Serendipity on radio and streaming: Between musical discovery and recognition

Introduction

Billions of people tune in to radio stations and music streaming platforms in search of a soundtrack for their daily activities. Listening takes place with different levels of attention and engagement. A person who listens to his or her favourite radio station may be looking for familiar songs and, occasionally, some musical discovery. On streaming, that same listener may be looking for a new album by his or her favourite artist or a playlist to boost concentration. On radio or streaming, the concatenation of thousands of tracks provides diverse experiences, which the audio media industry increasingly seeks to capture and drive. Among these experiences, what we are interested in discussing in this article is serendipity, the fortuitous discovery that (as we advocate) mobilizes affections and helps to build bonds – either with broadcasters or with platforms. Along the way, we will address similarities and differences between music programming on radio and streaming.

The personalization of discovery, leading to an experience of serendipity, is a kind of Holy Grail for the media and computing industries. Large companies, such as Google, have publicly expressed the ambition to build ‘engines of serendipity’ (Morrison, 2010). Music streaming platforms seek to map tastes, listening profiles and music consumption trends to expand and retain their user bases, optimizing their financial results. Digital platforms have taken on a growing role in the music industry, surpassing, as of 2015, revenues from physical sales (CDs, DVDs, etc.). ¹ By the end of the decade, there were 443 million subscribers to music streaming worldwide (International Federation of Phonographic Industry [IFPI], 2021), pointing to a profound cultural shift from an anti-establishment activist logic of online music sharing to one disciplined by consumer relations. ²

In this context, the major record labels rebranded themselves as music companies and detached themselves from their previous role as providers of physical musical products (Herschmann, 2010). As owners of the intellectual property of millions of phonograms, labels started to profit from the licencing of their catalogues to music streaming platforms, in a strategy that included acquiring minority stakes in these new players in the digital music environment (Music Business Worldwide [MBW], 2018). ³

In this article, we explore the operationalization of serendipity in key points of music circulation history through audio media: radio and streaming. Based on exploratory bibliographic research, our goal is to question how and why serendipity stays at the centre of current streaming platforms’ rhetoric and musical culture in general, and call attention to the lack of scholarship on the topic in media studies. By comparing the music programming approaches in both media, we discuss how serendipity articulates discovery, memory and recognition in the current push towards personalization in music streaming platforms. We start by delineating a brief genealogy of the concept of serendipity. Then, we discuss the relationship between memory and the serendipitous affect and how audio media language articulates recognition as an important part of music circulation. Finally, we compare the aspects of music curation and recommendation that these media use to operationalize serendipity as a musical consumption experience.

We articulate several Ibero-American references that have been standing out in radio studies and audio media in general but often go unnoticed by the leading scientific journals. Breaking this silence and expanding the reach of these accounts in a decolonial perspective is a secondary goal of the reflections we present here, which hopefully will reach researchers from different countries equally affected by issues of music consumption on radio and streaming platforms.

The ideas we have systematized here do not involve the pretension of building a global, comprehensive History of sound media. We understand that, just as we have many radios, in plural, there are also many radio Histories, with local and national specificities. And, although music streaming has been platformized since the birth of the industry, we understand that the same applies to this segment, which should be understood from different regulatory realities that condition access to these services.

Serendipity origins

Some words travel across time and space in peculiar ways. It is the case of the neologism ‘serendipity’, registered by the Cambridge Dictionary as ‘the fact of finding interesting or valuable things by chance’. The etymology refers to Serendip or Serendib, from the Arabic Sarandib, the ancient name for Sri Lanka. The word was first registered in English in 1754 by the writer Horace Walpole in a letter to a friend in which he quotes the tale The Travels and Adventures of the Three Princes of Sarendip. In this Persian story, the young princes embark on a challenging journey, a passage rite, in which they end up making unexpected discoveries.

Although, in the tale, the princes’ performance in solving problems is due more to their sagacity and observation skills, over the years, the word has been resignified to denote fortuitous and happy findings. In 1943, the derived expression ‘serendipitous’ emerges, a property attributed to people contemplated by luck or to the discoveries themselves.

In the second half of the 20th century, serendipity evolved from an obscure expression (‘esoteric, not to say mysterious’, notes Merton in the introduction of his book ⁷ ) to an almost scientific cliché. Many technological innovations, from penicillin to non-stick cookware, result from casual and happy discoveries. You look for something in a laboratory experience, and you find something completely different but of great interest, something that becomes valuable for solving a problem or a need.

The term ended up being used in a generalized way to designate all kinds of discoveries, evoking a somewhat fanciful narrative, which puts the individual genius, the scientist who screams ‘eureka!’, above public policies for scientific, technological and innovation development.

In recent years, though, the concept of serendipity has been re-elaborated in different areas of knowledge, such as in information science (see Foster and Ellis, 2014; Makri et al., 2014; Vechiato and de Farias, 2020; Smets, 2022 among others). Makri et al. (2014) interviewed 14 content creation professionals to understand how they aspired to design digital environments to achieve serendipitous experiences. Smets (2022), in a theoretical essay, seeks to map ‘intents to design for serendipity’.

Serendipity is a factor considered even in studies focused on urbanism, which seek to verify whether the growing datafication and information and communication technologies in the smart cities that are being developed will leave room for fortuitous encounters (Sauer and Copeland, 2021).

There is a growing interest, especially among computer scientists, on the practical applications of the concept of serendipity, which results in a huge volume of papers related to the offer of music on streaming platforms. However, what interests us in this article is a specific appropriation of the word by media industries, including radio and music streaming platforms, and its connection with memory and recognition.

The state of the art on correlated research makes explicit the commercial approach of most of the investigations. There is a tendency in these studies to regard serendipity as a moment, as a cliché of fortuitous discoveries, which should be engineered or designed, part of a media plan. However, we must understand the serendipity phenomenon, as McCay-Peet and Toms (2010) suggest, in a holistic way, as a process, instead of a moment of insight. The authors offer a less mechanistic approach to the concept and suggest that different triggers can activate serendipitous experiences.

It is still a minority perspective, but it is already possible to state that a part of the studies on serendipity are already starting to move beyond clichés. In this article, we seek to follow this trend, bringing a contribution to the field of audio research.

It is puzzling, though, that we did not find a single scientific article that takes as its axis the discussion of serendipity in radio, within the scope of media studies. We hope we will be able to address some of these issues.

Memory rearrangements

If the concept of serendipity applied by AI-based RS depends deeply on novelty in tension with relevance, we highlight that, for musical listening, another form of affectivity is also able to promote this encounter. We are talking about a rediscovery, the encounter with an already known song, but that proves to be appropriate to the listening context.

Serendipity may reveal itself as a fortuitous reencounter when listening to a song that we know, evoking emotions from another time. It is not necessarily a nostalgic experience, the pain of a lost time often idealized – although we understand there are different modalities of nostalgia (Pickering and Keightley, 2020). The radio flashback offers a kind of map of personal experiences, in which music emerges as a soundtrack of a past that does not convey any specific meaning but for which we build (never rigid) new meanings. A hit from decades ago can be resignified from an individual perspective, serving as a trigger to rearrange memory.

The past is a relevant business for the music industry and broadcasting. In the United States, between 2014 and 2018, Classic Hits and Classic Rock formats alternated as the fastest growing segments in the radio market (Nielsen, 2018). In 2021, Brazilian broadcasters in the Adult Contemporary format were among the most listened to on FM and web radios. ⁸ As a significant actor in the audio media ecosystem, radio evokes and rethreads individual and collective experiences like a show we attend.

Live music performances are memory-setting events and support the negotiation of meanings and establishment of identities. The proportions of a concert tell us almost nothing about the impact it can have on our symbolic project of constructing the self (Thompson, 1995). A concert for 20 or so in a small independent nightclub can be as striking as a famous band show at a mega-festival with an audience of over 200,000 people.

Michael Pollak (1992) emphasizes that the constitutive elements of memory encompass, first, the events experienced personally, but also those experienced ‘second-hand’ by the group or collectivity with which the individual builds bonds, an idea of belonging. Through socialization, political or historical, a phenomenon of projection or identification with a particular past may occur, resulting in an ‘almost inherited memory’:

These are events in which the person did not always participate but which, in the imaginary, have taken on such importance that, in the end, it is almost impossible for them to know whether they participated or not. If we go further, to these events indirectly experienced are added all the events that are not situated within the space-time of a person or a group. (Pollak, 1992: 201)

Of course, memory is selective, a construction, a permanent effort of organization of experiences, emphasizing some and forgetting others. Pollak stresses that ‘memory is a constituent element of the sense of identity, both individual and collective’ (Pollak, 1992: 204, emphasis in the original). Identities are equally negotiated and should not be understood from an essentialist perspective. We gather memories that help us to build identities that integrate more or less coherent narratives of the self.

Musical scenes constitute collective experiences that often transcend physical borders. Memories of old concerts, which we attend (or not), are triggered as a factor of social distinction, as testimonies of a historical moment, a milestone, that evokes ‘happy memories’, a vow of fidelity to an often idealized past (Ricoeur, 2007: 502).

Herom Vargas and Mozahir Salomão Brück (2017: 237) refer to this notion to discuss the idealizations of memories around the bossa nova, a Brazilian popular music genre. ‘The “happy memory” emerges, therefore, moulded in the comfort and desire of the memorialistic activation and in the claim fed back by the passage of time whose enunciations are arranged according to how pleasant they are to repeat’. ⁹

The fact that the reunion with a musical success from the past escapes the novelty requirement can be addressed by a third component of serendipity considered by RS: the surprise. In recommender systems, according to Denis Kotkov et al. (2016: 182), surprise or unexpected recommendation is expressed by the presence of an item that differs from the profile without its novelty being considered. As Octavio Paz (1984: 19) notes, in a classic essay on tradition and ruptures, novelty and unexpected are related terms but not equivalent; novelty may contain continuity, and offer new and surprising combinations of the same elements.

Therefore, serendipity is one of the possibilities of RS’s orientation, composed of a tension between personalization and discovery. It works by substituting a set of emotional responses that is impossible to measure by the relationship between these formalized characteristics associated with the phonogram. The tension will be more evident once we compare programming processes with platforms’ recommendation.

Resignifying recognition in audio media

Critical theory elected radio, the primary mass medium in the 1930s–1950s, as a priority target. Radio music was seen as disposable merchandise and a pastiche of noteworthy artistic manifestations. For Theodor Adorno (1991), recognition is a symptom of the regression of listening, a response to the organization of mass popular music around formulas, tested and repeated to exhaustion. He identifies this regression process when discussing the internal musical characteristics, its relationship with the commercial demands of distribution and the social and psychological effects these technical and aesthetic impositions would cause in the population.

Commercial interests would favour serialization and standardization: reduction of costs, increase in the number of goods produced, capacity for dissemination and promotion. In a society of commodities, the human need would only be a side effect of the search for profit and that ‘tends to the concentration of capital’, which produces more standardized products (Adorno, 1945). Technical formats, both radio broadcasting and recording standards, are developed to match these purposes, as well as aesthetic formats, in a rational process of industrialization in which there is no difference between the resulting products, be they automobiles, household appliances, films or songs (Adorno and Horkheimer, 1985).

For the author, the regressive listening would result in a paradox: the mass audiences would be angered by having their intelligence underestimated by the constant repetition. However, they would surrender to the pleasure of standardized music as a regression to the comfort of an infantile listening phase associated with lullabies. The combination of these two perspectives, aesthetic and psychoanalytic, results in the Adornian evaluation of mass music. The author refers to mass music as the apex of infantilization of the listening of a population incapable of reacting to industrial impositions, even in the face of immense disgust for the simplification and vulgarization of musical creation.

However, we call attention to the value the author puts on musical creation as a symbolic means of human emotional development and the interesting contribution to the material dependence of ideas in musical artwork. Alison Stone (2016) argues that these contributions are valuable for studying musical aesthetics, even if we are aware that Adorno ignores the creative possibilities of the mass musical assemblage, which is also dependent on the historical/material context. In this sense, we consider recognition not only as a simplifying force of aesthetic experience but also as a factor dependent on socially mediated listening conditions and as a trigger for memory, revesting melodies – not rarely simple, repetitive – of affective memories.

From a very different perspective, Ricoeur (2007: 502) classifies ‘recognition as the little miracle of memory’. We understand here that the cognitive impact of listening to a song that somehow marked our lives or a specific collectivity is similar to the reunion with our loved ones or the experience of leafing through a photo album. ‘All memory-making thus sums up in recognition’, Ricoeur (2007: 502) points out.

In the following two subsections, we will discuss the characteristics of music discovery (and rediscovery) on radio and streaming.

Radio music and streaming: serendipity from discovery and rediscovery

We review, in this section, how audio media operationalized the notion of serendipity as resulting from novelty, discovery and recognition in music prescription, either non-explicitly by music radio programming or as central rhetoric on music streaming recommendation. In these media, serendipity becomes a pursued value because it can be understood as stemming not only from new encounters but from recognition and memory. Since media language and technology are highly dependent on self-references, pre-established formats, intermedial connections and reconfigurations of creative practices, different media can present themselves as tools for engineering serendipity. For music circulation, radio and platforms share the playlist format (Ferreira, 2020) and its medialities, ¹⁰ and they both deal with the chance encounters that feed off discovery and rediscovery.

Radio programmers do not explicitly defend their role in serendipity. On the contrary, they usually adopt radio programming philosophies (Ahlkvist, 2001), claiming for themselves the role of cultural mediators rather than designers of serendipitous experiences. Possible chance encounters might come from the random combinations of the medium (self)references and cultural experiences any person or group might have. While streaming platforms resort to serendipity to aid in its necessity of captivating an audience for data register and processing, chance encounters would come as users discover new modes of experience that encourage them to stay on the platform.

Tensions between promotion and creativity in music radio programming

Research about radio programming traditionally takes serendipity as its self-evident result (Villermet et al., 2021) which would supposedly dispense further reflection. However, new music discoveries and rediscoveries appear at the centre of the music and recording industries’ business models. In this sense, justifications for radio resulting in a special kind of listening cannot be neglected because it served as promotional tool within commercial partnership with the recording industry (see among others, Gambaro et al., 2018; Kischinhevsky, 2011b), while also propping up the creative aspects of curation, wherein an educational or artistic expression guided the possibilities of taste building and expansion.

Major record labels based their business models throughout the 20th century on an industrial logic of permanently commercializing novelties, even though most of these releases explored success formulas already tested (De Marchi, 2016; Vicente, 2014). Especially in the second half of the century, phonographic and radio industries developed in a symbiotic way (Kischinhevsky, 2011a): record companies released music, now in Long Play (LP) format, with greater storage capacity and higher fidelity reproduction, and radio aired it, providing a laboratory for experimentation with popular taste. Live performances and dissemination by radio stations were primarily aimed at boosting record sales, the primary source of revenue for record companies and artists. In this scenario, the artistic director, then the DJ and, finally, the music programmer, each with their specific labour conditions, become critical actors in the construction of a continuous flow of music that seeks to arouse identification and build an environment conducive to consumption – of the advertising conveyed by the station as well as the radio itself, understood as a social institution (Gambaro, 2019).

We observe a tension between commercial and creative programming in two key moments of the shifting prescriptive roles of music radio: first, the organization of the growing flux of releases, whether by charismatic DJs or anonymous programming directors; and second, the editors or curators’ management of new participatory audiences. These tensions reveal the breaches for serendipity, as they promote a wide range of approaches to programming, all dependent on subjective decisions.

Programming or curating a radio or streaming playlist entails knowledge about musical conventions, aesthetics and genres. Music radio programmers and streaming curators classify songs into more or less specific categories, evoking music genres established in common sense and considering items such as tracks’ tempo (slow, medium, fast), melodic line, feelings (minor chords evoke sadness; diminished chords induce tension, discomfort), release date (in stations targeting young audiences, for example, a hit from 3 years ago can already be understood by listeners as a flashback), the relevance of the artists to their target audiences, novelties and so on. As Simon Frith (1996: 94) reminds us, genres are not static, they are permanently shifting, being negotiated and resignified – just like our memories and identities.

Musical genres operate between two poles: innovation, which is necessary for the music industry but eventually disorienting for the audience, who may react negatively; and consolidation, which provides some stability and predictability, associated with repetition and recognition. The role of programmers and curators is to creatively balance these two extremes in a coherent listening experience. As the discovery of a new song (or a different version of a track one already likes) is intrinsic to the act of listening, shifting priorities could frustrate the listeners if the programming does not bring surprises, limiting itself to repeating tracks and artists recognized by the public and/or specialized critics. The excessive engineering and quantification of listening stats towards commercial goals could hinder serendipity, as it curtails an artistic musical messaging. In this sense, curation assumes that a good musical experience is based on these fortuitous encounters only made possible by maximizing a qualified, creative offer for this type of experience – something threatened by commercial radio programming based on the repetition of successful formulas.

Before the 1950s, music radio programming was dominated by the DJs, who branded time on air with their voices and songs’ selection. Harnessing the power of an exclusive medium for attention, these DJs created close relationships with advertisers, record companies and artists, bypassing their stations. The emergence of the format radio playlist characterizes the development of a curatorial and filtering work in the face of the constant expansion of the catalogue of available recordings and the limited time for dissemination. Thus, the introduction of novelty in programming became not only motivated by a commercial logic of circulation of new products but also by a non-explicit but present search for serendipity in the face of the new experiences that radio could provide.

The novelty in constructing the radio playlist attested to the DJ or programmer’s ability to manage an ever-growing offer of tracks. During radio hegemony, between 1930 and 1950, the medium was the main source of musical novelty. In this period, the professionalization of programming directors, with the development of different programming techniques, especially that of Top 40, served to organize the control of musical prescriptions in the stations.

However, the creation of the playlist format, which was based on a technical discourse of selection and no longer on the creativity and personality of the DJ, did not address the question of novelty, as it produced an even more pronounced limitation for music dissemination (Brewster and Broughton, 2014). If before, the controversy was the excessive power of DJs and the lack of transparency in music selection, now the issue is the impersonal professionalized selection. How can it promote an emotionally satisfying musical experience, given that programming directors are not artists of music selection like DJs? These professionals would be concerned with building audience figures and solidifying partnerships with the recording industry, caring little for a musical messaging that would provide the audience with the best music for the moment (Denisoff, 1973; Gallego Pérez, 2015).

Eric Rothenbuhler (1985), Jar A. Ahlkvist and Robert Faulkner (1995), Ahlkvist and Gene Fisher (2000) and Ahlkvist (2001) demonstrate, however, that the cultural process of music programming did not, at that point, translate itself to an impersonal practice. In studies that sought to understand the role and motivations of radio music programmers, these authors argue that programming directors use knowledge bases about the music, the audience, the radio station and the brand they represent to develop strategies, or philosophies, to select and organize what they offer to the listener. Moreover, these professionals had different levels of freedom in alternative stations, such as college radios or FMs which sought to distinguish themselves from the AM Talk and News format.

In this sense, programmers would promote novelty in a planned context of discovery, involving an imagined audience that has expectations about what they will find in the programming. But this is an audience whose serendipity is never taken for granted, since listeners’ consumption practices are not explicitly related to music and since radio music histories constituted a background that programmers interpreted according to their own agenda. The logic of scarcity of time for promotion and the tension with the public’s preconceived demand for novelty will provide the centre of the attention focused on serendipity in radio. The medium was treated as the promoter of musical emotion, sense of community and taste formation, and at the same time as the spearhead of the recording industry’s marketing efforts. This tension will be relieved with the radio spectrum expansion, especially with the digitization of access to phonograms. The number of stations increases tremendously, and novelty becomes more individualized.

According to Friedrik Stiernstedt (2015), audience participation has always been highly relevant for radio communication. Nevertheless, this new context allows for new forms of reception, interaction and participation that have transformed programmers and DJs into moderators who select and assemble programming partially produced by the audience itself, mainly through the accumulation of behavioural data and choices of these listeners in social networks and interactions with the station. Stiernstedt believes the automated music programming reduced the intensity of the work of editors, who only needed to organize a playlist and insert interventions throughout the programming.

This kind of programming, however, created demands of social network management and interaction with the public, profoundly changing the way prescription operated: from a relatively tight professional and creative attitude that informed music selection to mediation between programming and the public, with the radio programmer becoming the creator of a coherent text between the station’s identity and the demands and content produced by the public itself. These new kinds of audience participation will promote a new tension between demands for personal experiences and the logic of prescription in streaming platforms.

Tensions between personalization and discovery in streaming services

Music streaming recommendation advances the centrality of the listener in music programming. Commercial interests are still relevant, but discovery is employed to guarantee retention, specifically user data gathering and processing, instead of a communicative and commercial act that can result in serendipity. Platforms work on a stimulus–response logic of inciting a psychological pleasure, hopefully of serendipity, guiding listening behaviour. As Robert Prey (2019: 3) notes, streaming platforms organize music ‘around listener-centered descriptors’, revealing a ‘functionalist’ and ‘behaviourist’ perspective ‘that regards music as both accompaniment and actant: that is, background to our everyday routines, amplifier for experiences and enabler of moods’. While the ‘Fairness, Accountability and Transparency’ (FACT) community from Computer Science is dedicated to challenging homogenizing features of recommendation (see Ekstrand et al., 2017; Epps-Darling et al., 2020; Ferraro et al., 2021), serendipity still seems mostly dependent on personal recognition, not of genre, past listening or musical conventions, but of some psychological attributes of taste exciting a biological response (De Marchi et al., 2021). A happy encounter with a song is thus dependent on what the system will consider as an unknown or forgotten song, but that has features that can please the user, considering their consumption history and track ‘family trees’ (networks of musicological links between artists, songs and genres mapped by music streaming platforms). The challenge for RS is to balance its main metrics of operation. The prediction of listening to a song is based on past listening behaviour, among which is the history of playback and evaluation, with the introduction of tracks for which main prediction metrics do not exist since they have never been played by the user.

RS evaluate the serendipity linked to discovery according to their ability to predict the user’s inability to recognize a track, relegating the accuracy of listening prediction to the background. According to Kotkov et al. (2016: 182), this means adding to the dissimilarity of the song with the user’s profile, an intrinsically social feature of the track, its ‘unpopularity’. The reasoning is that songs that are more popular and similar to those in the listener’s history are more likely to be recognized. Thus, the mathematical formalization of RS limits novelty into measurable aspects as it attempts to build predictability. However, platforms do not capture an individual’s entire musical experience or background. Music listening is not limited to one service or channel, so previous listening practices available to the system do not accurately represent a listener’s cultural repertoire.

It would be easy to offer a long playlist randomly built from the millions of tracks available in the streaming platforms catalogues, with high chances of providing novelty. However, as we pointed out, these systems justify their existence with their ability to predict a personalized and specific listening experience for each user. The simple discovery of a new song is not enough; serendipity requires the emotional results of identification or the positive affect of the encounter. This balance between novelty and relevance makes serendipity a difficult concept to apply to mathematical prediction. Eric Drott (2018) observes that platforms use their curation and recommendation to create constant anticipation of ‘the next song’, building a short-term listening experience and modulating users according to thematic contexts. According to Drott, the main difference between RS and traditional radio’s subjectification of listeners would be that RS are dependent on the individual user and music data, asymmetrically concentrated by the platforms. As these platforms rely on the attention to the selection process instead of the listening experience, the surprise necessary for serendipity seems more dependent on the anticipation than the actual listening. Jeremy Morris and Devon Powers (2015) also argue that the ‘digital music commodity’ promoted by platforms focuses more on control than personal ‘branded musical experiences’ and more on modulating user experiences than on establishing musical audiences as music radio did. Serendipity, in this way, is measured by the effectiveness of influencing listening behaviour.

The results of this new tension between personalization and discovery ripple through artists’ actions and music circulation. To promote the recognition necessary for personalization while catering to the ‘new’ experiences that RS seeks to create, a large number of established artists have been investing in reinterpretations of their records released two, three or more decades ago, taking advantage of ephemerides to capitalize on tours, as well as tribute records, in which an artist presents versions of the work of others. The further dependency of the recording industries on faster novelty cycles added to the permanent need for promotion of artists through regular releases, and finally, music streaming platforms need to keep users involved, favour these rereadings, revisiting a successful catalogue endorsed by the public and critics. And these songs generate an effect of recognition, which can cause an experience of serendipity. We can identify this phenomenon with the considerable increase of attention that the so-called catalogue songs have received on platforms. According to a report by MRC Data and Billboard (2021) in the United States, streams of newly released songs – less than 18 months ago – fell by 3.7% for the first time in the history of music streaming year-on-year. Meanwhile, listening to catalogue songs released at least 18 months ago reached 70% of the entire volume of streams, a growth of 19.3%.

Therefore, music streaming platforms’ recommendations seem to equate serendipity with their ability to affect users’ listening behaviour, responding to the tension between personalization and discovery. In this way, they can create measurable parameters for the effectiveness of discover and rely heavily upon their engineering power to justify their existence, exert influence over musical experiences and create value for their product. Since music listening, in this sense, is somewhat data driven, they can equate this information to their ability to promote individual discovery – based on recognition – which would, in this rhetoric, cause serendipity.

Concluding remarks

In this article, we discussed the concept of serendipity within the scope of radio, audio media and memory studies, emphasizing the role of human agency in musical discovery. We advocate the importance of recovering the ability for curation (Kischinhevsky et al., 2021), even if conditioned by the increasing use of artificial intelligence in digital platforms, as well as automation software in radio. We understand that this is just a step in debating the role of happy findings in radio listening or platforms beyond algorithmic taste (De Marchi et al., 2021) without incurring reductionist approaches.

It is essential to highlight that serendipity is always relational, and depends on the sociocultural background of each individual, even if symbolic references of wide circulation, often elaborated as collective memories, contribute to its occurrence. Listening to the track The chain, by Fleetwood Mac, in the film Guardians of the Galaxy vol. 2, evokes different sensations for those who lived in the 1970s and those born in the 2000s. Audiences over 50 recognize the music from one of the most listened-to rock albums of all time, Rumours, released in 1977, with over 45 million copies sold since then (Runtagh, 2017). On the contrary, those born in the 2000s probably associate the song and its thrilling crescendo with the most important scenes of the Marvel Universe blockbuster movie. ¹¹

The personalization of music recommendation, instead of celebrating the aesthetic tensions of recognition and novelty, values the non-reflective decrease of the risks of the encounter. It is not that it is incapable of promoting fortuitous encounters. Yet, it has difficulties in perceiving that the cultural direction and associations of the curator are communicative acts and that, therefore, always carries a specific dose of uncertainty. Serendipity can be found in this uncertain space of communicational exchanges.

The idea of serendipity, so mechanically pursued by computer science, opens doors for us to think about critical issues for musical consumption, the conditions of listening, memory, identity, representation, the affections triggered and audio media agency. These are central themes to address when focusing on the restructuring of the music industry, increasingly intertwined with information technology and telecommunications, in a context of platformization of communication and culture.