled with copyright, and makes mashups especially vulnerable to content moderation, whereby a platform may take down or block a mashup, or suspend or terminate a mashup producer’s user account, even when the mashup’s use of copyright material is not proven to be infringing.³ Without content identification, content moderation would be blind to many uses of copyright material.

I present this article in two parts. Firstly, I explain that most US and EU cases concerning music sampling and copyright do not establish that all or even most unlicensed music sampling would be copyright infringement. This leaves mashup music in a legal grey zone. Secondly, I show how mashups have become expression displaced off platforms and expression foregone. I argue that the pervasive use of content identification without consideration for copyright exceptions has improperly treated mashups in a legal grey zone as likely infringement, leading to takedowns and blocking that cause producers to abandon creating and sharing mashups.

In this second part, I draw on empirical research—30 semi-structured qualitative interviews with producers and a survey of 92 producers—that my colleagues and I at the University of Oslo’s MASHED project conducted in 2019.⁴ For the interviews, mashup producers were selected from 14 EU countries and the US to represent a diversity of age, gender, and mashup-related attributes, including online platforms used to distribute mashups, mashup types, level of popularity and current level of involvement in mashups. Around two in three interviews were conducted via online video, on the videoconferencing platform that suited the interviewee, and the remaining third were in person. Most interviewees chose to be quoted pseudonymously, though some chose to remain anonymous. For the survey, we recruited producers via social media posts (mainly Twitter), personal messages, online forums and word-of-mouth. The survey involved a combination of open and closed scaled questions.⁵ Responses were anonymous, in accordance with EU General Data Protection Regulation requirements. In conducting this empirical legal research, we agree that ‘often problems within the legal system, best practice insights and the effect of policy shifts can only be examined using in-depth, qualitative methods’ (Webley, 2010, p. 948). Our interviews and survey follow many other interviews and surveys conducted in the fields of empirical legal studies (Cane & Kritzer, 2010) and critical information studies (Vaidhyanathan, 2003). We have conducted this empirical research to understand contradictions and complexities in the operation of copyright law. Our interviews and survey have also enabled us to test assumptions in existing literature and to infer the social impact of laws that cannot be found through traditional black-letter approaches, such as statutory interpretation and close analysis of court judgements. In interpreting the information gathered through interviews and survey responses, we have been mindful not to conflate stories and personal accounts with legal facts and evidence.

While this article recognises that obtaining sample licences is an avenue for lawful sampling, it focuses on other avenues, namely US and EU copyright limitations and exceptions in the first part and licences between platform and music rights holders, not specific to samples, in the second part. I consider obtaining sample licences to be a genuine option in some cases, particularly where the mashup producers or sampling artists are signed to a label or have significant means to pay. Where licensing is the chosen avenue for lawful sampling, at least two licences are required from the respective rights holder for each sample: one for the composition, known as a ‘musical work’, and another for the recording of that musical work, known as a ‘sound recording’ (or a ‘phonogram’ or ‘phonorecord’).⁶ The same copyright item can have different rights holders in different jurisdictions, and rights can be sold or otherwise transferred from right holder to right holder.

Generally, seeking permission is made easier by collective management organisations that represent many (and sometimes a majority of) rights holders in a jurisdiction. US and EU examples of collective management organisations include the American Society of Composers, Authors and Publishers (ASCAP), Gesellschaft für musikalische Aufführungs- und mechanische Vervielfältigungsrechte (the German society for musical performing and mechanical reproduction, GEMA) and Svenska Tonsättares Internationella Musikbyrå (the Swedish performing rights society, STIM).

However, even with collective management organisations, seeking permission for music sampling use of musical works and sound recordings remains difficult. I note that research in the early 1990s (Broussard, 1991) and 2010s (McLeod & DiCola, 2011) have shed light on the difficulties of obtaining sample licences, including unaffordable costs of up to USD 5,000 dollars per sound recording used and 50% of both royalty revenues and ownership per musical work. Other than noting that such costs could be prohibitive for mashup producers, who all sample at least two sound recordings and two musical works per mashup, this article does not further explore sampling licensing.

Mashups as expression regulated

Music sampling and copyright law have long been entangled. While US and EU copyright laws grant a set of exclusive rights to copyright-protected expressions, music sampling is a form of expression that reuses such prior expressions. As such, music sampling and copyright exist at the intersection of important legal rights. In the US, these include the right to freedom of speech and intellectual property rights established under the progress clause (Constitution of the United States of America, art.I, §.8, cl.8 and amendment I), and in the EU, these include the rights to freedom of expression and information and to property (Charter of Fundamental Rights of the European Union, 2012, arts. 11 and 17).

Amongst expressive forms relying on prior expressions, mashups and other forms of music sampling are especially entangled with copyright. Even amongst forms of music sampling, mashups can be particularly tied to copyright. First and foremost, mashups use at least two samples, which involves the use of at least two musical works and two sound recordings. In addition, mashups preserve the recognisability of samples, making detection of their use more likely to rights holders.⁷ This recognisability is preserved through the use of samples of extensive length, the preference for sampling from well-known recordings, the restrained use of audio effects and audio manipulation on the samples, and the titles of mashups which sometimes reference the sampled tracks’ titles. The recognisability of samples in mashups makes the use of copyrighted music particularly obvious and makes mashups a common target of copyright disputes, as this part explains, and copyright-related content identification and content moderation, as the next part explains.

Copyright-related court judgements featuring music sampling make clear that some sampling uses are unlawful but leave many other instances in legal uncertainty. As the following discussion explains, this case law shows that under some US and EU copyright law limitations and exceptions—particularly the de minimis limitation, the US fair use exception and the EU quotation exception—sampling infringes copyright.⁸ However, legal uncertainty remains because this case law leaves significant gaps in our understanding of how copyright limitations and exceptions apply to mashups and other music sampling.

Sampling, including mashups, is often not de minimis

Unlike many other forms of creations, mashups and other forms of sampling are not reliably permitted by the concept of de minimis, short for the Latin de minimis non curat lex which translates approximately to ‘the law cares not for trifles’. Typically, copyright permits the use of tiny parts—de minimis uses—of copyright expressions. As a result, music creators using only a tiny part of a musical work, such as two notes or a single chord, do not use a copyright work under the law and do not need to seek a licence. For example, the US Ninth Circuit Court ruled the use of three notes from James Newton’s musical work Choir in Beastie Boys’ Pass the Mic was not copyright infringement (Newton v. Diamond, 2003, 349 F.3d 591 (9^th Cir.)).⁹ Copyright grants rights holders no ability to prevent or otherwise interfere with such uses of musical works.

However, the de minimis concept has permitted just a handful of cases of sampling considered by existing US and EU copyright jurisprudence. This is a significant impediment. If de minimis were to reliably permit sampling under copyright law, it would provide material space for unlicensed but lawful sampling expression. The crux of the de minimis is whether the part used is tiny; there is no in-principle restriction on how many times that tiny part is used in the remix, mashup or work of sampling. Common practices in sampling, such as looping of samples and constructing new beats, grooves and riffs from slices of sound, can use such tiny parts, though some sampling forms such as mashups typically rely on longer samples.

Though de minimis is undoubtedly a quantitative concept based on temporal length, it also has qualitative aspects, at least under EU law. In 2019, the Court of Justice of the EU rejected de minimis in a copyright dispute over a two-second sample of Kraftwerk’s ‘Metall auf Metall’ in Moses Pelham, Martin Haas and Sabrina Setlur’s ‘Nur mir’. The CJEU ruled that sampling is generally not de minimis, stating ‘the phonogram producer’s exclusive right . . . allows him or her to prevent another person from taking a sound sample, even if very short’ (Pelham v. Hütter, C-476/17, 2019, para. 39). The Court left some space for freedom of the arts, considered in EU law to be part of the fundamental right to the freedom of expression, by opining that samples ‘unrecognisable to the ear’ are free from copyright protection. Thus, this quantitatively short sample of two seconds was not de minimis for a qualitative reason, that is, the recognisability of the sampled work. This judgement effectively limits the de minimis concept from permitting uses of a recognisable sample throughout all EU countries, restricting mashups and other recognisable sampling. I note some sampling artists intend for samples to be obscured; as Tricia Rose writes, ‘prior to rap, the most desirable use of the sample was to mask the sample and its origin; to bury its identity’ (1994, p. 73).

A sampling artist relying on de minimis to lawfully sample sound recordings may have better prospects under US than EU case law. In 2016, the Ninth Circuit Court confirmed that the sampling of a 0.2 second horn hit in Madonna’s ‘Vogue’ was de minimis and therefore did not infringe the copyright in Shep Pettibone’s ‘Love Break’ (VMG Salsoul v. Ciccone, 2016, 824 F.3d 871 (9^th Cir.)). The Ninth Circuit went further to remind parties that US copyright statute limits the exclusive rights in sound recordings by permitting the making of a sound recording that independently mimics, imitates or simulates a copyrighted sound recording (17 U.S.C. § 114(b)). However, the Ninth Circuit precedent does not provide sampling artists with reliable access to de minimis in US courts. Because this is not a Supreme Court precedent, courts in other ‘circuits’ (or legal jurisdictional regions under the US federal law) are not bound by this precedent. In fact, the Sixth Circuit Court explicitly rejected a de minimis defence in its 2005 judgement in Bridgeport Music v. Dimension Films with a clear legal rule: ‘Get a license or do not sample. We do not see this as stifling creativity in any significant way’ (Bridgeport Music v. Dimension Films, 2005, 410 F.3d 792, (6^th Cir.), 801). In doing so, it found the sampling of a short, three-note electric guitar riff by rapper N.W.A.’s ‘100 Miles and Runnin’ was not a de minimis use of Funkadelic’s ‘Get Off Your Ass and Jam’.

US courts have never found unlicensed sampling to be fair use

If an instance of unlicensed sampling is not de minimis, a sampling artist may nonetheless argue that the sampling is permitted by a copyright exception. The fair use exception under US copyright statute (17 U.S.C. § 107) and certain exceptions discussed below in EU copyright statute (Information Society Directive, 2001, art. 5(3)) are particularly relevant. US copyright law permits the ‘fair use’ of copyright material, determined by considering four factors (amongst any other factors a court considers relevant): the purpose and character of the use, the nature of the copyrighted work, the amount used in relation to the copyrighted work as a whole and the effect of the use upon the potential market for or value of the copyrighted work (17 U.S.C. § 107).

In our survey and interview research, many of the mashup producers we interviewed claimed fair use permits their unlicensed mashups. Although most producers did not demonstrate detailed knowledge of fair use—that it is a US legal exception, the four fair use factors, court judgements affirming or rejecting a fair use defence—they discussed mashup characteristics that are relevant to fair use.¹⁰ For example, Happy Cat Disco, DJ Earworm and Raheem D argued in interviews with us that their mashups are transformative, which is relevant to the first and fourth US fair use factors.

To be clear, there is no US legal precedent finding any instance of unlicensed music sampling of a copyrighted sound recording to be fair use. The US Supreme Court’s Campbell v Acuff-Rose judgement is the closest that any US court has come to finding music sampling to be fair use (Campbell v. Acuff-Rose, 1994, 510 US 569). In Campbell, the court found that 2 Live Crew’s commercial and parodic remix of Roy Orbison’s musical work Oh, Pretty Woman was not an infringing use, being a fair use under the four factors courts must consider. However, the Campbell judgement gives sampling artists no fair use precedent for sampling of sound recordings, only for the parody of musical works. Whether the sampling of the sound recording was fair use was not an issue in the case, simply because Orbison’s 1964 sound recording was not protected by federal US copyright which protected sound recordings from 1972.¹¹ Even as a precedent for parody of musical works, the Court’s reasoning provides limited guidance. Because the reasoning focuses on the lyrics, which are part of the musical work under US copyright law, it gives little direct insight into what uses of melody, harmony, rhythm, or other common musical features that would constitute fair use parody of an instrumental composition (or instrumental part of a composition).

In fact, no US court—district, circuit or Supreme—has ever found unlicensed music sampling to be fair use. This is confirmed in three studies published between 2008 and 2019 (Beebe, 2008; Liu, 2019; Samuelson, 2009), surveying almost every US fair use case. Law scholar Pamela Samuelson explains, ‘Except in cases involving digital sampling of sound recordings, 285 courts have become more receptive to ‘quoting’ from songs, pictures, and videos’ (2009, p. 2578). Like Samuelson, law scholars Barton Beebe (2008) and Jiarui Liu (2019) do not point to any precedent for fair use of sampling. Some sampling may be fair use but, as law and economics scholar Peter DiCola puts it, ‘the case-by-case nature of fair use renders it unpredictable and expensive, limiting its utility for samplers’ (2011, p. 239). Patricia Aufderheide and Peter Jaszi, two of the strongest fair use proponents in copyright law scholarship, recognise that the existing precedents do not establish when sampling is fair use (2011, pp. 92–93).

As I will explain below, there are reasoned legal arguments in scholarship to find some music sampling to be fair use; here, I simply make the point that none of these arguments find authoritative precedents from US court judgements.

EU limitations and exceptions do not clearly permit unlicensed sampling

Across the Atlantic, EU copyright law has not been more permissive towards unlicensed sampling than US copyright law. Two decades of litigation in German and EU courts over whether a two-second sample from a Kraftwerk sound recording ‘Metall auf Metall’ culminated in the Court of Justice of the EU in 2019. The Court recognised the relevance of freedom of expression and freedom of the arts but ultimately left little space for unauthorised music sampling. The EU Court ruled the German ‘free use’ concept was not consistent with the EU and could not permit any uses of phonograms; instead a sampling artist must rely on a national copyright exception which implements a purpose-based exception listed in EU law (Pelham v. Hütter, 2019, C-476/17; Information Society Directive, 2001, art.5(3)). While many purposes are listed, the ones most relevant to sampling are ‘quotation for purposes such as criticism or review’ and ‘parody, caricature and pastiche’ (Information Society Directive, 2001, art. 5(3)(d),(k)). As intellectual property legal scholars Axel Metzger and Martin Senftleben write, quotation and parody provisions ‘strike a balance between copyright protection and freedom of expression’ (2020, p. 12).

In 2020, the German Federal Court of Justice applied the EU Court’s 2019 precedent to rule that Moses Pelham and Martin Haas’ two-second Kraftwerk sample does not qualify for any relevant purpose-based exceptions in German copyright law (Pelham v. Hütter, 2020, I ZR-115/16 applying precedents from Pelham v. Hütter, 2019, C-476/17).¹² The German court ruled out a parody exception because the sample at question was not humorous and did not mock ‘Metall auf Metall’. Under EU law, parody exceptions require a use to ‘evoke an existing work while being noticeably different from it’ and ‘constitute an expression of humour or mockery’ (Deckmyn v. Vandersteen, 2014, para. 33). Some producers argued their mashups could be permitted under this exception. Producers CFLO, DJ Earworm and Poolboy argued that mashups are parody or satire, and to producer dicksoak, mashups were comedy and ‘inherently mocking’. However, as Jacques explains, it remains unclear whether many instances of mashups and sampling are parody, especially in cases of a commercial purpose or long samples (Jacques, 2016, p. 7).

The German Court also confirmed a quotation exception under EU law—which mashup producers we interviewed and surveyed did not mention and were possibly unaware of—does not permit music sampling in many cases. EU countries can provide an exception for ‘quotation for purposes such as criticism or review’ (Information Society Directive, 2001, art. 5(3)(d)). However, several of the conditions make it difficult for the exception to permit sampling. The sample must be for a ‘purpose such as criticism or review’, be no longer than required for the purpose, in accordance with ‘fair practice’, ‘enter into dialogue’ with the quoted material, make the sampled work identifiable to the listener of the sample and be attributed to the source (Pelham v. Hütter, 2019, C-476/17, 87).¹³ The German Court considered the Kraftwerk sample did not meet these conditions. Ultimately, the EU and German Courts explicitly discussed freedom of expression and freedom of the arts in their judgements, but left little space for unauthorised music sampling.¹⁴

A legal grey zone for mashups

While the US and EU law canvassed here establish several cases of infringing music sampling, there remains a significant legal grey zone for unlicensed mashups. This grey zone exists for several reasons. Firstly, mashups have never been the subject of an EU or US copyright judgement, in spite of the prevalence of unlicensed mashups being uploaded by US and EU mashup producers, and the many instances of blocking and take down of mashups by platforms. This may be surprising, given the willingness of copyright owners to litigate against sampling artists generally, and the common practice of mashup producers not to seek licences for mashups, as our interviews with producers establish.

Secondly, copyright cases in the US and EU law have focussed on rap and hip-hop that climbed commercial music charts, not mashups which generally exist on content platforms but are not on music charts. This is true of the sampling artists who have been the subject of US and EU judgements on copyright and sampling: Beastie Boys in Newton v. Diamond, 2 Live Crew in Campbell v. Acuff-Rose, and Sabrina Setlur, Martin Haas and Moses Pelham in Pelham v. Hütter.

Thirdly, and perhaps most importantly, some copyright exceptions remain untested for unlicensed music sampling. In the US there remains no US case law on whether a mashup or sampling of a sound recording was fair use. As such, US case law on copyright and music sampling cannot be interpreted as precedent that unlicensed sampling is never or even rarely fair use; in this regard, I respectfully differ from music scholar Joanna Demers who claims that ‘For the majority of musicians who appropriate, fair use is dead’ (Demers, 2006, p. 121). Some legal scholars consider the broader case law on fair use to make it likely that some sampling may be fair use. For example, Samuelson argues that the Blanch v. Koons judgement—finding fair use in Jeff Koon’s transformative adaptation of women’s legs from magazine images into a new image commenting on a ‘particular type of woman frequently presented in advertising’ (467 F.3d 244, (2^nd Cir., 2006), 248) —‘bodes well for fair use as applied to transformative remixes and mashups’ (Samuelson, 2009, p. 2554). If a mashup is transformative, there are reasoned arguments for a future court to find it to be fair use. In another example, Schuster and colleagues argue that music sampling should be seen more favourably under the first and fourth fair use factors, based on their comprehensive econometric study finding that music sampling had a positive impact on sales of sampled songs from the US Billboard Hot 100 hits (Schuster et al., 2019).

Many copyright exceptions, including for pastiche and parody, remain likewise untested under EU law. With the Deckmyn guidance on parody, and the introduction-in-progress of the Article 17 parody and pastiche exceptions by EU member states, it remains unclear whether unlicensed mashups would be lawful parody or pastiche under EU law.¹⁵ Noting the two decades for the Pelham litigation to reach a clear CJEU decision on whether unlicensed sampling was lawful quotation, it would be premature to assume whether unlicensed mashups might be lawful parody or lawful pastiche under EU law.

In summary, under both US and EU copyright exceptions, there may still be room for unlicensed but nonetheless lawful mashups. In the US, the existing copyright case law on music sampling simply does not establish the circumstances under which unlicensed music sampling is fair use. In the EU, the Pelham court explained that a particular instance of unlicensed music sampling was not lawful quotation. While music sampling has been the subject of several US and EU copyright judgements, mashups have been the subject of none.

Mashups as expression displaced and foregone

The legal precedents applying copyright to music sampling provide no bright-line rules for online platforms to apply when a mashup is uploaded. With mashups occupying this legal grey zone, this section considers how regulation by online platforms impacted mashup expression. Our research with mashup producers demonstrates that in most cases, both content identification and content moderation have harmed motivations to create and upload mashups.¹⁶

How mashups have become expression foregone

Online platforms not only blocked and took down individual mashups, but also impacted mashups as a form. Through the aforementioned survey of 92 mashup producers and interviews with 30 mashup producers, it was clear that three platforms made it particularly difficult to share mashups and demotivated mashup producers. Two in three (69%) surveyed producers named SoundCloud and one in two (51%) named YouTube.¹⁷ In addition, several producers we interviewed—CFLO, DJ Earworm, DJ Prince, DJ Surda and Poolboy—singled out Spotify. Almost every mashup producer we surveyed (98%) experienced a notice or takedown of their mashup. Three in five (59%) mashup producers experiencing blocking or takedown of their mashups did not dispute the outcome, and the majority of disputing producers were unsuccessful, meaning their mashups were expressions foregone, at least to audiences on that platform. Almost three in five (57%) producers had a user account blocked or deleted by a platform they used, causing multiple mashup expressions to be foregone at once, and causing their producers to lose their connections with other producers and mashup audiences.

Platform regulation of mashups, including both content identification and content moderation, led producers to forego the exact mashup they intended to create and share. In some cases, producers attempted to evade content identification by altering samples, changing pitches of samples in otherwise completed mashups and choosing alternative samples. Our survey with producers confirmed that platforms’ notice and takedown practices changed the content and sound of mashups; one in four (27%) said this was to ‘a great extent’ and a further two in five (39%) said ‘somewhat’. In other cases, the platform altered the mashup expression without producer consent; one in three (35%) producers said their content had been altered automatically by the platform, for example by removing audio from mashups posted as videos. While content identification and content moderation operators are not producers of mashups, they nonetheless have a real and tangible impact on mashups as a form.

Because their mashups were taken down or blocked, producers lost motivation to create and share mashups, leading to ongoing expressions foregone. Over half (56%) of the producers we surveyed felt less motivated to make mashups after experiencing content moderation, and two in three mashup producers we surveyed (66%) had stopped using a platform because of notice and takedown systems. One of these producers (who we refer to in this and other published research as Producer A) started their own platform (which we refer to as Mashup Site A) to provide a safe haven for mashups, and an alternative to SoundCloud and other platforms. However, Mashup Site A itself was taken down by its UK-based domain name registrar when a right holder lodged a complaint under US copyright law; this severed links to the website permanently and displaced Mashup Site A to another web domain.¹⁸

These examples show two ways that mashups displaced become expression foregone. Firstly, when a mashup expression is displaced off a platform, it is gone forever unless someone reuploads it on that or another platform. Secondly, when some producers discouraged by platform regulation stopped making and uploading mashups, future mashup expressions were also sacrificed. The prevalence of foregone mashups should raise concerns about whether content moderation decisions reflect the lawfulness of mashups under copyright law. It seems only appropriate for such a proportion of mashups to be foregone if mashups were likely copyright infringement. However, given the significant gaps in US and EU case law on whether unlicensed music sampling infringes copyright, it is premature to assume that mashups are likely infringement.

Legal safeguards for copyright exceptions are ineffective

The prevalence of expression displaced and foregone also suggests that EU and US legal safeguards created to ensure that online platform users can rely on copyright exceptions have been ineffective. For example, US copyright ‘safe harbor’ provisions, introduced by the Digital Millennium Copyright Act, punishes people who fail to consider exceptions before claiming infringement in a takedown notice (17 U.S.C. § 512(f)).¹⁹ In a dispute about a YouTube home video with a Prince recording playing in the background that was posted by a mother and taken down after a copyright notice from Prince’s music publisher, the US Ninth Circuit ruled ‘Copyright holders cannot shirk their duty to consider — in good faith and prior to sending a takedown notification — whether allegedly infringing material constitutes fair use’ (Lenz v. Universal, 2016, 815 F.3d 1145 (9^th Cir.), 1152). As the previous section noted, there remains no US case law on whether a mashup or sampling of a sound recording was fair use. Nonetheless, under the US Lenz precedent, copyright owners have a responsibility to consider whether each upload is fair use before issuing a takedown notice. Likewise, EU countries were required by 7 June 2021 to ensure users can rely on copyright exceptions on ‘online content-sharing services’ such as YouTube and SoundCloud (Digital Single Market Directive, 2019, art. 17 and recital 70). Specifically, EU countries are now required to ensure users of platforms can rely on copyright exceptions permitting quotation, criticism, review, caricature, parody and pastiche. EU countries are also required to ensure that platforms put in place ‘an effective and expeditious complaint and redress mechanism that is available to users of their services in the event of disputes over the disabling of access to, or the removal of’ uploaded content (Digital Single Market Directive, 2019, art. 17(9)). While a handful of EU member states have already implemented these requirements, most have yet to do so.

The widespread takedown, blocking and displacement of mashups shows that, in practice, these safeguards for copyright exceptions have been difficult for platforms to implement. One reason for this lies within the laws themselves. Platforms face increasing legal liability for failing to take down and block uploads that potentially infringe copyright. This legal liability increases the chance that platforms will remove or disable access. US and EU copyright laws create legal liability for platforms that do not expeditiously remove or disable access to content uploaded in infringement of copyright. In the US and EU, there can be liability if the platform does not act when they receive a valid copyright notice or obtain knowledge of infringement (17 U.S.C. § 512 ; e-Commerce Directive, 2000, art. 14; Digital Single Market Directive, 2019, art. 17). As legal scholar Aleksandra Kuczerawy puts it, ‘refusal to take down content puts the intermediaries at risk of being held liable. Obviously, the most cautionary approach is to act upon any indication of illegality’ (Kuczerawy, 2020, p. 527).

Another reason is the ubiquity of content identification, which are the automated systems which detect when copyright material has been uploaded to an online platform.²⁰ For example, YouTube uses Google’s in-house Content ID system, and SoundCloud, Facebook and Twitch use Audible Magic’s content identification system. Music rights holders use content identification services like Zefr and Pex to further scour online platforms for uploads that contain copyrighted content, driving efforts to generate revenue and issue takedown notices.²¹ This greatly amplifies the reach of content moderation to find uses of copyright material, but without amplifying the ability to discern the purpose of uses that are essential for determining whether copyright exceptions apply. Currently, there are few purpose detection tools—which aim to automatically discern the purpose of an upload—to help automate copyright exceptions at the scale of online platforms.²² One mashup producer we interviewed, Raheem D, called out the lack of such tools on YouTube: ‘YouTube, it doesn’t really take into account fair use . . . Because, obviously, it’s an automated system. The system can’t know if it’s a parody, or if it’s transformative, or if it’s a mashup’. Raheem D’s claim is well-founded and confirmed by the platforms themselves. At recent public stakeholder meetings about forthcoming EU copyright law reforms, YouTube said:

because they are machine, they need also human [sic] to understand the context. That’s something that I think that in science is experimented, they are trying toward doing it like the law but we are not there yet, at least at the scale that we need to be managing a system like YouTube (European Commission, 2019a).

Facebook conceded ‘Our matching system is not able to take context into account; it is just seeking to identify whether or not two pieces of content match to one another’ (European Commission, 2019b, n.p.). Audible Magic—which conducts content identification for Facebook, SoundCloud and others—admitted:

Copyright exceptions require a high degree of intellectual judgement and an understanding and appreciation of context. We do not represent that any technology can solve this problem in an automated fashion. Ultimately these types of determinations must be handled by human judgement (European Commission, 2019b).

Lacking purpose detection to bring copyright exceptions to platforms, some platforms turn to human moderators to identify when copyright exceptions might apply.²³ However, human moderators provide limited assistance to mashup producers. They do not prevent a platform from taking down or blocking in the first place, because human moderators typically only review uploads if the uploader disputes the takedown or blocking. Because most mashup producers informed us that they do not dispute takedowns or blocking, as the previous section discussed, human moderators often do not have the opportunity to intervene before a mashup is taken down or blocked.

Conclusion

Throughout this article, I have used the term ‘expression foregone’ to make clear that the copyright regulation and content moderation of mashups has failed in many instances to provide the incentive for mashup producers to create these ‘expressions’. In both US and EU law, one of the central purposes of copyright is to provide an incentive for the creation of (and investment in) creative expressions (Constitution of the United States of America, art.I, §.8, cl.8 and amendment I, Information Society Directive, 2001, preamble paras. 9-11). It is clear that copyright regulation and content moderation push mashup producers to forego both their imagination, as Pappalardo and colleagues describe, but also their expressions, thereby failing the incentive to create. Content identification and content moderation of mashups are a prime example of what a United Nations Special Rapporteur for freedom of expression and opinion highlighted: ‘Demands for quick, automatic removals risk new forms of prior restraint that already threaten creative endeavours in the context of copyright’ (UN Special Rapporteur, 2018, p. 7). While empirical studies may show that, generally, ‘the notice-and-takedown regime is working’ (Erickson & Kretschmer, 2019, p.16), our empirical research draws attention to specific, negative impacts of such regimes on mashups and mashup producers.

In this article, I have set out how mashups become expressions displaced and foregone. Firstly, while there are no US or EU judgements on unlicensed mashups, the existing judgements on unlicensed music sampling do not clarify whether unlicensed mashups are lawful. Under US copyright law, it is unclear whether unlicensed music sampling is fair use. Under EU copyright law, it is unclear whether unlicensed music sampling would be lawful parody or pastiche. In light of unclear legal authorities, unlicensed mashups find themselves in a genuine legal grey zone.

I explained how content moderation has been applied on platforms without sufficient regard for copyright exceptions. Unlicensed music sampling has been blocked and taken down on platforms on the basis of content identification, not on the basis of copyright exceptions. Platforms and content identification firms admit that they cannot distinguish whether particular uploads qualify as copyright exceptions. As the first-hand accounts of mashup producers from our empirical research confirms, platforms typically treated unlicensed mashups as infringement, despite unlicensed music sampling being a legal grey zone. Discouraged mashup producers abandoned not only these wrongfully blocked and taken down mashups, but also the possibility of making mashups in the future. Ultimately, both mashups and their producers were displaced off platforms.

The displacement of mashups poses difficult issues for copyright policy and content moderation to address. Foregone mashup expression is often impossible to replace because mashups make timely comments and interventions in popular culture and discourse, and creators blocked or taken down decide not to create again. Overcoming shortcomings in copyright law and platform regulation does not restore the moment lost for creation and conversation. Moreover, expression foregone is difficult to detect because the absence of expression is not proof of expression foregone. Without other evidence—such as the empirical research my colleagues and I have conducted, with the trust and cooperation of mashup producers—mashup expression and creator motivation may be suppressed in silence. Still, understanding how expression is foregone is a worthwhile pursuit, not just in mashups but other online forms. As mashup producer BringMeTheMashup lamented in his interview with us, ‘if mashups were taken away I’d lose my own creativity, my own expression’.