ArticlePDF Available

Balancing Categorical Conventionality in Music

July 2022
American Journal of Sociology 128(1):224-286

July 2022
128(1):224-286

DOI:10.1086/719937

Authors:

Daniel Silver

University of Toronto

Show all 5 authorsHide

Research on the relationship between categorical unconventionality and popularity has produced mixed results. While many accounts sug- gest that unconventionality is penalized, much sociological theorizing indicates that success comes from a delicate balancing act between con- ventional and unconventional offerings. Using data on the genre self- classifications of over 2 million musicians and bands across the United States, the authors find broad support for this balancing act. Yet the shape it takes is also conditioned on local contexts, across both high-order complexes of musical genres and geographic space. The authors high- light the local metropolitan area characteristics that shift the relationship between unconventionality and popularity. They also create a typology of cities based on how their unconventional offerings are rewarded and punished. An online visualization tool enables further investigation of these relationships. The authors close by proposing an agenda for how to study local heterogeneity in the relationship between unconventionality and popularity.

Content uploaded by Daniel Silver

Content may be subject to copyright.

Balancing Categorical Conventionality in Music

Daniel Silver and Clayton Childress

University of Toronto

Monica Lee

Facebook

Adam Slez

University of Virginia

Fabio Dias

University of Toronto

Research on the relationship between categorical unconventionality

and popularity has produced mixed results. While many accounts sug-

gest that unconventionality is penalized, much sociological theorizing

indicates that success comes from a delicate balancing act between con-

ventional and unconventional offerings. Using data on the genre self-

classiﬁcations of over 2 million musicians and bands across the United

States, the authors ﬁnd broad support for this balancing act. Yet the

shape it takes is also conditioned on local contexts, across both high-order

complexes of musical genres and geographic space. The authors high-

light the local metropolitan area characteristics that shift the relationship

between unconventionality and popularity. They also create a typology

of cities based on how their unconventional offerings are rewarded and

punished. An online visualization tool enables further investigation of

these relationships. The authors close by proposing an agenda for how

to study local heterogeneity in the relationship between unconvention-

ality and popularity.

In 2007, before being nominated for six Grammy Awards and rising to inter-

national stardom, Janelle Monáe was a weirdo. In fact, in our data on over

Silver and Childress are coequal ﬁrst authors. For helpful feedback on this work, the au-

thors thankAndre Boutyline, ChadBorkenhagen, the Gangof Seven (Anne Bowers, Laura

Doering, Christopher Liu, Sida Liu, Kim Pernell, András Tilcsik), Peter McMahan, Craig

Rawlings, and Vaughn Schmutz. Prior versions of this work were presented at the 2017

American Sociological Association Annual Meeting and the 2018 Social Science History

Association Annual Meeting. All mistakes are our own. Direct correspondence to Daniel

Silver, Department of Sociology, University of Toronto Scarborough, 1265 Military Trail,

Scarborough, Ontario M1C 1A4, Canada. Email: dan.silver@utoronto.ca

Chicago Press. https://doi.org/10.1086/719937

224 AJS Volume 128 Number 1 (July 2022): 224–286

2 million musicians and bands working at that time, when measured by the

techniques described below, the combination of genres Monáe used to de-

scribe herself were in the 99th percentile for unconventionality. While Monáe’s

unconventionality was already on full display—at the time she was in the mid-

dle of releasing a three-part concept album about a messianic android—less

well known was her history as a weirdo coming up in the Kansas City scene

and the room for experimentation that it gave her. As Monáe later explained

of her band in Kansas City, The Weirdos, “beingapartof[TheWeirdos]

and touring with them around Kansas City helped me to discover my inner

weirdo, if you will”(Gross and Hart 2013). Monáe here draws a close connec-

tion between categorical conventionality, success,and the local context. This

article develops concepts and methods for examining these interrelation-

ships more generally.

Research on the relationship between categorical unconventionality and

popularity often—although not always—ﬁnds that individuals, organizations,

and products are penalized, punished, and discounted for breaking from the

norm (Zuckerman 1999; Negro, Koçak, and Hsu 2010; Durand, Granqvist,

and Tyllström 2017). Yet an emergent and ongoing question in the categories

literature that is also exempliﬁed by Monáe’s rise to stardom is how—given

a strong categorical imperative—innovation actually occurs (Romanelli 1991;

Clemens and Cook 1999; Pontikes 2012; Pontikes and Hannan 2014; Sgourev

and Althuizen 2014). As noted by Zuckerman (2017, p. 61), an emphasis on

penalization for categorical norm violation “may have diverted the literature

from appreciating the factors that provide relief from the categorical imper-

ative, making innovation possible.”

Relying on a data set of over 2 million musicians and bands, we investi-

gate how the relationship between categorical unconventionality and popu-

larity shifts, depending on the geographic and relational context in which a

band operates. To do so, we ﬁrst discuss the categories literature, highlight-

ing contrasting ﬁndings on the return in popularity for categorical uncon-

ventionality. We then assemble a wide swath of social theorizing that sug-

gests an inverted U shape in the relationship between unconventionality

and popularity. This proposed relationship between conventionality and

popularity has been theorized in the sociological literature for well over a

century and dates back even farther in Western philosophy. For example,

Nietzsche attributes the success of artists, poets, and writers to their ability

to balance conventionality and unconventionality, a creative process that he

refers to as “dancing in chains”(1986, p. 343). In this view, unconventional

category blending tends to be rewarded up to a point, beyond which increas-

ing unconventionality becomes a liability.

Testing this hypothesis is the ﬁrst goal of this article, and we ﬁnd strong

evidence for it across multiple musical domains and hundreds of metropolitan

areas. Yet beyond the generally inverted “U-ness”of this shape, potential

Categorical Conventionality in Music

225

heterogeneity in its form has not drawn as much theoretical attention. As a

second major goal of the article, rather than controlling away this potential

heterogeneity in form in search of a singular average effect, we investigate it.

First, we generate a theoretical graph illustrating different shapes the rela-

tionship between unconventionality and popularity may take. Exploiting

the potential of big data, we then take a “forensic social science”approach

(Goldberg 2015; McFarland, Lewis, and Goldberg 2016; see also Vaughn

2014) to learn from our data in a structured yet open-ended way. After estab-

lishing that the universe of bands in our data is structured into multiple

musical “worlds”that we term “rock,”“hip-hop,”and “niche,”we show that

while the basic inverted U shape occurs within those worlds, the relationship

between unconventionality and popularity differs between them. Whereas

bands in the niche world are more harshly penalized for being conventional

than are bands in other worlds, bands in the hip-hop world operate along a

wider range of conventionality, not only with less penalty but also, unlike

bands in the other worlds, even seeing a second uptick in popularity for be-

ing more unconventional. Scaling down to the geographic level, we then gen-

erate a typology of cities (“normalists,”“traditionalists,”“experimentalists,”

and “specialists”) based on the relationship between unconventionality and

popularity for bands from them. More generally, we ﬁnd that bands from

metropolitan areas with features like higher population density, record indus-

try concentration, and younger residents are more rewarded for more uncon-

ventional musical offerings. We also provide open-access, interactive visual-

ization tools by which others can extend our observations in potentially novel

directions, more fully investigating the micro- and mesolevel mechanisms at

work in rewarding and penalizing unconventionality. In this spirit, the dis-

cussion proposes an agenda for how to study the relationship between uncon-

ventionality and popularity while still allowing for heterogeneity in local con-

texts, highlighting Nashville and Honolulu as representative of what may

be more general phenomena. We close the article with two examples of how

our approach can be readily imported into other research domains beyond

music, before discussing the contributions of our work to research on catego-

ries, cultural sociology, urban studies, big data, and forensic social science

more generally.

IS IT BETTER TO FIT IN, STAND OUT, OR BOTH?

Conventions are the artistic means acquired for the understanding of the hearer;

the common speech, learnt with much toil, whereby the artist can really com-

municate his ideas. ...What the artist devises beyond convention he offers

of his own free will and takes a risk, his success at best resulting in the setting-

up of a new convention. As a rule originality is marveled at, sometimes even

American Journal of Sociology

226

worshipped, but seldom understood. (Friedrich Nietzsche, Human All Too

Human, aphorism 122)

As summarized by Pontikes (2012, pp. 81–82), a “consensus is building that

organizations with multiple market identities are ...less likely to be success-

ful ...than more focused competitors.”Often—although not always—referred

to as the “categorical imperative”(Zuckerman 1999), empirical evidence of a

penalty for unconventionality has been found across a wide swath of mar-

kets, including those for securities (Zuckerman 1999), online lending (Leung

and Sharkey 2013), Hollywood ﬁlms (Hsu 2006), technology (Carroll et al.

2010), and wines (Negro, Hannan, and Rao 2010). As this line of research

has been summarized by Kovács and Hannan (2010, p. 176), for producers

“the threat of external punishment ...presents a signiﬁcant barrier to par-

ticipation in multiple categories.”

Although most scholarship has focused on the disciplinary mechanism of

clear categorization, the literature is much less settled on this question than

it may initially appear (Schneiberg and Berk 2010; Zuckerman 2017). As

Keuschnigg and Wimmer (2017, p. 449) put it, the “negative effects [from cat-

egory spanning] are far from universal.”Indeed, under the right conditions

there is no penalty for category mixing for law ﬁrms (Phillips, Turco, and

Zuckerman 2013) or diversiﬁcation across industries (Baker 1992; Campa

and Kedia 2002; Ruef and Patterson 2009), and category blending can bere-

warded in venture capital markets (Navis and Glynn 2011; Pontikes 2012; Wry

and Lounsbury 2013), for law and technology ﬁrms (Fleming 2001; Paolella and

Durand 2016), and for elite artists (Sgourev and Althuizen 2014) and estab-

lished actors and musicians (Zuckerman et al. 2003; Lena and Pachucki 2013;

Johnson 2017; van Venrooij and Schmutz 2018). Considering these mixed

ﬁndings, Pontikes and Hannan (2014, p. 312) summarize the “underlying ten-

sion”in the literature as one in which “social structures are simultaneously

rigid enough to induce widespread conformity and malleable enough to

change as a result of nonconformist behavior.”

Yet rather than viewing producers and products as facing a single categor-

ical imperative of varying directions and strengths, for over a century, schol-

ars across different theoretical paradigms have argued for the need to balance

between conventionality and unconventionality in market presentation. No-

tably, while this suggestion can be observed at an industry level (Peterson and

Berger 1975; Lopes 1992; Dowd 2004), organizational level (e.g., Mears 2011;

Childress 2017; Zhao et al. 2017), the level of individuals (e.g., Becker 1982;

Accominotti 2009; Goldberg et al. 2016; Wohl 2019), or the level of discrete

cultural objects such as songs (de Laat 2014; Askin and Mauskapf 2017), if

there is an imperative, the imperative is in understanding trade-offs and

ﬁnding the right mix of conventionality and unconventionality in any given

Categorical Conventionality in Music

227

context. As Simmel (1957, p. 542) wrote more generally of fashion, a “com-

promise”must be struck between “adaptation to society [i.e., conventionality]

and individual departure from its demands [i.e., unconventionality].”This

general idea has also been referred to as a “balance between innovation and

conformity”(Alston 1989, p. 153), a “trade-off between the emancipating as-

pects of entrepreneuring and the accommodation of constraints”(Rindova,

Barry, and Ketchen 2009, p. 483), and an offsetting between “universal isomor-

phism and distinctive enumeration”(Negro et al. 2010, p.18). As Zuckerman

(2017, p. 35) summarizes it, “differentiation imperatives stand in fundamen-

tal tension with the categorical imperative,”as “candidates must distinguish

their offerings from others but not have their offerings screened out forbeing

outside the conventional differentiation.”

For Bourdieu (1993, p. 123), this balancing act is expressed as a combina-

tion of “tradition and tempered innovation,”whereas for White and White

(1993, p. 10) entrants are assessed “both in terms of known standards and

in terms of originality,”and for Roy and Dowd (2010, p. 192) producers must

ﬁnd space somewhere in between in “a shifting mixture of precedence and

uniqueness.”In social psychology—and work on categories and markets in-

spired by it—thesamebasicprincipleisreferredtoas“optimal distinctive-

ness,”which is expressed through an inverted U-shaped relationship between

unconventionality and market success: unconventionality is rewarded up to a

point, but when offerings become too unusual their popularity fades (Brewer

1991; Askin and Mauskapf 2017; see Zhao et al. [2017] for review). As such, in

this type of “strategic balance”(Deephouse 1999), products must be “legiti-

mately distinctive”(Navis and Glynn 2011) without being too distinctive, such

as Uzzi et al. (2013, p. 468) ﬁnd for highly conventional articles containing “in-

trusion[s] of unusual combinations”and Lena and Pachucki (2013) ﬁnd for

sampling choices in rap songs. Taken together, this disparate theorizing leads

to our central hypothesis, which, following Nietzsche, we call the dancing in

chains (DIC) hypothesis:

HYPOTHESIS 1.—Unconventional category blending will be rewarded up to

a point, beyond which increasing unconventionality will become a liability.

This process produces an inverted U-shaped curve describing the relationship

between unconventionality in category blending and popularity.

Nietzsche’s evocative phrase offers a useful label to capture an implicit em-

pirical hypothesis embedded in the diverse theorizing reviewed above. Nietz-

sche himself developed the notion of “dancing in chains”in rich philosophical

reﬂection on the creative artistic process, emphasizing the temporal process

of balancing freedom and constraints, the weight of old conventions and the

search for new ones, and the seeming paradox of freely choosing to bind one-

self to conventions. We extend the general insight of a dynamic and shift-

ing relationship between “dancing”and “chains”by examining how the rela-

tionship between unconventionality, conventionality, and popularity varies

American Journal of Sociology

228

depending on a band’s context. What it means for a band to be unconven-

tional depends on where it sits within the broader musical landscape. To ex-

amine how context operates to deﬁne the meaning of “balancing,”we fea-

ture two key contexts that plausibly (but not fully) shape the expectations

bands meet: the larger genre complexes and the geographic location with

which they are associated.

BALANCING ACTS AND CONTEXTS

Unlike a clear and consistentcategoricalimperative, a balancing act between

two competing imperatives better allows for the possibility of nonlinear re-

lationships between conventionality and popularity. While a wealth of socio-

logical theorizing across a range of subﬁelds suggests some form of the DIC

hypothesis, beyond a widespread theoretical support for this insight, variation

in its form has remained uninvestigated and untested. This means that with

regard to variation on an inverted U we cannot appeal to a well-articulated

theoretical tradition to generate clear sets of ex ante hypotheses to be tested.

Moreover, empirical investigation of such variations has been hampered by

the fact that modeling them requires complex techniques such as nonlinear

interactions, which can be difﬁcult to interpret andsometimes unreliable out-

side of a big data context.

To meet these challenges, now operating outside of a hypothetico-deductive

framework, we ﬁrst develop intuitions about some of the possible forms the

relationship between unconventionality and popularity might take. We do

so in the form of a theoretical ﬁgure (Swedberg 2016; Silver 2020), supported

with illustrative examples of the types of settings and contexts in which these

different relationships between unconventionality and popularity could possi-

bly occur. We then consider the potential role of two types of contexts in shift-

ing the relationship between unconventionality and popularity—relational,

supralevel complexes of three different musical “worlds”(Silver, Lee, and

Childress 2016) and metropolitan-level local scenes (Silver and Clark 2016).

These contexts may provide some, but not all, of the plausible sources of var-

iation that we ﬁnd. We use these insights to help us better interpret the pat-

terns we inductively discover in our data via various techniques geared toward

revealing how nonlinear relationships between unconventional category pair-

ings and popularity may vary across contexts.

Figure 1 shows nine theoretical relationships between conventionality

and popularity. The X-axis runs from highly conventional combinations of

categories on the left to highly unconventional combinations on the right; the

Y-axis goes from low popularity to high popularity. Our goal in ﬁgure 1 is

to be illustrative, not exhaustive, and to illuminate the ways the relationship

between unconventionality and popularity can vary, the point along the spec-

trum of conventionality at which unconventionality becomes a strong liability

Categorical Conventionality in Music

229

could differ, the strength of the penalty could vary, the linearity of the relation-

ship could vary, and more. The ﬁrst row shows three simple versions of this

relationship: no relationship, linearly increasing punishments for unconven-

tionality, and linearly increasing rewards for unconventionality. Whereas

scholarship operating in the categories literature has produced ﬁndings in sup-

port of ﬁgures 1A–1Cor something like them (e.g., Hsu 2006; Pontikes 2012;

Phillips et al. 2013), ﬁgure 1D, or something roughly like it in its nonmonotonic

form, is supported by various literatures on balancing between conventionality

and unconventionality (e.g., Simmel 1957; Nietzsche 1986; Brewer 1991). The

inverted U shape in ﬁgure 1Ddescribes contexts in which optimal distinctive-

ness predominates: the most and least conventional offerings are penalized,

FIG.1.—Nine forms of the relationship between conventionality and popularity. Each

subgraph has the same Y- and X-axes, with low to high popularity on the Yand conven-

tional to unconventional genre combinations on the X. Variations on themes in the sub-

graphs can be read across the rows.

American Journal of Sociology

230

and an inﬂection point exists at the midpoint, in which the “conventionally

unconventional”receive the most market attention.

The heterogeneous ﬁndings discussed above suggest considering possible

contextual variations on the basic form of a nonlinear relationship between

conventionality and popularity. For this reason, ﬁgures 1E–1Ipresent differ-

ent variants on the basic principle of a balancing actand the rewards or pen-

alties for where one falls within it. Figure 1E—the inverse of ﬁgure 1D—is a

scenario in which extreme conventionality and unconventionality are strongly

rewarded, while offerings that balance the two are penalized. Something like

a U shape has been observed in the industry for video game platforms (Cen-

namo and Santalo 2013) and may also be found in the bifurcated market for

fashion models in New York as described by Mears (2011), in which one side

of the market rewards conventional attractiveness (i.e., conventionality),

whereas the other rewards “edginess”(i.e., unconventionality), with perhaps

a discounting for falling in between. Figure1F, which we term an “optimized

plateau”model, is a variation on the theme of optimal distinctiveness: a

wider range of unconventionality is rewarded equally, withouta perfect bal-

ance that creates a clear peak. For example, think of the world of big budget

blockbuster movies, which since their emergence in the 1970s and 1980s have

so regularly genre blended across action, drama, comedy, romance, thriller,

and so on, that it is more noteworthy when they do not do so. Even among

the more limited subset of big budget superhero movies, unconventional-

ity of genre combinations can widely range, extending without penalty from

the perhaps more conventional blending of war ﬁlm and political thriller in

Captain America: Winter Soldier to the perhaps less conventional blending

of space opera and comedy in Guardians of the Galaxy. Yet an openness to

genre blending in the world of blockbuster ﬁlmsislesscommonintheworld

of Oscar-bait movies—which are almost exclusively intimate dramas and

tend to be highly conventional biopics and period pieces (Rossman and Schilke

2014)—as well as more niche worlds of ﬁlm, like low-budget genre ﬁlms such as

horror movies in which, as with romance novels (Radway 1984), breaking

from convention at all may be sharply penalized. As these examples suggest,

even within a single cultural domain (i.e., movies), norms about what is and

is not conventional may differ depending on what part of the domain one is

examining.

Finally, the third row presents three more nonsymmetrical forms of non-

linear relationships between popularity and conventionality. In ﬁgure 1G,

which we title “recombinative tolerance,”extremely conventional offerings are

penalized, whereas moderately and extremely unconventional combinations

are equally privileged. Focusing on geographic differences, this type of curve

might be found in the 2010s stand-up comedy scene in Brooklyn, which is cel-

ebrated for category blending across different types of stage performance. As

Categorical Conventionality in Music

231

New York Magazine cultural critic E. Alex Jung (2018) explains, Brooklyn-

style comedy “isn’t just inﬂuenced by stand-up, but also drag, cabaret, bur-

lesque, experimental theater, viral videos, and variety shows. ...[It is] an

amalgam of performance art, skit, musical, and storytelling.”In this ex-

ample, place identity in part comes to inﬂect the shape of the relationship

between conventionality and success such that the scene in Brooklyn is gener-

ally a better home for, generates more, and is more widely known for unconven-

tional stand-up comedy than is, for example, Dallas. A further extension of this

idea would be an “eclectic imperative”(ﬁg. 1C) in which increasing unconven-

tionality is increasingly rewarded. This type of relationship could perhaps be

found in Zurich, Switzerland, in the 1920s during the early formation of the

anti-art artistic movement. Figure 1H,“recombinative intolerance,”is the in-

verse of ﬁgure 1Gand represents a context in which conventional and rela-

tively typical recombinations are privileged, whereas more abstruse recom-

binations are harshly penalized; it is a variant on the categorical imperative

that is sensitive to some degree of a balancing act. This could be found in the

focus on authentic ity and not breaking too far from t he norm in blues perfor-

mance expected of Chica go bands (Grazian 2005). Finally, we term ﬁgur e 1I

a“dual innovation”model. In this framework, extreme conventionality and

most unconventional combinations are punished, whereas less unusual re-

combinations are privileged. Yet the most atypical offerings also attract in-

creased popularity compared to their slightly less atypical brethren. This

would be a world or scene that follows an optimal distinctiveness model but

also contains experimental trailblazers who receive their own, if lesser, re-

wards; think of a market for ﬁne art in which the most rewarded artists bal-

ance conventionality and unconventionality but that also containsan avant-

garde at the fringes of what is acceptably art who are rewarded more than

other unconventional artists (Rawlings 2001). This can be considered one

variation on markets that contain what Zhao et al. (2017) call multiple con-

vergence points in which, rather than a single inverted U shape, there may

be multiple peaks and valleys for popularity across the full range of conven-

tional and unconventional offerings.

Rather than being exhaustive, the examples from ﬁgure 1 are intended to be

illustrative of the types of relationships between unconventionality and popu-

larity that could exist in any given context. Our goal in illustrating these pos-

sibilities is not to delimit the consideration set of relationships that may exist

but rather to expand it, given that extant theories can sometimes “act as blind-

ers”that “do not allow for serendipitous ﬁndings or unexpected discoveries”

(McFarland et al. 2016, p. 30). In that vein, we imagine that the forms of the

relationships we outline in ﬁgure 1 may vary, just as the examples we use

to illustrate those forms may vary, such that one can imagine different re-

lationships based on different audiences (e.g., Pontikes 2012), creators in

American Journal of Sociology

232

different relational positions within the same overall domain (Bourdieu

1993), and creators from different geographic locales with their own repu-

tations for what typically comes from different places and what is locally typ-

ically done there (e.g., Bennett 2000; Leschziner 2007; Phillips 2011; Hoppe

2020). The central task implied by ﬁgure 1, then, is to uncover how the balanc-

ing act between unconventionality and conventionality varies given speciﬁc

contexts.

Guided by Griswold (2014, p. 142), we think of bands as existing in com-

munities that may be deﬁned both relationally (i.e., through similarity and

dissimilarity in relational “webs of association”) and spatially (i.e., geograph-

ically, or “something we can locate on a map”). Notably, Zhao et al. (2017)

similarly observe that variation on inverted U shapes may be found both

(1) based on psychographic market contexts and (2) geographically. Follow-

ing this logic, ﬁrst, balancing between conventionality and unconventional-

ity may take different forms within different relationally deﬁned supralevel

musical worlds. Given that from a sociological perspective (unlike a musico-

logical perspective) genres are community based rather than purely sonically

based (Lena 2012), we would expect these different communities to have dif-

ferent established conventions (Becker 1982; Crossley and Bottero 2015), par-

ticularly as it relates to what is and is not conventional. Put another way, just

as producers’frames of reference are relationally deﬁned from the position of

their niches (White 2002), we might expect that bands working in the world of

hip-hop genres may have an orientation to both unconventionality and pop-

ularity different from bands in the rock world (and related genres), who may

have their own norms and deﬁnitions.

Similarly, we expect that bands from different urban contexts could, at the

margins, be held to different standards depending on the meanings connected

to the type of cultural scenes (Silver and Clark 2016) associated with their cit-

ies. In various contexts, place-based cultural identities have been shown to

shape how both locals and nonlocals evaluate people and events associated

with a given place (Molotch, Freudenburg, and Paulsen 2000), from poten-

tial tourists applying different stereotypes to Vermont and New Hampshire

based on their local reputations (Kaufman and Kaliner 2011) to tech workers

considering a move to San Francisco rather than Los Angeles partly on the

basis of the former’s bohemian countercultural reputation (Storper et al.

2015). These general place associations extend to music as well, and schol-

ars of popular music have closely examined the degree to which musicians

are stamped by the expectations and conventions of the local scenes with

which they are associated (Bennett 2017). Different metropolitan-level scenes

have different reputations not only for different styles of music (e.g., grunge

from Seattle or hyphy from Oakland) but also for being more traditional or

avant-garde in their orientations (e.g., Athens, Georgia, as a hotbed of musical

Categorical Conventionality in Music

233

innovation in the rock world in the early 1980s). A rock band from Cleveland

that experiments with unusual genre combinations runs the risk not only of

being penalized for pushing the boundaries of conventionality for a generic

placeless rock band but also for being too unusual for a Cleveland band—by

contrast, an Athens, Georgia, band may be positively expected to combine

more unusual genres.

Beyond speciﬁc cities and their unique identities, types of local contexts

may generate different expectations for bands associated with them. This

is in line with a rich tradition in popular music research and the geography

of music (e.g., Ellis and Beresford 1994; Bennett 2000; Hudson 2006; Phillips

2011; Wynn 2015; Johnson 2017; Mellander et al. 2018). This literature also

points us toward variables to include and explore in the analysis, even if it

does not give clear predictions. For example, dense cities with critical masses

of racially diverse, young, and cosmopolitan residents and subcultures (Fi-

scher 1995) tend to be associated with experimentation and personal self-

expression (Silver and Clark 2016), whereas smaller more homogenous loca-

tions tend to generate relatively stronger expectations of conventionality for

their cultural offerings. Likewise, being from highly educated college towns

(such as Ann Arbor or Madison) may raise expectations for unconventional

musical offerings (Kruse 2010) for some musical styles, while hailing from

marginalized communities and “crossroad”cities such as New Orleans, Mem-

phis, Detroit, Chicago, and Kansas City (where Janelle Monáe is from) might

do so for others (Florida and Jackson 2010, p. 311). New musical forms have

often emerged in poor and African-American communities, and location in

such settings may be a signal to reward certain forms of experimentation that

might be discouraged elsewhere (Chang 2005; Phillips 2011). Moreover,

while being located in commercial industry cores might encourage confor-

mity (Peterson and Berger (1975), cities with major record industry concen-

trations can also attract large numbers of talented and highly professionally

competent musicians across many genres, leading to increased opportunity

for novel creations, much like the creation and rise of Funk music as de-

scribed in Lena (2012). This is also the case with Nashville, where the country

music industry attracts talented musicians, some of whom then go on to be

celebrated for unconventionally combining the local “house style”of country

music with other genres.

Taken together, this literature suggests that the rates of return on uncon-

ventionality should differ across cities on the basis of factors such as record

industry concentration, median household income, racial composition, and

the percentage of college graduates and students (Fischer 1995; Florida

and Jackson 2010; Kruse 2010; Phillips 2011; Silver and Clark 2016). This

also implies that we should be able to classify cities by the form of the

unconventionality-popularity relationship their bands experience, whether

at the extremes of experimentalism or traditionalism across genre worlds, closer

American Journal of Sociology

234

to the norm, or specializing in experimentation in one genre world while re-

maining traditionalist across the others. In terms of the types of curves illus-

trated in ﬁgure 1, for bands from experimentalist cities, the punishment for

unconventionality should occur further to the right of the graph, and the fall

from most to least popular should be shorter; by contrast, hailing from tra-

ditionalist cities should generate the opposite pattern (an earlier penalty and

a steeper drop).

Combining the above review and synthesis of multiple literatures, our

analysis pursues two major questions:

Question 1.—What is the typical shape of the relationship between con-

ventionality and popularity among musicians and bands? In brief, in accor-

dance with the DIC hypothesis, we would expect to ﬁnd variants on the sec-

ond and third rows of ﬁgure 1, rather than the linear relationships between

unconventionality and popularity represented in the ﬁrst row of ﬁgure 1.

We would expect this to be the case both across musical worlds and across

metropolitan-level musical scenes.

Question 2.—How does this shape vary across metropolitan areas? The

literature on the role of place in forming the identity and reputation of cul-

tural producers suggests that where bands are from should alter expectations

around how unconventional they can be. This may be based on features of

locales that might be associated with different rates of return on unconven-

tionality, such as their racial and educational composition and their record

industry concentration. And even beyond these types of geographic resources

for bands, it may be based on the presence of different metropolitan-level

musical “scenes,”such that bands from cities with more experimentalist scenes

are afforded more space for unconventionality before being penalized, whereas

bands from cities with more traditionalist scenes are expected to conform more

strictly to conventional genre combinations.

DATA

We investigate these questions by uncovering patterns in a large data set of

musicians and bands. More speciﬁcally, our data consist of 2.88 million mu-

sical artists and groups with proﬁles on MySpace.com in January 2007. These

data were originally collected by the University of Chicago Cultural Policy

Center (see Rothﬁeld et al. 2007). At the time of data collection, MySpace

had existed for 4.5 years and was not only the most visited social network in

the United States, but it had also surpassed Google Search and Yahoo Mail

as the most visited website in the United States (Cashmore 2006). In 2007,

industry analysts worried that MySpace had become a natural monopoly

(Keegan 2007), and through an analysis of users’online behavior before

and after visiting the site, the takeaway was that “people are really using

Categorical Conventionality in Music

235

[MySpace] to discover artists and bands”(Wray 2007). At the time of data

collection, MySpace was thought of as “the de facto home page for the music

industry”(Kirkpatrick 2007). Also key to our analysis is that when these data

were collected MySpace wasstill over a year away from incorporating an al-

gorithmic recommendation engine into its platform and had not yet debuted

a“newsfeed”onto the site, meaning that music listeners on MySpace were

not algorithmically directed toward listening to some bands and not others,

nor was their listenership inﬂuenced through native advertising on the site

(Vincent 2007).

For terminological simplicity we refer to both solo artists and groups on

MySpace as “bands.”All bands in our data set self-classiﬁed their musical

offerings by selecting up to, but not more than, three genre designations from

a standardized list of 121 genres (see table A1 for a full list).Regarding those

genres, as outlined by DiMaggio (1987, p. 441), “Literally, a genre is a ‘kind’

or ‘type’of art ...[and the] notion of genre presumes that some aggregation

principle enables observers to sort cultural products into categories.”The

creation and assignation of genres can be accomplished by communities of

creators (Lena 2012), inﬂuenced through industry interests or conventions

(Negus 2013; Skaggs 2019), or deﬁned by critics (van Venrooij and Schmutz

2018) or audiences (DiMaggio 1987). While cultural sociologists rightly note

heterogeneity in genre assignments (both in where they come from and poten-

tial mismatches in assignment across those assigning), those working in the cat-

egories literature have also relied on category (i.e., genre) assignments across a

wide range of sources, all of which, of course, have their own strengths and

weaknesses. Here we brieﬂy touch on three strengths and a potential weakness

of our data on genre assignments.

A strength of our data is that on MySpace the same genres were available

for self-identifying to all bands. Hence, their self-classiﬁcations occur within

the same categorical system, consisting of the same availabilities and con-

straints, while still allowing for high levels of granularity in self-deﬁnition

(Hsu, Hannan, and Koçak 2009). As a result, unconventionality in categor-

ical self-presentation was available to them but not forced on them. With 121

genres afforded and three available slots in which to apply genres to repre-

sent themselves, bands had around 300,000 unique categorical identities avail-

able to them. While additional genres to pick from might provide bands some

greater leeway to express a wider range of genre combinations, the MySpace

platform offered signiﬁcant opportunities to both afﬁrm their strong identi-

ﬁcation with conventional styles and display themselves in unusual ways.

While for our question it is only important that bands have a range of conventional and un-

conventional genre combinations available to them (and in our case they have a wide range

with around 300,000 possible combinations available to them), using critic’sclassiﬁcations for

American Journal of Sociology

236

In fact, as discussed below, all bands did not use all available genre slots, nor

were combinations across these genres equally distributed.

A second strength of our data is their size and scope. If we think of bands

in 2007 as an iceberg, the vast majority of bands in our data set would be in

the (vast majority of the) iceberg that is below the surface. This submerged

portion of the iceberg is for the most part entirely ignored by those who

would otherwise be engaging in category assignments (e.g., critics, industry

observers, record labels, or communities of superfans). Simply put, producer

self-classiﬁcation allows us to investigate millions of the otherwise unclassi-

ﬁed bands that are therefore unobserved in most research. Third, while we

do not know whether audiences agree with our bands’self-deﬁnitions, we

know, as we discuss below, that they were exposed to them as a feature of

bands’self-presentations. While there will always be heterogeneity in audi-

ence classiﬁcation systems (e.g., Pontikes 2012), we know both how bands

use labels to self-present to potential audiences and that those audiences were

exposed to those self-presentations.

The major potential limitation of self-designation is that it can be aspira-

tional (Granqvist, Grodal, and Woolley 2013; Pontikes and Kim 2017) and

possibly not reﬂect true category membership or competency. While still a

concern, this is less of a worry in our data than is often the case. The vast ma-

jority of bands in our database are “paid”not in monetary compensation from

consumers but through the “psychic goods”of identity as an artist (Menger

1999, p. 556; see also Bourdieu 1993). This remuneration through artistic

identity means that misrepresenting one’s artistic identity for perceived stra-

tegic gain is perhaps less common in our data than in data on more humdrum

goods and products; regardless, there is no reason to believe it is a systematic

problem in our data.

On MySpace, self-selected genre designations appeared directly below the

band’s name on the top leftof the page (the most prime real estate for left-to-

right languages such as English). These genre designations, should there be

more than one selected, were separated by slashes (e.g., “rap/hip-hop/R&B”).

Likewise, beneath and to the right of genre designations, the next text pre-

sented to potential listeners is the city and state in which the band is located.

While we cannot be assured that every visitor to every page treated these cat-

egory and city designations as meaningful information, because they were

clearly displayed directly beneath the name of the band on the top left of

the page guarantees that this information was observable, readily available,

and treated as meaningful on the platform. In brief, our analysis of relational

147,000 albums on allmusic.com and without a constraint on the number of genres assigned,

Hannan et al. (2019, p. 122) ﬁnd that assigning three genres is modal (58% of all cases) with

only one genre being applied to about 20% of all cases.

Categorical Conventionality in Music

237

musical worlds and geographic musical space mirrors the information that

we know was available to potential listeners of these bands.

Our database also includes for each band its total number of plays, views,

and fans. Unfortunately, additional band-level information is not available,

although future research would beneﬁt from incorporating such information.

After several transformations from these raw data, which we discuss below,

we merge the data set with data from the U.S. census and County Business

Patterns (CBP). These metropolitan-level variables allow us to investigate

how the relationship between unconventionality and popularity shifts across

urban contexts.

ANALYTIC STRATEGY

The various subgraphs in ﬁgure 1 illustrate scenarios in which the relation-

ship between (un)conventionality and popularity takes on different shapes,

which in turn indicate diverse modes of rewarding categorical conventions

in the form of popularity. Our analytic strategy exploits opportunities within

our data set to observe this relationship within and across different contexts.

We examine two major spaces in which bands operate: (1) the relational struc-

turalspaceofmusicalworldsand(2)the physical geographic space of metro-

politan areas. Our analysis inductively traces the relationship between bands’

popularity and the conventionality of their genre choices across these contexts,

with an eye to ways in which the resultant curves approximate the theoretical

constructs in ﬁgure 1, as well as other relationships that emerge in the data.

After describing how we measure popularity, we discuss our procedures for

investigating these two contexts in order.

Deﬁning Popularity

During the time of data collection, MySpace did not have a mechanism for the

purchase or sale of music. Instead, our outcome variable is a construct of three

indicators of popularity: total number of views, total number of fans, and total

number of plays in the embedded music player on each band’s page. These

measures are highly correlated, with a Cronbach’s alpha of .91, well above

the standard .7 threshold for construct validity. To combine these threevar-

iables into a single construct, we take natural logarithms of each component

variable (views, plays, and fans) to normalize their distributions (which are

highly skewed), calculate a Z-score for each, and then add them. This helps

ensure that components are given equal weight in the popularity score de-

spite their differences in numerical magnitude (i.e., bands will get many

more page views than people signing on as fans).

Table 1 shows the mean, median, and standard deviation for plays, views,

and fans. Figure 2 shows plays, views, and fans across the three worlds (derived

American Journal of Sociology

238

according to methods described below). On average bands in the rock world

are more popular than bands in the hip-hop world, which are slightly more

popular than bands in the niche world. The large disparity between the

means and medians indicates that, as is generally true in markets for cultural

goods, a relatively small number of bands are extremely popular, while most

are not. For example, the median band has three fans and 120 plays.

As music in 2007 was not available for sale on or through MySpace, like

Salganik, Dodds, and Watts (2006) we are able to measure a relatively pure

form of popularity that is not conditioned on factors like pricing, variation in

disposable entertainment income, or the possible prevalence of piracy over

purchasing in some worlds more than others. Likewise, in 2007 MySpace

was an unmediated market (Hirsch 1972), allowing for direct interaction be-

tween bands and potential fans of their music without any sorting, segregat-

ing, evaluations, or “coverage mismatch”(Zuckerman 1999) from third-party

intermediaries or some third-party intermediaries over others (Dowd et al.

2021).

Deriving Three Musical Worlds: Rock, Hip-Hop, and Niche

We create a large and complex genre network by mapping the band-provided

(self-identiﬁed) colistings of (up to three) genres. Genres are considered “related”

when a band lists them together. For the full population of bands that listed

more than one genre designation, greedy modularity optimization is employed

to identify genre communities. These procedures resulted in three higher-

orde r genre complexes, which we term “worlds”: rock, hip-hop, and niche, which

are made up of 16 lower-order genre communities. These genre worlds are

As a robustness check, we ran our regression models with a dummy variable for bands

with the lowest popularity score, which did not alter our ﬁndings.

While we believe ours is a “purer”measure of popularity than sales for the above-stated

reasons, in a supplementary analysis of the top 150 bands on MySpace (available on re-

quest) popularity does track well with sales.

TABLE 1

Mean, Median, and Standard Deviation for Plays, Views, and Fans

Plays Views Fans

Mean . . . . . . . . . . . . . 3,788 2,137 222

Median . . . . . . . . . . . 120 125 3

SD . . . . . . . . . . . . . . . 147,595 67,886 3,582

NOTE.—Data are across all MySpace bands. Large SDs indicate that while

the preponderance of bands have relatively few plays, views, and fans, popular-

ity is spread across a great range of values including smaller numbers of very

popular bands.

Categorical Conventionality in Music

239

FIG.2.—Popularity varies by musical world. Mean values and 95% conﬁdence intervals for three popularity metrics (plays, views, fans) within

three musical worlds (rock, hip-hop, and niche). Given the size of the data set, conﬁdence intervals are extremely narrow. Rock musicians are con-

sistently more popular, on average.

robust across multiple network-based community detection algorithms.

They

result from the fact that musicians are more likely to pair some genres together

than others. Genres are not isolated units combined haphazardly but are an-

chored in higher-order musical groupings—islands of musical inbreeding,

each of which is structurally distinct from the next, mixing more within their

boundaries than between. The hip-hop world includes genres such as rap,

R&B, and crunk; rock includes genres like metal, alternative, and punk; and

niche includes genres like electronica, experimental, and Hawaiian. More

details regarding these procedures and the rationales for them are available

in appendix D and Silver et al. (2016).

To be clear, the construction of these worlds does not drive the overall in-

verted U shapes we document below. Rather it allows us to investigate whether

within the basic framework of an inverted U shape we see different patterns

across worlds in the relationship between popularity and unconventionality,

just as we subsequently investigate whether our results differ by geographic

place. The conceptualization of multiple musical worlds as deﬁned by genres

within the broader universe of music making is not foreign to sociologists

of music, who emphasize that in music there are multiple “genre worlds”

(Frith 1998), musical “worlds”(Crossley and Bottero 2015), and “music cul-

tures”(Negus 2013) that are simultaneously operating according to different

conventions, evaluations, norms, or standard practices. Yet our interest in

whether the worlds of rock,hip-hop, and niche genres may substantively dif-

fer from one another when it comes to the relationship between unconven-

tionality and popularity is not solely derived from a broader acknowledgment

that music is organized by genres or that those families of genres come with

different communities and conventions (which is true across genres and

supralevel families of genres across all creative goods). Rather, in addition

to being relationally derived through webs of association (Griswold 2014,

p. 142), our three worlds empirically differ in ways that would reasonably play

a role in our central question on the relationship between popularity and

unconventionality.

Most importantly for the question of the relationship between popularity

and conventionality, the baseline level of popularity in the rock, hip-hop,

Our analysis showed that the clustering produced by greedy modularity optimization

was robust across algorithms. In comparing clusterings from the greedy modularity op-

timization algorithm to alternatives, similarities are in the range of 93% (walktrap com-

munity detection) to 96% (multilevel modularity optimization). The (small) differences

tended to revolve around the relative size of clusters, with alternatives producing less bal-

ance. For instance, alternatives might place a few genres into the larger niche cluster that

greedy optimization placed into the hip-hop cluster. Despite the fact that multiple algo-

rithms would generate similar clusters, we chose greedy modularity optimization in part

because it is the most computationally efﬁcient and also because its logic reﬂects in a

straightforward way our intention to ﬁnd densely interconnected areas of the genre net-

work (discussed further in Silver et al. 2016).

Categorical Conventionality in Music

241

and niche worlds empirically differs, as do each world’s baseline character-

istics of genre blending. Overall, as ﬁgure 2 shows, bands in the rock world

are on average the most popular, whereas bands in niche world are the least.

Likewise, these worlds have different degrees of internal differentiation: our

community detection algorithm found no subcommunities in the hip-hop

world and no musical core around which other genres operate in the niche

world (see app. D). The worlds also have different levels of permeability at

their boundaries: while bands in the niche world still inbreed with other

niche world genres more than they crossbreed into genres in other worlds,

they do extend out to other worlds in their genre combinations more than do

bands in the rock and hip-hop worlds (Silver et al. 2016). Given that these

worlds differ both for overall popularity and for genre blending within and

between them, we would, in fact, be somewhat surprised if there was not

variation in how the relationship between popularity and conventionality

operates across these worlds.

A reasonable question concerns the substantive meaningfulness of the

niche world, which contains genres that are more sonically heterogeneous

than those in the rock or hip-hop worlds, as well as including more regional

specialty genres such as “Spanish pop,”“regional Mexican”and “J-pop.”De-

spite this internal heterogeneity, these niche genres in our view still consti-

tute a world in our sense of the term (1) because bands more frequently blend

genres within the niche world than into either of the other worlds and (2) be-

cause the genres in the niche world occupy the same structural position in

relation to the two dominant worlds of popular music: rock and hip-hop.

In fact, the logic of structural equivalence (Burt 1976; White, Boorman,

and Breiger 1976) is central to the types of relational measures we employ

in constructing these worlds and is particularly important if the structural

equivalence of entities may be capturing some of the variation in one’sre-

search question.

In our case, bands in the niche world are structurally equiv-

alent not only in how they combine genres in relation to the rock and hip-hop

worlds but also with regard to their relationship to popularity. Simply put,

because in 2007 rock and hip-hop were the two major worlds around which

popular music orbited, by working in niche genres, bands were creating ceil-

ings on how popular they could become (e.g., regional Mexican bands are rea-

sonably doing something that is not oriented around trying to make it onto

the Billboard Top 100). This is to say that if one is interested in the relation-

ship between genre combinations and popularity, a world of bands that are

It is worth noting that the point of structural equivalence is tosee beyond the types of cat-

egorical groupings that are more readily apparent. Recalling Mohr (1994) on 19th-century

welfare relief, unwed mothers and returning soldiers were the structurally equivalent “de-

serving poor”in relation to tramps, despite unwed mothers never also being returning sol-

diers and vice versa.

American Journal of Sociology

242

structurally equivalent in their relationship to unconventionality in genre com-

binations and who also have opted to work in comparatively less popular

genres is substantivelymeaningful for our research question on both counts.

Nevertheless, we reiterate that we do not enforce meaningfulness in our mea-

sures on the niche world in relation to the other two worlds but rather allow

for meaningfulness to be observable should it exist.

Deﬁning Genre Conventionality

In the categories literature, categorical typicality is most often treated linearly

through count data, in which each additional category assigned correlates

with a one-unit increase in categorical complexity (Hsu et al. 2009; Kovács

and Hannan 2010; Pontikes 2012). While this measure is elegant in its sim-

plicity, it elides research on classiﬁcation systems in which conceptual dis-

tances between categories are not equal. For instance, a count data approach

would treat the blending of “black metal”and “death metal”as equivalent to

the blending of “death metal”and “opera.”Following recent developments in

this stream of research (e.g. Kovács and Hannan 2015; Wry and Castor 2017;

van Venrooij and Schmutz 2018), we therefore treat musical genres as rela-

tionally positioned to one another: those genres that regularly co-occur are

closer and more conventional (e.g., black metal and death metal), whereas

those that do not (e.g., death metal and opera) are more distant and uncon-

ventional (Church and Hanks 1990; Mohr 1998; Gärdenfors 2004; Pachucki

and Breiger 2010). While a relational approach to classiﬁcatory systems is far

from novel (Bourdieu 1993; Mark 1998; Ruef 2000; Ruefand Patterson 2009),

it is, we believe, too irregularly applied to work on category blending.

To this end we adapt Lizardo’s (2014) measure to our case and data. In this

measure, genres that are routinely connected to one another are considered

more conventional, while more unusual combinations count as more uncon-

ventional. Technical details about how we construct the measure are in ap-

pendix B.

The resulting unconventionality measure ranges from 0 to 1, with

values closer to 1 indicating a band with a more unconventional set of genre

Lizardo’s (2014) measure was developed to quantify unconventional genre preferences

in consumers. However, the measure itself is neutral with respect to its object, as are other

plausible metrics such as species of the broader family of Jaccard-style distance metrics,

variants of which have also appeared in the categories literature (e.g., Kovács and

Hannan 2015). Accordingly we compared results based on our min-clustering approach

with a standard Jaccard measure and Kovács and Hannan’s (2015). In all cases the in-

verted U appears, and the results of the multilevel regression models are nearly identical.

However, for both the standard Jaccard and Kovács and Hannan’s (2015) variation, un-

conventionality is compressed toward the right side of the distribution due to Jaccard-

based measures’collapsing disparity in size between genres with the unconventionality

of the pairing of those genres. Our approach allows us a direct path to our research inter-

est about the combining of genres in atypical ways.

Categorical Conventionality in Music

243

combinations. The most categorically conventional bands in our data, which

have an unconventionality score of 0, self-present as fully committed to a

single genre by either listing it multiple times across each available genre slot

(e.g., death metal/death metal/death metal) or only listing one genre (e.g.,

jazz/[blank]/[blank])—these are indications of conforming to clear categorical

conventions.

Scores in the middle range from .3 to .7 tend to include fairly

familiar genre combinations, such as metal/thrash/death metal (rock world),

pop/R&B/hip-hop (hip-hop world), techno/trance/turntablism (niche world).

These are the types of conventionally unconventional genre pairings we would

expect toward the middle of the distribution: they are not the entirely conven-

tional genre purists who set the baseline (e.g., punk/punk/punk), nor are they

the highly unconventional weirdos who appear at the high values. Those very

unconventional genre combinations become extremely unusual above around

.9, including self-presentations such as electro/surf/powerpop and crunk/indie/

techno (rock world), ambient/death metal/hyphy and club/afrobeat/ska (hip-

hop world), and Christian rap/jungle/thrash and dub/house/thrash (niche world).

To return to our opening example, in our data Janelle Monáe’s genre com-

binations placed her in the 99th percentile of rock world unconventionality.

Linking Bands to Cities

Given the linkages suggested by the literature between various place charac-

ter and musical conventions, we locate bands in metropolitan areas and ex-

amine the extent to which the general relationship between conventionality

and popularity shifts, depending on the city a bandis associated with. To as-

sign bands to geographic areas, we exploit the fact that bands could list their

city and state on their MySpace proﬁle. While in most cases thiswas straight-

forward, in some cases bands did not enter a city or state or entered terms that

could not be interpreted as a city or state except on a case-by-case basis. Of

the approximately 2.5 million band pages that were identiﬁed as being in the

United States, we were able to match about 2.2 million to a census-designated

primary metropolitan statistical area (PMSA; for convenience we use the ge-

neric terms MSA, metro, or metro area throughout this text), and thus they

could be included in the geographic analysis.

Because the number of genres chosen may itself be associated with bands’popularity

(e.g., in that completing all three genre slots is a sign of being more committed to public

self-presentation), we control for this in our regression models.

Matching bands to PMSAs required a few different approaches. While the vast major-

ity of bands could be matched to PMSAs after correcting for common errors, typos, and

stylized spellings in their provided city/state information, about 150,000 were matched by

SAS’s soundex routine based on pronunciation. In addition, city names that had an iden-

tiﬁed state and were used 10 or more times were manually investigated and matched to a

metro area when possible.

American Journal of Sociology

244

Although bands are embedded in or have emerged from local metropolitan

scenes, especially since the advent of virtual spaces like MySpace, they also

operate in translocal and virtual scenes (Peterson and Bennett 2004).

Like-

wise, the advent of virtual distribution channels like MySpace allowed bands

to reach widespread audiences outside of traditional industry distribution net-

works. Accordingly, our measure of popularity is not strictly local—a San

Francisco–based band can and will have plays, views, and fans from else-

where. Nevertheless, wherever they are located, we know that, compared

to today, audiences were much more likely to have at least some knowledge

of the bands they were listening to online. This is because our data exist in a

brief window in which MySpace was the most popular music destination on

the Internet, but before the advent of newsfeeds, algorithmically generated

taste suggestions, promotional advertising on the site to attract more atten-

tion, or even multiartist playlists. Likewise, we know that audiences are ex-

posed to a band’s location information on their MySpace page, which can

potentially shift expectations for how much unconventionality is appropriate

for, for example, a rap artist from New York or Los Angeles or a rock musician

from Cleveland or Athens. Analytically, this straightforwardly suggests consid-

ering the impact on popularity of the interaction between the place label asso-

ciated with a band and its unconventionality.

While our data cannot account for fan location or the degree to which a

given band tours and participates actively in virtual spaces, we note that re-

search suggests that even in a digital setting most bands’online audiences

come from the local milieu out of which they operate (Allington 2014). Given

that the vast majority of MySpace bands have very small followings, we ex-

pect this to be especially the case in our data set. Moreover, to the extent that

the typical MySpace band does reach extralocal audiences, those audiences

would accordingly tend to be highly knowledgeable about the band and the

local scene in which it operates. Overall, the low median popularity indicates

most bands in the MySpace universe are “garage bands”operating within rel-

atively local scenes with little translocal reach.

Still, we cannot differentiate

In a supplementary analysis for the top 150 bands on MySpace, we investigated their

relationship to the cities they listed on their proﬁles: 74% listed places that they were ei-

ther born or raised in and launched their careers from (e.g., Ice Cube and Los Angeles or

Regina Spektor, whose family moved to New York when she was eight and who came up

through the antifolk scene in New York); 22% listed cities they moved to in young adult-

hood (usually after high school or college) and developed their sound in for a signiﬁcant

amount of time before their ﬁrst release (e.g., Blake Shelton, who moved to Nashville at

17 and worked and performed in Nashville for nine years before his ﬁrst release); and 4%

listed cities they had moved to after their ﬁrst release, with this type of infrequent residen-

tial mobility being inversely related to popularity even in the top 150 bands ranked for

popularity (i.e., the top .005% of bands in our data ranked for popularity).

Very popular bands by contrast likely have much broader reaches, suggesting that their

location may be less important, although not entirely unimportant: regardless of their location,

Categorical Conventionality in Music

245

between our metropolitan-level effects being somewhat “noisy”because of

the existence of some nonlocal fandom or nonlocal fans holding unconven-

tional bands from places with characteristics that might signal or produce

openness to experimentation (or the opposite) to different standards. Never-

theless, as we return to in the discussion, pursuing with suitable data a more

robust analysis of how these multiple spaces operate separately and in inter-

action is an exciting avenue for extending our research.

Metropolitan Variables

Our metro area variables come from three data sources: the U.S. census, CBP

(aggregated to the PMSA level), and our MySpace data (also aggregated). Be-

cause metro boundary deﬁnitions changed in 2010 and our PMSA deﬁnitions

are from 2000, we use data from the 2000 census. The CBP data are from

2001. Given the highly aggregated character of these data, it is extremely un-

likely that relative differences among metro areas would change substantially

between 2000 and 2007.

The census provides population data that the literature has identiﬁed as

deﬁning urban contexts in which various musical styles may be differentially

favored. These population variables include population density, median house-

hold income, education (the percentage of the populationwithacollegedegree),

youth concentrations (the percentage of the population ages 25–34), the per-

centage of the population that are college students, and the nonwhite percent-

age of the population. To capture local music industry concentration, CBP

contains information about recording industry establishments and music ex-

hibition and consumption spaces. Speciﬁcally, we include two indexes. One

is a metro per capita record industry index, which sums four six-digit North

American Industrial Classiﬁcation System (NAICS) categories: “Record

Production,”“Music Publishers,”“Sound Recording Studios,”and “Other

Sound Recording Industries.”The other is a per capita index measuring con-

centrations of music consumption and exhibition opportunities: “Musical In-

strument and Supplies Stores,”“Prerecorded Tape, CD, and Record Stores,”

“Promoters of Entertainment Events with Facility,”and “Promoters of En-

tertainment Events, without Facility.”

Finally, we construct a metro-level variable from our MySpace data. This

variable measures the degree to which each metro specializes in a particular

audiences may implicitly or explicitly treat rap bands, e.g., from Los Angeles, Atlanta, or

New York, differently.In line with this insight, we found that restricting our regression anal-

ysis to the top 5% of bands in terms of popularity reduced the explanatory power of metro-

politan variables to some degree.

Details about organizations included in NAICS categories may be found at https://

www.census.gov/naics/.

American Journal of Sociology

246

musical world (as deﬁned above). To construct this variable, we calculate an

entropy index (speciﬁcally, Shannon’s entropy H) based on the number of bands

in each of the three worlds. Lower values indicate greater specialization.

We use several different techniques to examine how the relationship be-

tween unconventionality and popularity varies by musical world and metro-

politan area. Speciﬁc details about these methods are provided in the course

of the analyses.

RESULTS

Testing the Dancing in Chains Hypothesis

We ﬁrst examine the relationship between bands’conventionality in genre

recombination and their popularity across musical worlds. Figure 3 visual-

izes generalized additive models (GAMs) showing the relationships between

unconventionality and popularity for all bands across all three musical

worlds.

The DIC hypothesis holds. Across all three musical worlds we ﬁnd

an inverted U shape. Two-line tests (Simonsohn 2018) conﬁrmed the non-

monotonic, inverted U-shaped function in each musical world.

The inverted U is the predominant shape taken by the relationship be-

tween unconventionality and popularity, strongly conﬁrming our central hy-

pothesis. Yet ﬁgure 3 also suggests minor variations on this dominant motif.

Rock and niche worlds largely approximate an optimal distinctivenessmodel

(ﬁg. 1D), with rock somewhat resembling the optimized plateau model in

which the range of acceptable unusual combinations is relatively wider, and

the peak is somewhat ﬂatter. More speciﬁcally, rock and niche both exhibit

curves that are slightly skewed to the right. The niche curve starts lower (i.e.,

a stronger penalty for no genre blending at all) than the rock and hip-hop

worlds. In these musical worlds, bands reach an inﬂection point at an un-

conventionality score of around .65 or so, and then they experience sharper

As Beck and Jackman (1998) describe, GAMs are especially well suited for cases like

ours in which the shape of the relationship between the independent and dependent var-

iable is an open question. While standard forms of regression allow for the possibility of

nonlinearity, in our case existing research provides little theoretical guidance when it

comes to specifying the appropriate parameterization a priori, as required when using

more conventional techniques. A GAM gets around this problem by allowing the relation-

ship between the independent and dependent variable to vary locally across the range of

the predictor. The smoothness of the resulting curves depends on the value of the user-

deﬁned smoothing parameter k, which can be selected using generalized cross-validation

(Wood 2017). As a robustness check, we experimented with alternative values of kand

found consistent support for the claim that the relationship between unconventionality

and popularity is characterized by an inverted U shape, with the returns on unconvention-

ality tending to fall off at eitherextreme. The visual evidence produced by the GAMs was

subsequently corroborated using more conventional parametric procedures including two-

line tests and quadratic multilevel regression, as discussed below.

Categorical Conventionality in Music

247

FIG.3.—Genre conventionality and popularity across three musical worlds: rock, hip-hop, and niche. All three exhibit an inverted U-shaped char-

acteristic of optimal matching. However, the shapes vary, indicating different ways of balancing these imperativ es. Curves are estimated using a GAMin

which the relationship between popularity and unconventionality in a given world is expressed in terms of a penalized cubic regression spline. Penalized

cubic regression splines were implemented using the shrinkage smoother included as part of the mgcv library in R (Wood 2017). The visual evidence

produced by the GAMs is corroborated using more conventional parametric procedures including two-line tests and quadratic multilevel regression.

declines in popularity as theyget more unconventional from there. The types

of genre combinations chosen by relatively popular bands at these inﬂection

points are in general not extremely conventional, nor are they extremely in-

novative. For example, techno/trance and punk/glam combinations are near

the inﬂection point in the niche world, as are rock/alternative/grunge and

thrash/metal/hardcore in the rock world. These are semifamiliar genre com-

binations that stand in between categorical purity on the onehand and more

unconventional recombinations on the other. In these worlds they are the

conventionally unconventional genre pairings that attract fans.

The hip-hop world alsois well described by an inverted U shape, although

it takes a slightly different form. It follows the same pattern as rock and niche

through the ﬁrst three-quarters of the range, with the inﬂection point occurring

around a .5 unconventionality score—earlier than the other worlds. The imper-

ative to convention is somewhat stronger in the hip-hop world, although the

curve is somewhat ﬂatter, suggesting that the range of unconventionality that

is rewarded is relatively wider. Bands’popularity peaks at a relatively higher

level of conventionality (e.g., hip-hop/pop/R&B) and diminishes when they

stray into more unconventional territory, such as neo-soul/techno or hyphy/

jazz. At the same time, the “punishment”for such unconventionality is not

as steep as it is in rock and niche, in that the distance between peak and

trough popularity is substantially smaller in the hip-hop world.

The relationship between unconventionality and popularity in the hip-hop

world further complicates at the tail of unconventionality, where it resembles

the dual innovation model envisioned in ﬁgure 1I. After the ﬁrst major uptick,

a second one appears, where even more unconventional combinations result

in relatively heightened popularity, before, in this case, trailing off again when

their unconventionality becomes too extreme.

Examples of genre combina-

tions chosen by relatively popular bands at this second inﬂection point in the

hip-hop world include hyphy/rock/punk and indie/crunk. Popular artists at this

second peak in the hip-hop world in 2007 included Cody ChestnuTT (perhaps

most well known for his collaboration with The Roots on “The Seed (2.0),”

which was described as a “hybrid of distorted rock, hip-hop and psychedelic

soul”[Petridis 2002]) and Zilla Rocca (an unconventional rapper who would

go on to later self-identify as making what he called “noir-hop”).

Popularity and Conventionality Vary by Urban Context

So far, we have demonstrated a strong and consistent inverted U shape across

three musical worlds, which describes a nonlinear process whereby bands’pop-

ularity increases as they incorporate more unconventional genre combinations,

The right tail of the rock world exhibitsan uptick, where extremely unconventionalgenre

combinations (e.g., psychedelic/death metal/southern rock) tend to increase in popularity.

Categorical Conventionality in Music

249

before hitting an inﬂection point where additional unconventionality reduces

their popularity. We also pointed toward some evidence that this general pattern

comes in somewhat different forms across musical worlds and that the ways

genre recombination is rewarded and punished correspond to scenarios en-

visioned by our theoretical ﬁgure. We now move to the metropolitan context.

We examine the extent to which conventionality is rewarded differentially

depending on where bands are in geographic space. To do so we pursue a

two-step strategy. In the ﬁrst step, we compare metros by the shapes of their

conventionality-popularity curves across musical worlds. In a second step,

for each genre world we ﬁt a multilevel regression model that accounts for

both within and between metro variation in the conventionality-popularity

relationship. These models are designed to highlight variables that shift the

relationship between conventionality and popularity across metros. More de-

tails about both methods are provided below.

Turning to the ﬁrst step, we again use GAMs to generate separate curves for

each musical world, this time for each metro separately. This produces three

curves for each metro, with one curve for each musical world. This is illustrated

in ﬁgure 4 using the case of San Francisco. We construct similar sets of curves

for all metro areas.

Comparing ﬁgure 4 to ﬁgure 3, it is striking that the macropattern of the

inverted U recurs in this local context, again conﬁrming the generality of

the DIC hypothesis. Yet ﬁgure 4 also indicates that across all three musical

worlds the rate of return onunconventionality for bands located in San Fran-

cisco differs from the typical pattern. Nationally, the inﬂection point for pop-

ularity occurs between about .5 and .65. For San Francisco bands, however,

the inﬂection point occurs between .875 and .925. To be sure, even for bands

from San Francisco extreme unconventionality is eventually penalized, but

this discounting is lessextreme than it ison average. Compared to othermet-

ropolitan locales, across all three musical worlds it pays to be an unconven-

tional bandconnected toSan Francisco’s reputation for experimentation and

innovation.

To facilitate comparison and exploration of these types of curves across

metros and musical worlds more generally, we built an online interactive vi-

sualization tool, which also provides more methodological details (see https://

unconventionality.github.io/). Figures 5 and 6 are stylized snapshots of this

tool, but we encourage readers to view the interactive visualization and its

corresponding appendix to better understand how to use the interface and

the methods by which we produced it.

Points are positioned in the central panel to reveal similarities and differences

in the shapes of their corresponding curves. Figure 5 illustrates the overall re-

sults, which again can be explored in more detail at https://unconventionality

.github.io/. In both ﬁgure 5 and the online tool, the left panel shows the three

basic patterns describing the conventionality-popularity relationship across

American Journal of Sociology

250

FIG.4.—Atypical receptivity to unconventional genre combinations across rock, hip-hop, and niche musical worlds, for the San Francisco PMSA.

Curves are derived from GAMs with cubic splines, with R’s mgcv package. We construct similar curves for all metro areas and compare them.

all metros. Each pattern is present within each metro to varying degrees. Per-

centages that accompany each pattern indicate the contribution of that pattern

to the total shape of all curves across all metros.

Pattern 1 is accordingly the most common factor, contributing about 54%

of the total shape of the curves and representing the basic inverted U of op-

timal distinctiveness in ﬁgure 1. Pattern 2 contributes about 30% and ap-

proximates the dual innovation model of ﬁgure 1. The distribution of points

on the graph indicates that, independent of metro characteristics, the niche

world tends to resemble pattern 1, the hip-hop world tends to resemble pat-

tern 2, and rock is in the middle, with a somewhat greater tilt toward pattern 1

than pattern 2. This aspect of the decomposition thus corroborates the results

of ﬁgure 3, showing that musical worlds constitute the major axes around

which our universe of over 2 million bands turn.

Pattern 3, however, reveals an additional, metropolitan axis of comparison

that cuts across musical worlds. Accounting for about 16% of the popularity-

conventionality curves within metros, this pattern describes the degree to

which unconventionality or conventionality is rewarded for bands from a

FIG.5.—Comparing the conventionality-popularity relationship across metro areas and

musical worlds. All metro-musical world combinations are placed into a three-dimensional

space derived from a nonnegative matrix factorization of their local unconventionality-

popularity curves. Metro-world combinations are positioned by how closely they approx-

imate the three patterns on the left. Bubble sizes are proportional to metro population

size. There are three bubbles for each metro and one bubble for each musical world.These

relationships can be exploredin more depth through an online interactive tool accessible at

https://unconventionality.github.io/.

American Journal of Sociology

252

FIG.6.—Illustrating normalists, traditionalists, experimentalists, and specialists. Four metro areas highlighted as examples of general types of

metropolitan formations of the conventionality-popularity relationship. These relationships can be explored in more depth through an online in-

teractive tool accessible at https://unconventionality.github.io/.

given metro. Looking at the right panel of ﬁgure 5, within each world, there is

a clear distribution of metros ranging from the bottom left of the panel (more

receptive to unconventional genre blending) to the top right (less receptive).

Taken as a whole, ﬁgure 5 empirically illustrates the core theoretical intui-

tions behind this study: that, within the basic framework of a balancing act

between conventionality and unconventionality, the relationship between

conventionality and popularity comes in many forms and that the shape

and distribution of those forms track (to varying degrees) the (musical) cat-

egorical context (i.e., genre worlds) and geographical context (i.e., metros).

These methods also allow us to identify metros that occupy distinctive

positions within the space of conventionality-popularity relationships, such

as San Francisco. In the online visualization, hovering over or clicking on a

point in the central panel changes the left panel. The three patterns are

ranked according to their prominence for that metro and musical world.

For example, clicking on New York City’s niche curve shows that pattern 3

contributes about 32% of its shape, more than pattern 2 does (18%). These rel-

ative contributions account for the position of each point in the central panel.

Based on an examination of the positions of metros via the interactive

tool and the relative shares contributed by pattern 3 to their shapes, table 2

presents a typology of four different types of metropolitan areas. Figure 6

adds graphical illustration, while also providing further examples of simi-

larly positioned metropolitan areas in table 2. Bands from normalist metros,

illustrated by St. Louis, Missouri, in the top left of ﬁgure 6, have all three of

their points in the middle of the distribution, generally exhibiting a typical

relationship between conventionality and popularity across all genre worlds.

Bands from conventionalist metros, illustrated by Buffalo, New York, in the

top right of ﬁgure 6, are rewarded for conventional genre combinations and

punished for unconventional pairings with greater force than are bands from

other cities, across all worlds; for these metros, all three of their points tend to

cluster in the upper right of the diagram. Bands from experimentalist cities,

such as New York in the bottom left of ﬁgure 6, are more rewarded for uncon-

ventional combinations across all worlds: all three of their points tend to cluster

in the bottom left of the diagram. Last, bands from specialist cities experience

a sharp contrast across musical worlds, for example, by being less or more

TABLE 2

Typology of Cities

Normalists Traditionalists Experimentalists Specialists

St. Louis, Mo. Buffalo, N.Y. New York, N.Y. Baton Rouge, La.

Dallas, Tex. Gary, Ind. Ann Arbor, Mich. Birmingham, Ala.

Sacramento, Calif. Norfolk, Va. Nashville, Tenn. Miami, Fla.

Dayton, Ohio Vallejo, Calif. San Francisco, Calif. Greenville, N.C.

American Journal of Sociology

254

receptive to eclectic offerings in one world versus the others. For these types

of metros, there are great distances between their threepoints. In the bottom-

right corner of ﬁgure 6, this is exempliﬁed by Baton Rouge, Louisiana. Hip-

hop bands from here ﬁnd unconventional hip-hop music more rewarded,

whereas for rock and niche bands unconventionality is more rewarded.

Together, ﬁgures 5 and 6 show that the shape of the relationship between

conventionality and popularity varies in important ways from one metro-

politan area to the next. We now turn to multilevel regression to investigate

how the shapes of these metro-speciﬁc relationships vary as a function of the

characteristics of the metropolitan areas in which a band is located. For

each musical world, we estimate a varying intercept, varying slope model

in which the effects of unconventionality and unconventionality squared

are allowed to vary by a band’s metro area, thus producing a unique qua-

dratic curve for each. In addition to controlling for the number of genres se-

lected—represented here as an ordered factor—and the characteristics of

the metropolitan area, we include a series of cross-level interaction terms

representing the interaction between the unconventionality terms on the

one hand and the metro-level characteristics on the other. Following Gelman

(2008), continuous predictors are standardized to have a mean of 0 and a

standard deviation of 0.5, while dichotomous predictors are centered around

the mean.

The results of the analysis are shown in table 3. For context, we ﬁrst note

main effects beyond our focal variable, bands’relational unconventional-

ity. We ﬁnd that for the otherwise average band in any given world, popu-

larity is generally associated with lower population density, higher numbers

of college graduates and young people, lower incomes, and greater special-

ization in distinct musical worlds, all else being equal. The rock world stands

out in the extent to which popularity is associated with concentration of the

local record industry. Similarly, we ﬁnd that whereas popularity in the rock

and niche worlds is negatively associated with the size of the nonwhite pop-

ulation, the hip-hop world is unique in the sense that, on average, the racial

composition of the surrounding area has almost no effect on the band’s over-

all popularity. Across all three worlds, we ﬁnd that bands that choose two or

three genres tend to be more popular than those who choose only one. This

last ﬁnding is signiﬁcant in that it is a variant of the typical count data ap-

proach that has been used in the categories literature, where choosing more

total genres would indicate greater unconventionality. On this score, weﬁnd

that across over 2 million American bands there is more of an eclectic imper-

ative (ﬁg. 1C) than a categorical imperative (ﬁg. 1B; see also Pontikes 2012;

Johnson 2017).

Keeping this background in mind, we turn to the relationship between un-

conventionality and popularity. As expected, characteristics of a band’s city

such as population density, racial composition, median household income,

Categorical Conventionality in Music

255

TABLE 3

Varying Intercept, Varying Slope Models of Unconventionality by Musical World

Predictor

Rock

Popularity

Hip-Hop

Popularity

Niche

Popularity

Unconventionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.37*** .35*** 21.94***

(.03) (.02) (.05)

Unconventionality

............................. 23.57*** 21.20*** 24.09***

(.07) (.06) (.08)

Population density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.21* 2.38** 2.34***

(.10) (.13) (.10)

Percentage nonwhite . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.37*** .00 2.27***

(.07) (.07) (.07)

Median household income . . . . . . . . . . . . . . . . . . . . . . . 2.28*** 2.25** 2.32***

(.08) (.09) (.08)

Percentage ages 25–34 . . . . . . . . . . . . . . . . . . . . . . . . . . .27*** .15* .21**

(.07) (.07) (.07)

Percentage college graduates . . . . . . . . . . . . . . . . . . . . . .23** .21* .16

(.09) (.09) (.09)

Percentage college students . . . . . . . . . . . . . . . . . . . . . . .04 2.02 .04

(.04) (.03) (.04)

Metro musical world specialization . . . . . . . . . . . . . . . . 2.17*** 2.27*** 2.16***

(.05) (.04) (.05)

Record industry concentration . . . . . . . . . . . . . . . . . . . . .25** .00 .07

(.09) (.07) (.07)

Music consumption and exhibition concentration . . . . . .02 .04 .04

(.05) (.05) (.05)

Two genres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83*** .39*** .20***

(.02) (.02) (.04)

Three genres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.03*** .53*** .58***

(.02) (.02) (.04)

Unconventionality population density . . . . . . . . . . . . 2.04 2.23*** .17

(.09) (.05) (.13)

Unconventionality

population density . . . . . . . . . . . .20 .40* .62**

(.19) (.17) (.21)

Unconventionality percentage nonwhite . . . . . . . . . . .21** 2.13*** .26*

(.07) (.03) (.11)

Unconventionality

percentage nonwhite . . . . . . . . . .53*** .26* .66***

(.15) (.11) (.17)

Unconventionality median household income . . . . . . 2.10 2.29*** .05

(.09) (.04) (.13)

Unconventionality

median household income . . . . . .05 .38** .44*

(.18) (.14) (.21)

Unconventionality percentage ages 25–34 . . . . . . . . . .01 .16*** .04

(.07) (.03) (.10)

Unconventionality

percentage ages 25–34 ........ 2.21 2.15 2.18

(.15) (.11) (.17)

Unconventionality percentage college graduates . . . . .16 .22*** .39**

(.10) (.04) (.15)

Unconventionality

percentage college graduates . . . .14 .02 .33

(.19) (.15) (.23)

Unconventionality percentage college students . . . . . .04 .03 .03

(.04) (.02) (.06)

Unconventionality

percentage college students . . . . .07 .11* .05

(.08) (.05) (.10)

American Journal of Sociology

256

and record industry composition play an important role in shaping the rela-

tionship between unconventionality and popularity. For any given world,

the instantaneous returns on unconventionality ufor a band from the typical

MSA can be expressed in terms of the estimated marginal effect:

M5^g10 1o^g1kwk



12^g20 1o^g2kwk



u, (1)

where ^g10 refers to the main effect of unconventionality and ^g1krefers to

the cross-level interaction between unconventionality and the kth metro-

level covariate w

, with ^g20 and ^g2ksimilarly deﬁned for unconventionality

squared. To help ﬁx ideas, we begin by focusing on the payoff function for

an otherwise average band, as given by ^g10 12^g20 u. While the value of ^g10

represents the estimated marginal effect of unconventionality for an other-

wise average band that is equally average in its combination of genres, the

value of ^g20 estimates the degree to which the rate of return changes as this

combination of genres gets further from the world-speciﬁc mean. Whether

the average band in a given world beneﬁts from unconventionality is reﬂected

in the direction of ^g10 . In this case, positive effects denote popularity rewards,

TABLE 3 (Continued )

Predictor

Rock

Popularity

Hip-Hop

Popularity

Niche

Popularity

Unconventionality metro musical world

specialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.08 2.10*** 2.13

(.06) (.02) (.08)

Unconventionality

metro musical world

specialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.13 .22** 2.11

(.11) (.07) (.12)

Unconventionality record industry concentration . . . .14 .19*** .16

(.08) (.03) (.11)

Unconventionality

record industry

concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.00 .29** .12

(.17) (.11) (.17)

Unconventionality music consumption and exhibition

concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .01 2.00 .04

(.06) (.03) (.09)

Unconventionality

music consumption and exhibition

concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.03 2.02 2.01

(.12) (.09) (.15)

Observations (bands) . . . . . . . . . . . . . . . . . . . . . . . . . . . 1,014,966 1,002,941 468,872

NOTE.—Results of three separate varying intercept, varying slope models—one for each mu-

sical world. Intercepts are allowed to vary by metro area, as are the slopes for unconvention-

ality and unconventionality

, thus producing a metro-speciﬁc polynomial for each. Observa-

tions (metros) 5332.

*P<.05.

** P<.01.

*** P<.001.

Categorical Conventionality in Music

257

while negative effects denote popularity penalties, with the magnitude of each

given by the magnitude of the corresponding parameter estimate.

Looking at table 3, we ﬁnd that whereas unconventionality tends to pay

off for the average band in the hip-hop world, this is not the case in the rock

and niche worlds, where increased unconventionality comes at a price, as

evidenced by the direction of the corresponding main effect. This is not to

say that unconventionality is without its costs in the hip-hop world or that

unconventionality goes totally unrewarded among rock or niche bands. In-

deed, in all three worlds the relationship between unconventionality and

popularity for an otherwise average band is characterized by an inverted

U shape, as evidenced by the direction of ^g20 , which is consistently negative

from one world to the next. This is in line with results of the GAMs described

above, lending further support to the idea that regardless of the musical world

in which an otherwise average band resides, the DIC hypothesis holds.

What differs based on the musical world in which a band is located is the

point at which these payoffs become penalties, and these models allow us to

be more precise about that threshold. The notion of a world-speciﬁc balanc-

ing point is more than just a loose analogy. For any given world, the esti-

mated balancing point is given by 2^g10 =2^g20. Using this formula, we esti-

mate that for an otherwise average hip-hop band, unconventionality does

not become a liability until the unconventionality score is 0.30 standard de-

viations or more above the world-speciﬁc mean, as compared to 0.38 stan-

dard deviations below the world-speciﬁc mean for the otherwise average rock

band or 0.47 standard deviations below the world-speciﬁcmeanfortheother-

wise average niche band. As the formula for the estimated balancing point

2^g10=2^g20 suggests, these differences are driven not only by differences in

^g10 (i.e., the estimated rate of return for the average band) but by differences

in ^g20. The lower the absolute value of ^g20, the more slowly the estimated rate

of return changes, which is precisely what we observe when comparing the hip-

hop world to its counterparts. We ﬁnd that the absolute value of ^g20 in the hip-

hop world is noticeably smaller than it is in the other two, producing a ﬂatter

popularity curve in a world where the base rate of return on unconventionality

already tends to be higher, with the balancing point pushed up accordingly.

As equation (1) shows, these relationships are affected by characteristics

of a band’s geographic location in a straightforward way: all else being

equal, a 2 standard deviation increase in the value of the kth metro-level co-

variate w

increases the estimated rate of return at the mean by a value of ^g1k

and the effect of uon ^

Mby a value of 2^g2k. The latter quantity is signiﬁcant

insofar as it governs the ﬂatness of the popularity curve. Consider, for exam-

ple, the case of the rock world, where a 2 standard deviation increase in the

size of the nonwhite population in a band’s metro area increases the base

return on unconventionality by a value of 0.21, while simultaneously reduc-

ing the effect of unconventionality on the estimated returns in popularity by

American Journal of Sociology

258

a value of 2 0:53 51:06. In other words, as the size of the nonwhite pop-

ulation goes up, the payoff for unconventionality tends to increase, while the

popularity curve tends to get ﬂatter, suggesting a context in which the value

of the predicted payoff is less sensitive to differences in the level of unconven-

tionality, as in the optimized plateau (ﬁg. 1F). While we observe a similar

pattern in the niche world, things work somewhat differently in the hip-hop

world, where an increase in the size of the nonwhite population tends to reduce

the payoff for unconventionality at the mean. As is the case in the other two

worlds, the popularity curve for the hip-hop world tends to get ﬂatterasthe

size of the nonwhite population grows, although the degree of ﬂattening is less

pronounced, keeping in mind that the popularity curve in the hip-hop world is

generally much ﬂatter to begin with.

To be clear, it is possible for the base rate of return to vary as a function of

metro-level characteristics without inducing comparable changes in the steep-

ness of the underlying popularity curve and vice versa. This is most apparent

in the hip-hop world, where, for example, returns on unconventionality tended

to increase when bands are located in areas with a larger college-educated

population, with the steepness of the popularity curve varying only slightly.

Nonetheless, the cumulative effect of metro-level characteristics tends to be

more pronounced in cases in which both unconventionality and unconven-

tionality squared are subject to interactions, as can be seen in ﬁgure 7, which

depicts the changing pattern of marginal effects that results from the inclu-

sion of the cross-level interactions described above. For each metro-level var-

iable, we examine what happens to the marginal effect of unconventionality

for an otherwise average band as we shift the value of the variable in ques-

tion or the level of unconventionality, keeping in mind that payoff for increas-

ingly unfamiliar pairings depends on the level of unconventionality from which

a band starts. In both cases, the shift from low to high is deﬁned by a shift

from the 10th percentile to the 90th. Lookingat the results for the rock world,

for example, we observe a signiﬁcant difference in the estimated marginal ef-

fect of unconventionality as the size of the nonwhite population shifts from

low to high, as evidenced by the lack of overlap between the corresponding

conﬁdence intervals (CIs).

This is true for both conventional rock bands as

well as their less conventional counterparts.

Four key features stand out here. First, while changes in the metro-level

characteristics may shift the payoff a bandreceives for unconventional genre

blendings, it is universally true that the estimated returns on unconventionality

are positive at low levels of unconventionality and negative at high levels of

While an absence of overlap between two intervals implies a signiﬁcant difference in

the estimated marginal, the presence of overlap does not imply insigniﬁcance. To account

for this, we constructed a 95% conﬁdence band based on the difference in estimated mar-

ginal effects observed across the central 80% of conventionality scores (see ﬁg. C1).

Categorical Conventionality in Music

259

FIG.7.—Estimated marginal effects by unconventionality andmetro-level variable sta-

tus. Point estimates refer to the instantaneous marginal effect of unconventionality, with

the corresponding 95% conﬁdence interval, constructed using the delta method as imple-

mented in the margins package in R (Leeper 2018). For any given variable, low and high

values refer to the 10th and 90th percentile, respectively. Scales are unique to each panel in

order to increase legibility.

unconventionality. This change in direction is consistent with the idea that

the curve depicting the relationship between unconventionality and popu-

larity is characterized by an inverted U shape, providing further support

for the idea that unconventionality is neither universally positive nor nega-

tive. Second, while the relationship between metro-level characteristics and

the payoff for unconventionality differs across worlds in noticeable ways, it

is generally the case that the rewards and penalties associated with uncon-

ventionality are less pronounced in the hip-hop world. Third, the payoffs

for unconventionality are strongly affected by racial composition, regardless

of world. Finally, apart from racial composition, the payoffs for unconven-

tionality in the rock world appear to be less affected by observable metro-level

characteristics than they are in the other two worlds. This is perhaps most ap-

parent when comparing rock to hip-hop, where the returns on unconvention-

ality are clearly affected by not only racial composition but population density,

median household income, and record industry concentration. More speciﬁ-

cally, we ﬁnd that for conventional hip-hop bands, the returns on unconven-

tionality tend to be higher in metro areas that are less densely populated, have

lower levels of median household income, and have less record industry con-

centration, all else being equal. The opposite pattern holds, however, for less

conventional hip-hop bands that tend to be penalized more heavily in these

types of areas than they are in areas that are either more densely populated or

more afﬂuent or feature a more prominent local record industry.

DISCUSSION AND CONCLUSION

Back when she was the lead singer of The Weirdos, Janelle Monáe credited

openness to experimentation in her local Kansas City music scene with giv-

ing her the freedom to “discover”her “inner weirdo.”Using data on the genre

combinations of and fandom for over 2 million bands, we show that, within

the genres in which she was working, Monáe was right. As can be seen inthe

online visualization tool, Kansas City bands were atypically tolerated for be-

ing weirdos, a ﬁnding also predicted by work on music and urban contexts in

which Kansas City is a “crossroad”city in which new genre recombinations

in music have emerged and been distinctively rewarded (Florida and Jack-

son 2010, p. 311). Across the landscape of popular music, Monáe and the

Weirdos were not only exceptionally weird in their genre pairings; they were

also exceptionally rewarded (or at any rate, less penalized) for their uncon-

ventionality in part because of their embeddedness in a particular local con-

text. While Monáe’s eventual superstardom was not predictable, starting out

she was at least the right kind of weird in the right kind of place for her

weirdness to be tolerated or even encouraged.

Our analysis generalizes this insight to over 2 million bands. Bands from

some cities (such as New York or San Francisco) are rewarded for unusual

Categorical Conventionality in Music

261

genre combinations at relatively high rates, whereas those from others (such

as Buffalo) are less rewarded for unconventionality; others (such as St. Louis)

broadly mirror national patterns, while bands from still others (like Baton

Rouge) are rewarded or punished for unconventionality differently depend-

ing on the musical world in question. Further, we examined a range of met-

ropolitan characteristics that associate with a band’s likelihood of being re-

warded or penalized for eclectic musical offerings. Extending a literature on

the geography of popular music, we found that returns on unconventional

musical offerings also vary by the metropolitan characteristics and musical

worlds in which bands are embedded. Very unconventional rock and niche

bands enjoy more tolerance for experimentation when located in racially

diverse metros, while hip-hop bands in cities with a strong record industry

concentration are more rewarded for unconventional genre mixing than

hip-hop bands in cities without. When located in metros with highly educated

populations, unconventional hip-hop and niche bands are more likely toﬁnd

an audience relatively open to unusual genre combinations.

Our analysis reveals large-scale patterns in the relationship between un-

conventionality and popularity, highlighting the fact that the average or

typical pattern is a complex combination of diverse local processes. The

“average effect”is a special case that may in fact be rare. Social reality is in-

delibly marked by contingency and context, but this does not make it un-

knowable. While we have in this article been able to document some sources

of local variation from cities and genre worlds that produce distinctive forms

of relation between unconventionality and popularity, articulating and iden-

tifying the mechanisms by which these forms arise remains an open chal-

lenge. We conclude with reﬂections on what a forward-looking agenda in

search of such mechanisms might look like.

Central to such an agenda is to incorporate a more sociological perspec-

tive into how we conceptualize the social psychological processes identiﬁed

by past research of the balancing act between conventionality and uncon-

ventionality. From Simmel (1957) to Brewer (1991) and beyond, authors point

at a “desire to ﬁt in and stand out”resulting in a “sweet spot.”We add the

sociological proposition that there is no single sweet spot but rather multiple

spots that are contingent on what it means to ﬁt in or stand out in different

contexts. Figure 1 shows some shapes this contextualization may generate,

but additional work is necessary to identify how and why precisely one or

the other arises.

We suggest as an “accounting device”(Griswold 2014, p. 16) four sites that

we believe any prospector should be looking for when digging for explana-

tionsastowhyvarioussweetspotsemerge:(1) categories, (2) producers, (3) con-

sumers, and (4) places. Regarding categories, in advancing explanations of

the types of patterns shown in ﬁgure 1, the researcher may want to know

such things as how old the category is and whether it has differentiated into

American Journal of Sociology

262

avant-garde and commercial poles. Or how popular is the category, what

level of generality does it operate on, and how important is authenticity

within the category? For producers, how generally oriented are they to other

producers versus consumers (e.g., a “musician’s musician”or “comic’scomic”);

do they work full time in the domain or category, or is participation in it part

of a portfolio career; and are they generally compensated through psychic,

status-based, or monetary rewards?The workhorse variablesthrough which

we study cultural tastes such as age, education, childhood arts exposure, di-

versity, and connoisseurship may be the same types of measures that drive

audience’s attraction or repulsion to unconventional combinations. And ﬁ-

nally, places may vary in how much they are deﬁned by a particular category,

if they are stepping-stone places, industry centers,crossroads, immigrant com-

munities, or reputational magnets that draw producers and consumers while

also deﬁning reputations for knowledgeable nonresidents.

We expect that the shapes of the unconventionality-conventionality bal-

ancing acts that we observe will emerge from how features of these four sites

combine. For example, one might consider the degree to whichthe rock world

tended to allow for a greater range of unconventionality with less steep pen-

alties across the middle range of the distribution, while still at this time (as

would later be noted) being less ambitious in innovating newsounds through

recombinant experimentation (Baker 2013; Baroni 2015). This may be be-

cause of a maturity of genres that the rock world contained and a greater

range of conventionally unconventional combinations that are generally ac-

cepted as being legitimately rock-like in theirsounds. Likewise, one could ad-

vance conjectures about the sharper penalty for extreme conventionality in

the niche world. This could occur because to locate oneself outside of the two

major worlds around which popular music orbited in 2007 (rock and hip-

hop) is to already present oneself as catering to an unusual audience; to be

ﬁrmly dedicated to just one niche genre in the niche world could moreover

imply a hyperniche self-presentation not in the market for more popularity

beyond the already inclined (Mark 1998). In turn, the second uptick in pop-

ularity in the hip-hop world ﬁts with future recollective accounts of how mu-

sic was being transformed through genre blending with hip-hop (Richmond

2015; Guan 2017). While anecdotal and based on public perceptions, it is in

fact the outcome that would be predicted by our data from 2007. With the

right longitudinal data, scholars could also look at the trajectories of partic-

ular genres, perhaps with unconventional genre blending being more penal-

ized as scenes form (Wry, Lounsbury, and Glynn 2011) and again as they ma-

ture into traditionalist domains in which preservation takes primacy over

experimentation (Lena 2012)—this would be a genre-speciﬁc temporal ver-

sion of the more spatial inverted U we have documented in this article.

These genre category processes we expect would produce different out-

comes depending on how they interact with mechanisms rooted in audiences,

Categorical Conventionality in Music

263

producers, and places. Thus, part of the decreasing rewards for unusual

genre blending in the rock world could be explained by aging audiences and

amore“traditionalist”orientation to the genre and its communities (Lena

2012). Similarly, part of the trend toward cross-world genre blending in hip-

hop likely arises in part from hip-hop artists’increasing commercial success

creating opportunities for collaboration with other commercially successful

artists from diverse genre communities. And all of these interactions among

genre, producer, and audience can be inﬂuenced by the place in which they

occur. To illustrate, we discuss how two places shift the orientation of audi-

ences and producers toward blending genres in music that emerges there.

First, Nashville provides an intriguing case for building explanations

about shifting rewards for genre classiﬁcations around the distinctive local

characteristics of a place. Returning to our data, for bands in the hip-hop

world from Nashville, the online visualization tool shows that returns to un-

conventionality generally rise, peaking around .9, before dropping. Bands

at this range of conventionality in Nashville often combine country and re-

lated Americana genres with genres such as soul and R&B—for example,

country/soul/rock, alternative/southern rock/soul, and Christian/rock/soul.

Jason Eskridge is a case in point—a musician in our data set who lists soul/

acoustic/folk and was featured in 2014 by a local music blogger as an example

of top soul artists in Nashville ( Jay 2014). This speciﬁc example exempliﬁes

a broader point about the underlying mechanisms at work: with Nashville’s

growing country music industry concentration, it has become home to large

numbers of talented highly professionally competent musicians, attracting

musicians from other genres as well (Florida and Jackson 2010). Even so,

country and related genres still dominate. This combination—a genre spe-

cialist city with a large recording industry base and professional talent pool—

probably makes it more likely that country 1X combinations (which might

be unusual nationally) would occur here and be less penalized for such com-

binations. For example, while there was certainly an “Alternative Country”

virtual scene that positioned itself against “the corporate way of producing

country music centered in Nashville”(Lee and Peterson 2004, p. 202), bands

located in Nashville that make alternative/country are still on average sub-

stantially more popular (mean popularity 52:86, 95% CI [2.13, 3.52]) than

alternative/country bands from elsewhere (mean popularity 52:1, 95% CI

[2.26, .05]). This is one example of how local production and consumption

dynamics can encourage speciﬁc forms of genre combinations and, more

generally, of how place characteristics such as industry concentration and

local genre predominance can help explain why particular sweet spots be-

tween conventionality and unconventionality may be more likely to arise

in some places than others .

Honolulu provides another illustration of how to incorporate place into

explanations of the origins of patterns in U-shaped curves. Like Nashville,

American Journal of Sociology

264

Honolulu is strongly associated with a speciﬁc music genre (Hawaiian). But

Honolulu is less of an industry center and more of a meeting point of diverse

ethnocultural traditions along with a local cultural economy strongly shaped

by tourism. The result is a comparatively ﬂat and elongated relationship be-

tween unconventionality and popularity across the middle of the distribution

for the niche world. This may be because in Honolulu there is a long history

of the types of genre blending with Hawaiian music that are unconventional

on the national stage. For instance, the genre blending of reggae and Hawai-

ian music (e.g., replacing the guitar with a ukulele) is a robust phenomenon

in Hawaii (Kale 2017), and the blending of Hawaiian music with country

music dates to the 1920s with Sol Hoopii and his lap steel slide guitar (Silva

2017). In fact, Honolulu, a relatively small metro area, accounts for roughly

one-quarter of all bands who chose combinations of Hawaiian, reggae, and

country genres, andHonolulu bands with these combinations are on average

substantially more popular (mean popularity 5:31, 95% CI [.09, .52]) than

bands with these combinations elsewhere (mean popularity 52:68, 95%

CI [2.83, 2.53]). The prevalence of Hawaiian music there likely means that

genres that diffused to Hawaii from other places (like reggae and country)

have been adapted to blend with the “home”sound of the local scene (Ben-

nett 2000) and that audiences—some of whom are tourists looking for local

sounds—give Hawaiian bands from Honolulu more room for unusual com-

binations with Hawaiian music than from elsewhere. Overall, scaling down

in this way into smaller genre communities or smaller metropolitan scenes

could better allow for teasing out the relationship between psychological ex-

planations and mesolevel sociological mechanisms. Even so, the typology we

formulated in table 2 could be used to make relatively general hypotheses

about how local cultural scenes make particular curves more or less likely in

experimentalist or traditionalist places.

To be clear, while our observations about places like Nashville and Ho-

nolulu are to some degree speculative, they show how for scholars of urban

studies, culture, and creativity our ﬁndings and the accompanying online vi-

sualization tool can serve as a ﬁrst step “thin description”(Spillman 2014)

that precedes the thick description of qualitatively investigating meaning

making, mechanisms, and processes within local contexts or within speciﬁc

genres and the family ofgenres they are most combinedwith. Such may also

be true for online spaces (Peterson and Bennett 2004), which may be more

shaped by categories, producers, and consumers than they are by geographic

places, or as places themselves they may have their own features that drive

the local relationship to rewarding unconventionality. This suggests that in

the long-run, qualitative, historical, and networks-based studies (e.g., Gra-

zian 2005; Lena 2012; Cornﬁeld 2015; Crossley 2015; Skaggs 2019) may

be more equipped than our more macrolevel approach to tease out local

dynamics.

Categorical Conventionality in Music

265

We believe the approach developed in this article is highly portable to

other domains. For example, using data from Artprice and ArtFacts, global

artists could be identiﬁed as operating within worlds through the media and

movements in which they work and then indexed for conventionality within

those worlds while also being geographically located in cities or countries. As

in Buchholz (2018), with the use of these data sources, popularity could then

be measured two ways: through both auction sales and gallery exhibitions.

Some cities or countries may produce more popular unconventional artists

than others, just as unconventionality may be rewarded differently in the

auction market and gallery market (relatedly, see van Venrooij and Schmutz

2018) or in some artistic worlds over others.

As another example, using American Sociological Association section

membership data the researcher could index the conventionality of multiple

section memberships across each member (e.g., Culture and Theory section

membership may be highly conventional, whereas Culture and Rationality

and Society may not) and then nest members in their departments. From

this, as in our typology of cities, one could then create a typology of depart-

ments, potentially with conventionalists, normalists, traditionalists, and ex-

perimentalists classiﬁed by the conventionality of their members’section

memberships. Going a step further, one could then index the conventional-

ity of section memberships for participants in the employment fair, model-

ing over time whether unconventional job market candidates are more likely

to be employed by experimentalist departments than normalist or tradition-

alist departments. Like the example of the global art market, while porting

the basic contours of our approach to this setting may introduce a new set of

challenges, it may also have substantial payoffs.

Overall, we see this work as contributing to multiple literatures. For the

categories literature we hope our work serves as another vote in several re-

cently emergent trends. First, we hope this work serves as another call to

study the effects of categories on popularity beyond the scope of the categor-

ical imperative (Zuckerman 2017). Second, we believe this article signals a

direction in which research can head, as work on categories can be brought

back into conversation with a wider swath of sociological literature extend-

ing back to Simmel, which treats market actors as balancing between two

competing imperatives. With limited exceptions, work in the categories lit-

erature mostly operates along the ﬁrst row of proposed relationships in ﬁg-

ure 1. We additionally hope this work serves as a call for the further inves-

tigation of both nonlinear and nonmonotonic functions in the relationship

between categorical unconventionality and popularity. To do so would re-

quire treating categories as positioned in relational space, which along with

others (Kovács and Hannan 2015; Wry and Castor 2017; van Venrooij and

Schmutz 2018) we feel is a fruitful way to move forward in this line of re-

search. With regard to category assignments, given the varied purposes that

American Journal of Sociology

266

genre designations serve (DiMaggio 1987; Lena 2012; Negus 2013), while

our data are on producer’s self-claims, future research could home in on cases

in which there are classiﬁcatory mismatches across different classifying groups

(e.g., producers and consumers).

For the sociology of culture, this work brings further evidence to the en-

durance of regionalism in cultural American life (Griswold and Wright

2004; Leschziner 2007; Griswold and Wohl 2015) and even in virtual com-

munities (Peterson and Bennett 2004; Allington 2014). For work on urban

culture and the geography of music, this article makes several contribu-

tions. While most quantitative research in this area relies on industry and

ofﬁcial government statistics (Florida, Mellander, and Stolarick 2010), we

show the potential in using other data sources to capture the full spectrum

from local grassroots garage bands and solo mix-tape artists to superstars.

With so many bands invisible to ofﬁcial sources, we are able to capture dis-

tinctive local musical contexts such as those in Buffalo, Baton Rouge, and

St. Louis, which are too often overlooked for industry-based studies of New

York, Los Angeles, and Nashville. Moreover, we create a new measure of

local musical conventionality and unconventionality, which can be added

to more typical measures such as industry concentration. While in this work

we focus on the level of musician’s and band’s self-presentation, we hope

that future work merges our approach with exciting work happening at

the level of discrete objects such as songs (e.g., de Laat 2014; Askin and

Mauskapf 2017). Likewise, while we focus on the relationship between con-

ventionality and popularity as it relates to spatial and relational communi-

ties, we believe that other axes of variation such as artist sociodemographics

(e.g., de Laat 2019), location in a genre’s trajectory (Lena 2012), or industry

conditions embedded in different historical time periods (e.g., Peterson and

Berger 1975; Lopes 1992; Dowd 2004) might provide additional sites of

meaningful variation. In the same vein, future researchers could investigate

the degree to which a genre or particular combinations of genres (or fandom

for genres) is embedded within a particular locale versus being diffuse with-

out a geographic “home base,”perhaps documenting the degree to which

different musics and their fandoms live locally, translocally, or virtually

across geographic divides (Peterson and Bennett 2004). Measurement tech-

niques from economic geography such as location quotients might be use-

fully adapted to this end.

Finally, we show how big data sources can be used to advance older the-

oretical ideas, such as the Chicago school’s (both new and old) sensitivity to

local context and variation amid more general patterns (Sampson 2012),

while also illustrating what a forensic approach to big data can look like

(McFarland et al. 2016). By using a suite of quantitative techniques for con-

textualizing and visualizing local distinctiveness, we hope to bring sensibil-

ities traditionally associated with qualitative research to larger-scale studies,

Categorical Conventionality in Music

267

while at the same time encouraging qualitative and local researchers to sit-

uate their ﬁndings and cases comparatively, against other U.S. metropolitan

locations. Facilitating this kind of comparative dialogue between local and

general is one of the true advantages of big data. Our study shows the poten-

tial it holds.

American Journal of Sociology

268

APPENDIX A

TABLE A1

Genres by World

Hip-Hop Rock Niche

Club Acoustic Folk Rock Powerpop A’Cappella Dub Hawaiian Progressive House Visual

Crunk Alternative Funk Progressive Acousmatic Dutch Pop Healing & Easy

Listening

Psychobilly Western

Swing

Freestyle Ambient Fusion Psychedelic Afro-beat Electro House Regional Mexican Zouk

Hip-Hop Americana Garage Post-punk Big beat Electronica IDM Samba

Hyphy Bluegrass Gospel Punk Black Metal Emotronic Idol Spanish Pop

Latin Blues Grunge Religious Bossa Nova Flamenco Industrial Shoegaze

Salsa Christian Hardcore Rock Breakbeat French Pop Italian Pop Showtunes

Lyrical Classic Rock Indie Rockabilly Breakcore German Pop Japanese Classic Tango

Neo-soul Classical & Opera Jam Band Roots Music Celtic Ghettotech J-Pop Tape Music

R&B Comedy Jazz Screamo Christian Rap Glam Jungle Techno

Rap Country Lounge Ska Concrete Gothic K-Pop Thrash

Reggae Electroacoustic Metal Southern Rock Death Metal Grime Live Electronics Trance

Reggaeton Emo New Wave Surf Disco House Grindcore Melodramatic Popular Trip Hop

Soul Experimental Pop Swing Downtempo Happy

Hardcore

Minimalist Tropical

Folk Pop Punk Drum & Bass Hard House Nu-Jazz Turntablism

Categorical Conventionality in Music

269

APPENDIX B

Deﬁning Genre Unconventionality

Extending Lizardo’s (2014) measure of “effective omnivorousness,”we

measure genre unconventionality as follows. Given an npadjacency ma-

trix Adepicting the relationship between bands and genres, the degree of

numerical overlap between any given pair of genres jand kis given by

C5A0A. We use this information to construct a matrix of similarity scores

O, where

ojk 5cjk

min cjj,ckk

ðÞ

:(B1)

The resulting scores range from 0 to 1, with higher scores indicating a more

conventional pairing, much like a Jaccard coefﬁcient. The chief difference is

that whereas Jaccard-based measures standardize the value of c

by divid-

ing by the total number of unique entities to which jand kare collectively

tied (i.e., cjj 1ckk 2cjk), our measure standardizes the value c

by dividing

by the minimum value of c

and c

. This captures the degree of overlap

from the perspective of the smaller of the two genres.

The advantage of

this approach over the conventional Jaccard measure is that it avoids the

tendency to conﬂate inequity in the magnitude of c

and c

(i.e., the number

of entities tied to jand k, respectively) with a lack of overlap or clustering

(Latapy, Magnien, and Del Vecchio 2008).

Our approach diverges from Lizardo (2014) in two important ways. First,

instead of working with o

, which measures the degree of similarity be-

tween a given pair of genres, we focus instead on the dissimilarity given

by djk 512ojk. The resulting scores still range from 0 to 1, but higher scores

indicate more unconventional pairings. Second, instead of making a band’s

unconventionality score cumulative, adding to its score when more genre

choices are made, we take the mean of each band’s genre-pair unconvention-

ality. In effect each genre pairing receives an unconventionality score, and a

band’s overall unconventionality score is the mean of each genre pair it

chooses. We take means instead of sums because bands can self-select the

number of genres (up to three total), so the very number of genres listed (re-

gardless of how unusual any pairing is) can affect the score. As a result, if we

summed across all pairs, a band would get a higher score for describing itself

as “pop/rock/alternative”than as “grindcore/ghettotech/[blank].”

As a robustness check, we compared our results to the results produced using a number

of Jaccard-based measures. We found that, as expected, unconventionally scores tended

to cluster near 1 when using a Jaccard coefﬁcient to measure the degree of overlap be-

tween genres. Nonetheless, our results do not depend on the choice of measure. More spe-

ciﬁcally, we ﬁnd that the relationship between popularity and conventionality is charac-

terized by an inverted U shape, regardless of which measure we use. The results of our

multilevel regression are similarly unaffected.

American Journal of Sociology

270

Taking the mean of genre pair unconventionality scores avoids this prob-

lem. Accordingly, the unconventionality score uassociated with band iis

given by

ui5oj∈NiðÞ

ok∈NiðÞ

djk

mimi21ðÞ

, (B2)

where N(i) refers to the set of all genres chosen by band i(which Lizardo

refers to as the “cultural neighborhood”), and mi5ojaij refers to the total

number of genres selected by band i. In our regression models, we also sep-

arately control for the number of genres selected, in an effort to account for

the effect of adopting multiple genre labels, thus allowing the effects asso-

ciated with the number of genres to differ from the effects associated with

the particular combination of genres adopted.

Unconventionality scores are calculated using a band’s complete set of

genres. A band counts as part of a musical world so long as at least one of

its genres is in that world. If a band lists one genre in the rock world and

two in the hip-hop world, all three MySpace genres are considered in the cal-

culation for the score for both worlds. For bands whose genres span across

multiple worlds, we calculate separate scores for them for each world. The

fact that a band can belong to more than one world means that a band can

also have more than one unconventionality score, with the level of uncon-

ventionality depending on the world in which it is being viewed. By way of

example for the value of this approach, a band like Rage against the Machine

lists its genres as rock/rap/punk. This mirrors the on-the-ground reality of

Rage against the Machine self-identifying with and actively participating

in both the rock and hip-hop worlds: both its stated musical inﬂuences

(Black Sabbath, the Sex Pistols, Public Enemy, Cypress Hill) and its concert

touring (U2, Wu-Tang Clan) traverse these worlds (Everley 2017). In the

hip-hop world, it is a band with rap inﬂuences that also participates in rock

and punk genres, whereas in the rock world it is the opposite. Our measure

leaves open the real world possibility that the unconventionality of Rage

against the Machine is interpreted differently when performing with U2 and

when performing with Wu-Tang Clan, while also not enforcing that differ-

ence (e.g., the rock and rap worlds are allowed to be associated with genre

spanning combinations differently, but they are not forced to). Our measure

reﬂects these differences: in the rock world, Rage against the Machine’s genre

combination is conventionally unconventional, lying around the 50th per-

centile of rock world unconventionality. However, the same combination was

near the 83% of unconventionality in the hip-hop world. This, we believe,

better captures both the theoretical underpinnings of our approach and how

the real world operates.

As an alternative procedure and robustness check we reran our procedures with bands

assigned exclusively to the world of their ﬁrst self-reported genre designation, which did

Categorical Conventionality in Music

271

APPENDIX C

Signiﬁcance Tests for Nonlinear Interactions

Depicting the results of a nonlinear interaction is notoriously difﬁcult be-

cause the effect of interest differs as a function of both the value of the cor-

responding variable as well as the value of the variables that it is allowed to

interact (see Mize [2019] for a recent review). To address this problem, we

examined how the estimated marginal effect of unconventionality changed

when we changed the level of unconventionality or the level of a given metro-

level variable, focusing on the contrast between the 10th and the 90th per-

centile of each. As can be seen in ﬁgure 7, there is little doubt that if we hold

the value of the metro-level variable constant, the marginal effect of uncon-

ventionality at the 10th percentile of unconventionality is signiﬁcantly differ-

ent from what it is at the 90th percentile. What is less clear is whether the

estimated marginal effect of unconventionality differs signiﬁcantly across

levels of each metro-level variable, holding unconventionality constant.

As a ﬁrst approximation, we focused on the presence of nonoverlapping

intervals. The difﬁculty with this approach is that while a lack of overlap

between two intervals implies a signiﬁcant difference, the presence of over-

lap does not imply insigniﬁcance. Rather than comparing intervals, a more

appropriate approach is to construct an interval around the estimated dif-

ference. Toward this end, we used the delta method to construct a 95% conﬁ-

dence band based on the difference in estimated marginal effects that results

when we shift the level of a given metro-level variable from the 10th percen-

tile to the 90th. The results are shown in ﬁgure C1, which depicts the estimated

difference observed across the central 80% of conventionality scores for each

metro-level variable. A positive difference indicates that the estimated mar-

ginal effect of unconventionality at the 90th percentile of the metro-level

variable in question is larger than it is at the 10th percentile of that same

metro-level variable. A negative difference, however, indicates thatthe mar-

ginal effect of unconventionality at the 10th percentile is larger than it is at

the 90th percentile, with no difference denoted by the dashed line. When the

dashed line falls within the bounds of the conﬁdence band, it means that the

estimated difference in the marginal effects is signiﬁcant at that point.

not meaningfully change our results. By way of further exploration, we created a variable

for bands who are “purists,”where purists only select MySpace genres within a given

world. Overall, rock purists are more popular than world-crossing rock world bands,

whereas niche and hip-hop purists are less popular. This is likely because rock world mu-

sicians are more popular in general. Controlling for purists in our regression models does

not substantively alter our results.

American Journal of Sociology

272

FIG. C1.—Differences in estimated marginal effects

APPENDIX D

Constructing Musical Worlds

We follow a two-stage strategy to partition MySpace data into musical

worlds, as discussed in more detail in Silver et al. (2016), some of the text

from which we reproduce here. A ﬁrst step is to reject the null hypothesis

that there is a completely random relationship among a band’s genre choices.

Whatever a band chooses for genre 1 would be arbitrarily related to its choice

for genres 2 and 3 and vice versa. This is admittedly an unlikely scenario, but

it does provide a useful baseline. Using a community detection (modularity

optimization) algorithm, we ﬁnd that genre choices are far from random. Cer-

tain genres are paired with one another with great consistency. To determine

this, we catalog all band-supplied genre combinations as a network deﬁned

by the frequency with which bands coselect them. For instance, if one band

chooses rap and metal, there would be one edge between the rap and metal

nodes, and so on, with numerical frequency for rap and metal and all other

activated genre-by-genre ties.

Given that bands’genre choices do evince latent structural patterns, the

second step is to examine the nature of these patterns. By applying modu-

larity clustering to the MySpace genre hierarchically—that is, repeatedly

subdividing genre communities until it is impossible to do so again with sta-

tistical signiﬁcance—we are able to characterize in greater detail the MySpace

universe’s organization of musical genres. We ﬁnd a fundamental ﬁrst divi-

sion among rock, hip-hop, and niche musical worlds that breaks down fur-

ther into 16 distinct genre communities. The current study focuses on the

highest-order division (rock, hip-hop, niche) for the sake of simplicity and

clarity—we are examining variations between three worlds and hundreds

of metropolitan areas. Nevertheless, examination of the 16 lower-order com-

munities is an important further area of research and a useful way to under-

stand the meaning of the higher-order worlds.

To ﬁnd the latent structure of MySpace bands’genre selections, we create

a large and complex network by mapping the band-provided colistings of

(up to three) genres. Genres are considered “related”once when a band lists

them together. In the event that a musician chooses only one genre—thereby

providing no information about how genres are associated with one an-

other—the musician’s choices are not includedin the analysis. To make sure

this did not bias our results, we compared the distribution of genre nomina-

tions for genres that are listed alone versus genres that are listed in some sort

of combination. We found that the distributions do not differ substantially.

Thus, eliminating single genre selections from our analysis does not bias the

resulting genre clusters in any signiﬁcant way, nor would providing some

kind of unique score for single genres change the combinatory patterns that

we currently see.

American Journal of Sociology

274

For the full population of bands in the data that did list more than one

genre designation, greedy modularity optimization is employed to identify

genre communities. Greedy modularity optimization was developed by Clau-

set, Newman, and Moore (2004). This algorithm partitions a network by max-

imizing its modularity, a measure that quantiﬁes a network’s community struc-

ture by providing a value for every clustering within a given graph. The general

idea is to employ a random graph on the same vertex set that does not have

any community structure and compare the edge density of the clusters in the

original graph with the edge density of the clusters in the random graph. The

greater the difference between the two edge densities, the more community

structure the given clustering describes. We use the version operationalized

in R’s IGraph package, which outputs the best community structure (struc-

ture with the highest modularity score) possible.

But modularity algorithms, like most clustering algorithms, have no uni-

versally accepted signiﬁcance tests. In other words, there is no consensus as

to whether a modularity score of .1, .3, or any value indicates a “real”versus

an arbitrary community structure. In certain situations, however, it can be

relatively easy to apply statistical techniques that approximate a signiﬁ-

cance test. While it is unclear how we would deﬁne, let alone test, the sig-

niﬁcance of the entire community structure discovered in this study, it is

relatively straightforward to test whether a single identiﬁed community is

signiﬁcantly structurally “separate”from the rest of the large network. This

can be done with a Wilcoxon rank-sum test, which, applied to this context,

assesses whether there is a statistically signiﬁcant difference between the

number of in-edges and out-edges adjacent to members of a given commu-

nity. If a community has signiﬁcantly more in-edges than out-edges, the com-

munity is considered statistically signiﬁcant—a relatively uniﬁed group of

genres with relativelystrong boundaries. And if all its constituent communi-

ties are signiﬁcant, it is reasonable to consider an entire community structure

statistically signiﬁcant.

In this study, running the modularity optimization algorithm and signif-

icance testing was procedurally united. In order to identify the most speciﬁc

genre complexes possible, we do not simply run the modularity optimiza-

tion once. Instead, we run the modularity optimization and the rank-sum

test in direct succession and progressively until further dividing a commu-

nity into smaller, more speciﬁc groupings no longer yields statistically sig-

niﬁcant communities. Our results therefore present a community structure

in which all identiﬁed communities are indivisible into smaller signiﬁcant

communities and are themselves signiﬁcant at the P<:01 level.

We used a modularity-based approach to community detection because our

primary interest is to identify areas of density in a graph composed of weighted

and undirected edges. Our data are simple: musicians select genres, and genres

are considered related when they are coselected by many musicians. Areas of

Categorical Conventionality in Music

275

density therefore represent in a clear and straightforward way groups of genres

that are commonly associated with one another across the millions of musi-

cians in our sample. Given the nature of our data, structure comes from edge

weights, so the community detection algorithm chosen must be able to work

with edge weights. Since our edges are undirected, the community detection

algorithm must be chosen accordingly. Modularity-based approaches are a pri-

mary example of internal density approaches that operate on weighted, undi-

rected edges (Coscia, Giannotti, and Pedreschi 2011); here we use Igraph’sim-

plementation of greedy modularity optimization.

Figure D1 is a dendrogram that details how a ﬁrst-order clustering into

rock, hip-hop, and niche music worlds is broken down into 16 genre com-

munities, with provisional names to capture their main tendencies. For ev-

ery division, the community’s in-edges outnumber its out-edges at a statis-

tically signiﬁcant level (P<:01). Table D1 summarizes the MySpace genres

within each community.

We performed a placebo test to ensure that the modularity observed in

the MySpace network is not due simply to network density, randomly re-

wiring the network 1,000 times. While the modularity coefﬁcient for the

FIG. D1.—Summary of a dendrogram of MySpace genres with 16 genre communities

at its leaves.

American Journal of Sociology

276

actual MySpace network is 0.31, the average modularity for our 1,000 random

simulations is only .04. This very dense network exhibits almost no modu-

larity at all when its edges are randomly allocated. Thus, the clustering pat-

terns we observe in the MySpace network are highly unlikely to be due to ran-

dom chance.

Progressive modularity clustering moves from the left to right of the den-

drogram in ﬁgure D1. Furthest on the left, the ﬁrst-order breakdown sepa-

rates out three main worlds of popular music—rock, hip-hop, and niche

worlds. As we move to the right, the rock world breaks down to its “subcul-

tural”varieties, which we call countercultural rock, mainstream rock, and

punk offshoots, and then ﬁner categories therein. The hip-hop world, dom-

inated by rap, hip-hop, and R&B, breaks down no further into statistically

signiﬁcant communities. They comprise both a major musical world and an

“end”community. The niche world represents a variety of less popular (as

deﬁned by the frequency with which they are selected by musicians) genres

and communities. It is essentially a category encompassing music not strongly

TABLE D1

Genres within Communities

Genre Community MySpace Genre

Hip-hop Club, Crunk, Freestyle, Hip Hop, Hyphy, Latin, Lyrical,

Neo-soul, R&B

Avant-garde Ambient, Classical & Opera, Comedy, Experimental,

Electroacoustic, New Wave, Progressive, Psychedelic

Electronic/dance Breakbeat, Downtempo, Drum & Bass, Dub, Electro, IDM,

Tropical

Extreme metal Black Metal, Death Metal, Gothic, Grindcore, Thrash

Good old boy Americana, Bluegrass, Country, Rockabilly, Roots Music,

Southern Rock

Jammy Blues, Classic Rock, Funk, Fusion, Jam Band, Jazz, Lounge,

Swing

Japanese Healing & Easy Listening, Idol, Japanese Classic, Melodramatic

Popular

Keeping the beat alive A’Cappella, Afro-beat, Big beat, Christian Rap, Disco House,

Nu-Jazz

Other Ghettotech, Grime, Hawaiian, Regional Mexican, Showtunes,

Western Swing, Zouk

Pop/rock Pop, Powerpop, Rock, Alternative, Indie

Punk metal Emo, Screamo, Hardcore, Metal

Punk rock Garage, Grunge, Pop Punk, Punk, Ska, Surf

Rave Acousmatic, Electronica, Hard House, House, Industrial,

Techno, Progressive, House, Trance, Trip Hop

Spiritual Acoustic, Folk, Folk Rock, Christian, Gospel, Religious

Underground club Glam, Happy Hardcore, Jungle, Psychobilly, Shoegaze,

Turntablism, Visual

World music Bossa Nova, Breakcore, Celtic, Concrete, Dutch Pop, Emotronic,

Flamenco, French Pop, German Pop, Italian Pop, J-Pop,

K-Pop, Live Electronics, Minimalist, Samba, Spanish Pop,

Tango, Tape Music

Categorical Conventionality in Music

277

tied to the two dominantpoles of American music in the 2000s: rock and hip-

hop. These include most notably electronic music genres, dark or extreme metal,

and various underground and world music genres, which emerge and fur-

ther subdivide as we move to the right of the chart.

As an alternative procedure and robustness check, we reran our analysis

using these 16 communities and treating them as the “genres,”using the same

unconventionality formula as described in appendix B. In other words, we

examined degree of conventionality across these 16 “umbrella”categories

that operate at the same level and then plotted the resulting variable against

band popularity in ﬁgure D2. This approach accounts for the possibility that

the “balancing act”we observe is an artifact of differences in the number of

genres within worlds by treating lower-order genre communities themselves

as the relevant musical units.

FIG. D2.—Relationship between unconventionality and popularity, where unconven-

tionality is calculated with each of the 16 genre communities being treated as “genres.”

American Journal of Sociology

278

As can be seen, this does not meaningfully change our results for the basic

inverted U between unconventionality and popularity, which we also con-

ﬁrmed with two-line tests. Similarly, we examined how the relationship be-

tween this measure of unconventionality and popularity shifts across metro

areas. As ﬁgure D3 shows, we ﬁnd substantially similar basic patterns of

association, with race, income, and record industry concentration showing

signiﬁcant effects.

FIG. D3.—Variation in returns on unconventionality by metro-level characteristics, for

unconventionality calculated using 16 genre communities as “genres.”

Categorical Conventionality in Music

279

REFERENCES

Accominotti, Fabien. 2009. “Creativity from Interaction: Artistic Movements and the

Creativity Careers of Modern Painters.”Poetics 37 (3): 267–94.

Allington, Daniel. 2014. “The Geography of SoundCloud: Who’s Following Whom?”

Valuing Electronic Music. https://valuingelectronicmusic.org/2014/09/08/geography

-soundcloud-following/. Accessed February 25, 2019.

Alston, Jon P. 1989. The American Samurai: Blending American and Japanese Manage-

rial Practices. Berlin: de Gruyter.

Askin, Noah, and Michael Mauskapf. 2017. “What Makes Popular Culture Popular?

Product Features and Optimal Differentiation in Music.”American Sociological Re-

view 82 (5): 910–44.

Baker, Ernest. 2013. “Rock Music Sucks Now and It’s Depressing.”Pigeons and Planes,

August 27.

Baker, George P. 1992. “Beatrice: A Study in the Creation and Destruction of Value.”

Journal of Finance 47 (3): 1081–119.

Baroni, Nastassia. 2015. “Billy Corgan Says Modern Rock Isn’t Ambitious Enough.”

MusicFeeds, February 19.

Beck, Nathaniel, and Simon Jackman. 1998. “Beyond Linearity by Default: Generalized

Additive Models.”American Journal of Political Science 42 (2): 596–627.

Becker, Howard S. 1982. Art Worlds. Berkeley and Los Angeles: University of California

Press.

Bennett, Andy. 2000. Popular Music and Youth Culture: Music, Identity and Place. Lon-

don: Macmillan.

———. 2017. Music, Space and Place: Popular Music and Cultural Identity. Cam-

bridge: Routledge.

Bourdieu, Pierre. 1993. The Field of Cultural Production: Essays on Art and Literature.

New York: Columbia University Press.

Brewer, Marilynn B. 1991. “The Social Self: On Being the Same and Different at the

Same Time.”Personality and Social Psychology Bulletin 17 (5): 475–82.

Buchholz, Larissa. 2018. “Rethinking the Center-Periphery Model: Dimensions and Tem-

poralities of Macro-Structure in a Global Field of Cultural Production.”Poetics 71:18–

32.

Burt, Ronald S. 1976. “Positions in Networks.”Social Forces 55 (1): 93–122.

Campa, Jose Manuel, and Simi Kedia. 2002. “Explaining the Diversiﬁcation Discount.”

Journal of Finance 57 (4): 1731–62.

Carroll, Glenn R., Mi Feng, Gaël Le Mens, and David G. McKendrick. 2010. “Organi-

zational Evolution with Fuzzy Technological Formats: Tape Drive Producers in the

World Market, 1951–1998.”Pp. 203–33 in Categories in Markets: Origins and Evolu-

tion, edited by Greta Hsu, Giacomo Negro, and O

zgecan Kocak. Research in the So-

ciology of Organizations 31. Bingley: Emerald.

Cashmore, Pete. 2006. “MySpace, America’s Number One.”Mashable, July 11.

Cennamo, Carmelo, and Juan Santalo. 2013. “Platform Competition: Strategic Trade-

Offs in Platform Markets.”Strategic Management Journal 34 (11): 1331–50.

Chang, Jeff. 2005. Can’t Stop Won’t Stop: A History of the Hip-Hop Generation. New

York: St. Martin’s.

Childress, Clayton. 2017. Under the Cover: The Creation, Production, and Reception of a

Novel. Princeton, N.J.: Princeton University Press.

Church, Kenneth Ward, and Patrick Hanks. 1990. “Word Association Norms, Mutual

Information, and Lexicography.”Computational Linguistics 16 (1): 22–29.

Clauset, Aaron, Mark E. J. Newman, and Cristopher Moore. 2004. “Finding Community

Structure in Very Large Networks.”Physical Review E 70 (6): 066111.

Clemens, Elisabeth S., and James M. Cook. 1999. “Politics and Institutionalism: Explain-

ing Durability and Change.”Annual Review of Sociology 25 (1): 441–66.

American Journal of Sociology

280

Cornﬁeld, Daniel B. 2015. Beyond the Beat: Musicians Building Community in Nash-

ville. Princeton, N.J.: Princeton University Press.

Coscia, Michele, Fosca Giannotti, and Dino Pedreschi. 2011. “A Classiﬁcation for Com-

munity Discovery Methods in Complex Networks.”Statistical Analysis and Data

Mining 4 (5): 512–46.

Crossley, Nick. 2015. Networks of Sound, Style and Subversion: The Punk and Post-

punk Worlds of Manchester, London, Liverpool and Shefﬁeld, 1975–80. Oxford: Ox-

ford University Press.

Crossley, Nick, and Wendy Bottero. 2015. “Music Worlds and Internal Goods: The Role

of Convention.”Cultural Sociology 9 (1): 38–55.

Deephouse, David L. 1999. “To Be Different, or to Be the Same? It’s a Question (and

Theory) of Strategic Balance.”Strategic Management Journal 20 (2): 147–66.

de Laat, Kim. 2014. “Innovation and Diversity Redux: Analyzing Musical Form and Con-

tent in the American Recording Industry, 1990–2009.”Sociological Forum 29 (3): 673–

97.

———. 2019. “Singing the Romance: Gendered and Racialized Representations of Love

and Postfeminism in Popular Music.”Poetics 77. https://doi.org/10.1016/j.poetic.2019

.101382.

DiMaggio,Paul. 1987. “Classiﬁcation in Art.”American SociologicalReview 52 (4): 440–55.

Dowd, Timothy J. 2004. “Concentration and Diversity Revisited: Production Logics and

the U.S. Mainstream Recording Market, 1940–1990.”Social Forces 82 (4): 1411–55.

Dowd, Timothy J., Trent Ryan, Vaughn Schmutz, Dionne Parris, Ashlee Bledsoe, and

Dan Semenza. 2021. “Retrospective Consecration beyond the Mainstream: The Crea-

tion of a Progressive Rock Canon.”American Behavioral Scientist 65 (1): 116–39.

Durand, Rodolphe, Nina Granqvist, and Anna Tyllström. 2017. “From Categories to

Categorization: A Social Perspective on Market Categorization.”Pp. 3–30 in From

Categories to Categorization: Studies in Sociology, Organizations and Strategy at

the Crossroads, edited by Rodolphe Durand, Nina Granqvist, and Anna Tyllström.

Research in the Sociology of Organizations 51. Bingley: Emerald.

Ellis, Diane C., and John C. Beresford. 1994. “Trends in Artist Occupations, 1970–1990.”

Research division report 29. National Endowment for the Arts, Washington, D.C.

Everley, Dave. 2017. “‘We Were a F**king Livewire’: 25 Years of Rage against the Ma-

chine.”Louder, May 30.

Fischer, Claude S. 1995. “The Subcultural Theory of Urbanism: A Twentieth-Year As-

sessment.”American Journal of Sociology 101 (3): 543–77.

Fleming, Lee. 2001. “Recombinant Uncertainty in Technological Search.”Management

Science 47 (1): 117–32.

Florida, Richard, and Scott Jackson. 2010. “Sonic City: The Evolving Economic Geog-

raphy of the Music Industry.”Journal of Planning Education and Research 29 (3):

310–21.

Florida, Richard, Charlotta Mellander, and Kevin Stolarick. 2010. “Music Scenes to Mu-

sic Clusters: The Economic Geography of Music in the US, 1970–2000.”Environment

and Planning A 42 (4): 785–804.

Frith, Simon. 1998. Performing Rites: On the Value of Popular Music. Cambridge, Mass.:

Harvard University Press.

Gärdenfors, Peter. 2004. Conceptual Spaces: The Geometry of Thought. Cambridge,

Mass.: MIT Press.

Gelman, Andrew. 2008. “Scaling Regression Inputs by Dividing by Two Standard Devi-

ations.”Statistics in Medicine 27 (15): 2865–73.

Goldberg, Amir. 2015. “In Defense of Forensic Social Science.”Big Data and Society

2 (2): 1–3.

Goldberg, Amir, Sameer B. Srivastava, V. Govind Manian, William Monroe, and Chris-

topher Potts. 2016. “Fitting in or Standing Out? The Tradeoffs of Structural and Cul-

tural Embeddedness.”American Sociological Review 81 (6): 1190–222.

Categorical Conventionality in Music

281

Granqvist, Nina, Stine Grodal, and Jennifer L. Woolley. 2013. “Hedging Your Bets: Ex-

plaining Executives’Market Labeling Strategies in Nanotechnology.”Organization

Science 24 (2): 395–413.

Grazian, David. 2005. Blue Chicago: The Search for Authenticity in Urban Blues Clubs.

Chicago: University of Chicago Press.

Griswold, Wendy. 2014. Cultures and Societies in a Changing World, 4th ed. Thousand

Oaks, Calif.: Sage.

Griswold, Wendy, and Hannah Wohl. 2015. “Evangelists of Culture: One Book Pro-

grams and the Agents Who Deﬁne Literature, Shape Tastes, and Reproduce Region-

alism.”Poetics 50:96–109.

Griswold, Wendy, and Nathan Wright. 2004. “Cowbirds, Locals, and the Dynamic En-

durance of Regionalism.”American Journal of Sociology 109 (6): 1411–51.

Gross, Sylvia Maria, and Jon Hart. 2013. “How Janelle Monae Discovered Her ‘Inner

Weirdo’in Kansas City.”KCUR 89.3, November 23.

Guan, Frank. 2017. “Rap Dominated Pop in 2017, and It’s Not Going Anywhere Any-

time Soon.”Vulture, December 20.

Hannan, Michael T., Gaël Le Mens, Greta Hsu, Balázs Kovács, Giacomo Negro, László

Pólos, Elizabeth Pontikes, and Amanda J. Sharkey. 2019. Concepts and Categories:

Foundations for Sociological and Cultural Analysis. New York: Columbia University

Press.

Hirsch, Paul M. 1972. “Processing Fads and Fashions: An Organization-Set Analysis of

Cultural Industry Systems.”American Journal of Sociology 77 (4): 639–59.

Hoppe, Alexander D. 2020. “Strategic Balance or Imperfect Imitation? Style and Legit-

imation Challenges in a Semi-peripheral City.”Pp. 227–53 in Aesthetics and Style in

Strategy, edited by Gino Cattani, Simone Ferriani, Frederic Godart, and Stoyan V.

Sgourev. Bingley: Emerald.

Hsu, Greta. 2006. “Jacks of All Trades and Masters of None: Audiences’Reactions to

Spanning Genres in Feature Film Production.”Administrative Science Quarterly 51 (3):

420–50.

Hsu, Greta, Michael T. Hannan, and Özgecan Koçak. 2009. “Multiple Category Mem-

berships in Markets: An Integrative Theory and Two Empirical Tests.”American So-

ciological Review 74 (1): 150–69.

Hudson, Ray. 2006. “Regions and Place: Music, Identity and Place.”Progress in Human

Geography 30 (5): 626–34.

Jay, Anitra. 2014. “Soul Music in Nashville.”Acoustic Soul, December 6. https://anitra

jaymusic.wordpress.com/2014/12/06/soul-music-in-nashville/. Accessed March 16, 2020.

Johnson, Whitney. 2017. “Weird Music: Tension and Reconciliation in Cultural-Economic

Knowledge.”Cultural Sociology 11 (1): 44–59.

Jung, E. Alex. 2018. “The Bold and Bawdy New Queens of Comedy.”Vulture, May 31.

https://www.vulture.com/2018/05/the-rise-of-queer-comedy.html.

Kale, Sunaina Keonaona. 2017. “Localness and Indigeneity in Hawaiian Reggae.”Ph.D.

dissertation. University of California, Santa Barbara.

Kaufman, Jason, and Matthew E. Kaliner. 2011. “The Re-accomplishment of Place in

Twentieth Century Vermont and New Hampshire: History Repeats Itself, until It

Doesn’t.”Theory and Society 40 (2): 119–54.

Keegan, Victor. 2007. “Will MySpace Ever Lose Its Monopoly?”Guardian, February 8.

Keuschnigg, Marc, and Thomas Wimmer. 2017. “Is Category Spanning Truly Disadvan-

tageous? New Evidence from Primary and Secondary Movie Markets.”Social Forces

96 (1): 449–79.

Kirkpatrick, David. 2007. “As Facebook Takes off, MySpace Strikes Back.”Fortune,

September 19.

Kovács, Balázs, and Michael T. Hannan. 2010. “The Consequences of Category Span-

ning Dependon Contrast.”Pp. 175–201 in Categories in Markets: Originsand Evolution,

American Journal of Sociology

282

edited by Greta Hsu, Giacomo Negro, and O

zgecan Kocak. Research in the Sociology of

Organizations 31. Bingley: Emerald.

———. 2015. “Conceptual Spaces and the Consequences of Category Spanning.”Socio-

logical Science 2:252–86.

Kruse, Holly. 2010. “Local Identity and Independent Music Scenes, Online and Off.”

Popular Music and Society 33 (5): 625–39.

Latapy, Matthieu, Clémence Magnien, and Nathalie Del Vecchio. 2008. “Basic Notions

for the Analysis of Large Two-Mode Networks.”Social Networks 30 (1): 31–48.

Lee, Steve S., and Richard A. Peterson. 2004. “Internet-Based Virtual Music Scenes: The

Case of P2 in Alt.Country Music.”Pp. 187–204 in Music Scenes: Local, Translocal and

Virtual, edited by Andy Bennett and Richard A. Peterson. Nashville: Vanderbilt Uni-

versity Press.

Leeper, Thomas J. 2018. “Margins: Marginal Effects for Model Objects.”R package ver-

sion 0.3.23.

Lena, Jennifer C. 2012. Banding Together: How Communities Create Genres in Popular

Music. Princeton, N.J.: Princeton University Press.

Lena, Jennifer C., and Mark C. Pachucki. 2013. “The Sincerest Form of Flattery: Inno-

vation, Repetition, and Status in an Art Movement.”Poetics 41 (3): 236–64.

Leschziner, Vanina. 2007. “Kitchen Stories: Patterns of Recognition in Contemporary

High Cuisine.”Sociological Forum 22 (1): 77–101.

Leung, Ming D., and Amanda J. Sharkey. 2013. “Out of Sight, Out of Mind? Evidence of Per-

ceptual Factors in the Multiple-Category Discount.”Organization Science 25 (1): 171–84.

Lizardo, Omar. 2014. “Omnivorousness as the Bridging of Cultural Holes: A Measure-

ment Strategy.”Theory and Society 43 (3–4): 395–419.

Lopes, Paul. 1992. “Concentration and Diversity in the Popular Music Industry, 1969 to

1990.”American Sociological Review 57 (1): 56–71.

Mark, Noah. 1998. “Birds of a Feather Sing Together.”Social Forces 77 (2): 453–85.

McFarland, Daniel A., Kevin Lewis, and Amir Goldberg. 2016. “Sociology in the Era of

Big Data: The Ascent of Forensic Social Science.”American Sociologist 47 (1): 12–35.

Mears, Ashley. 2011. Pricing Beauty: The Making of a Fashion Model. Berkeley and Los

Angeles: University of California Press.

Mellander, Charlotta, Richard Florida, Peter J. Rentfrow, and Jeff Potter. 2018. “The

Geography of Music Preferences.”Journal of Cultural Economics 42:593–618.

Menger, Pierre-Michel. 1999. “Artistic Labor Markets and Careers.”Annual Review of

Sociology 25 (1): 541–74.

Mize, Trenton D. 2019. “Best Practices for Estimating, Interpreting, and Presenting

Nonlinear Interaction Effects.”Sociological Science 6:81–117.

Mohr, John W. 1994. “Soldiers, Mothers, Tramps and Others: Discourse Roles in the

1907 New York City Charity Directory.”Poetics 22 (4): 327–57.

———. 1998. “Measuring Meaning Structures.”Annual Review of Sociology 24 (1): 345–70.

Molotch, Harvey, William Freudenburg, and Krista E. Paulsen. 2000. “History Repeats

Itself, but How? City Character, Urban Tradition, and the Accomplishment of Place.”

American Sociological Review 65 (6): 791–823.

Navis, Chad, and Mary Ann Glynn. 2011. “Legitimate Distinctiveness and the Entrepre-

neurial Identity: Inﬂuence on Investor Judgments of New Venture Plausibility.”Acad-

emy of Management Review 36 (3): 479–99.

Negro, Giacomo, Michael T. Hannan, and Hayagreeva Rao. 2010. “Categorical Contrast

and Audience Appeal: Niche Width and Critical Success in Winemaking.”Industrial

and Corporate Change 19 (5): 1397–425.

Negro, Giacomo, Özgecan Koçak, and Greta Hsu. 2010. “Research on Categories in the

Sociology of Organizations.”Pp. 3–35 in Categories in Markets: Origins and Evolu-

tion, edited by Greta Hsu, Giacomo Negro, and O

zgecan Kocak. Research in the So-

ciology of Organizations 31. Bingley: Emerald.

Categorical Conventionality in Music

283

Negus, Keith. 2013. Music Genres and Corporate Cultures. London: Routledge.

Nietzsche, Friedrich. 1986. Human, All Too Human: A Book for Free Spirits. Cambridge:

Cambridge University Press.

Pachucki, Mark A., and Ronald L. Breiger. 2010. “Cultural Holes: Beyond Relationality

in Social Networks and Culture.”Annual Review of Sociology 36 (1): 205–24.

Paolella, Lionel, and Rodolphe Durand. 2016. “Category Spanning, Evaluation, and Per-

formance: Revised Theory and Test on the Corporate Law Market.”Academy of Man-

agement Journal 59 (1): 330–51.

Peterson, Richard A., and Andy Bennett. 2004. “Introducing Music Scenes.”Pp. 1–16 in

Music Scenes: Local, Translocal and Virtual, edited by Andy Bennett and Richard A.

Peterson. Nashville: Vanderbilt University Press.

Peterson, Richard A., and David G. Berger. 1975. “Cycles in Symbol Production: The

Case of Popular Music.”American Sociological Review 40:158–73.

Petridis, Alexis. 2002. “The Roots: Phrenology.”Guardian, December 6. https://www

.theguardian.com/music/2002/dec/06/popandrock.artsfeatures1.

Phillips, Damon J. 2011. “Jazz and the Disconnected: City Structural Disconnectedness

and the Emergence of a Jazz Canon, 1897–1933.”American Journal of Sociology

117 (2): 420–83.

Phillips, Damon J., Catherine J. Turco, and Ezra W. Zuckerman. 2013. “Betrayal as Mar-

ket Barrier: Identity-Based Limits to Diversiﬁcation among High-Status Corporate

Law Firms.”American Journal of Sociology 118 (4): 1023–54.

Pontikes, Elizabeth G. 2012. “Two Sides of the Same Coin: How Ambiguous Classiﬁ-

cation Affects Multiple Audiences’Evaluations.”Administrative Science Quarterly

57 (1): 81–118.

Pontikes, Elizabeth G., and Michael Hannan. 2014. “An Ecology of Social Categories.”

Sociological Science 1:311–43.

Pontikes, Elizabeth G., and Ruben Kim. 2017. “Strategic Categorization.”In From Cat-

egories to Categorization: Studies in Sociology, Organizationsand Strategy at the Cross-

roads, edited by Rodolphe Durand, Nina Granqvist, and Anna Tyllström. Research in

the Sociology of Organizations 51. Bingley: Emerald.

Radway, Janice A. 1984. Reading the Romance: Women, Patriarchy, and Popular Liter-

ature. Chapel Hill: University of North Carolina Press.

Rawlings, Craig M. 2001. “‘Making Names’: The Cutting Edge Renewal of African Art

in New York City, 1985–1996.”Poetics 29 (1): 25–54.

Richmond, Ben. 2015. “Big Data Conﬁrms Rap Was the Biggest Innovation in Pop since

1960.”Vice Motherboard, May 6.

Rindova, Violina, Daved Barry, and David J. Ketchen. 2009. “Entrepreneuring as

Emancipation.”Academy of Management Review 34 (3): 477–91.

Romanelli, Elaine. 1991. “The Evolution of New Organizational Forms.”Annual Review

of Sociology 17 (1): 79–103.

Rossman, Gabriel, and Oliver Schilke. 2014. “Close, but No Cigar: The Bimodal Re-

wards to Prize-Seeking.”American Sociological Review 79 (1): 86–108.

Rothﬁeld, Lawrence, Don Coursey, Sarah Lee, Daniel Silver, and Wendy Norris. 2007.

“Chicago Music City: A Report on the Music Industry in Chicago.”Report on the mu-

sic industry in Chicago prepared for the Chicago Music Commission by the Cultural

Policy Center at the University of Chicago.

Roy, William G., and Timothy J. Dowd. 2010. “What Is Sociological about Music?”An-

nual Review of Sociology 36 (1): 183–203.

Ruef, Martin. 2000. “The Emergence of Organizational Forms: A Community Ecology

Approach.”American Journal of Sociology 106 (3): 658–714.

Ruef, Martin, and Kelly Patterson. 2009. “Credit and Classiﬁcation: The Impact of In-

dustry Boundaries in Nineteenth-Century America.”Administrative Science Quarterly

54 (3): 486–520.

American Journal of Sociology

284

Salganik, Matthew J., Peter Sheridan Dodds, and Duncan J. Watts. 2006. “Experimental

Study of Inequality and Unpredictability in an Artiﬁcial Cultural Market.”Science

311 (5762): 854–56.

Sampson, Robert J. 2012. Great American City: Chicago and the Enduring Neighbor-

hood Effect. Chicago: University of Chicago Press.

Schneiberg, Marc, and Gerald Berk. 2010. “From Categorical Imperative to Learning by

Categories: Cost Accounting and New Categorical Practices in American Manufactur-

ing, 1900–1930.”Pp. 255–92 Categories in Markets: Origins and Evolution, edited by

Greta Hsu, Giacomo Negro, and O

zgecan Kocak. Research in the Sociology of Orga-

nizations 31. Bingley: Emerald.

Sgourev, Stoyan V., and Niek Althuizen. 2014. “‘Notable’or ‘Not Able’: When Are Acts

of Inconsistency Rewarded?”American Sociological Review 79 (2): 282–302.

Silva, Robert. 2017. “Country Music Goes to Hawaii.”Liveabout, November 26. https://

www.liveabout.com/country-music-goes-to-hawaii-931980. Accessed March 9, 2020.

Silver, Daniel. 2020. “Figure It Out!”Sociological Methods and Research 49 (4): 868–905.

Silver, Daniel, and Terry Nichols Clark. 2016. Scenescapes: How Qualities of Place

Shape Social Life. Chicago: University of Chicago Press.

Silver, Daniel, Monica Lee, and Clayton Childress. 2016. “Genre Complexes in Popular

Music.”PloSOne 11 (5): e0155471.

Simmel, Georg. 1957. “Fashion.”American Journal of Sociology 62 (6): 541–58.

Simonsohn, Uri. 2018. “Two Lines: A Valid Alternative to the Invalid Testing of U-

Shaped Relationships with Quadratic Regressions.”Advances in Methods and Prac-

tices in Psychological Science 1 (4): 538–55.

Skaggs, Rachel. 2019. “Harmonizing Small-Group Cohesion and Status in Creative Col-

laborations: How Songwriters Facilitate and Manipulate the Cowriting Process.”So-

cial Psychology Quarterly 82 (4): 367–85.

Spillman, Lyn. 2014. “Mixed Methods and the Logic of Qualitative Inference.”Qualita-

tive Sociology 37 (2): 189–205.

Storper, Michael, Thomas Kemeny, Naji Makarem, and Taner Osman. 2015. The Rise

and Fall of Urban Economies: Lessons from San Francisco and Los Angeles. Stanford,

Calif.: Stanford University Press.

Swedberg, Richard. 2016. “Can You Visualize Theory? On the Use of Visual Thinking in The-

ory Pictures, Theorizing Diagrams, and Visual Sketches.”Sociological Theory 34 (3): 250–75.

Uzzi, Brian, Satyam Mukherjee, Michael Stringer, and Ben Jones. 2013. “Atypical Com-

binations and Scientiﬁc Impact.”Science 342 (6157): 468–72.

van Venrooij, Alex, and Vaughn Schmutz. 2018. “Categorical Ambiguity in Cultural

Fields: The Effects of Genre Fuzziness in Popular Music.”Poetics 66:1–18.

Vaughan, Diane. 2014. “Analogy, Cases, and Comparative Social Organization.”Pp. 61–

84 in Theorizing in Social Science: The Context of Discovery, edited by R. Swedberg.

Stanford, Calif.: Stanford University Press.

Vincent, Fran. 2007. MySpace for Musicians: The Comprehensive Guide to Marketing

Your Music Online. Boston: Nelson Education.

White, Harrison C. 2002. Markets from Networks: Socioeconomic Models of Production.

Princeton, N.J.: Princeton University Press.

White, Harrison C., Scott A. Boorman, and Ronald L. Breiger. 1976. “Social Structure

from Multiple Networks: Blockmodels of Roles and Positions.”American Journal of

Sociology 81 (4): 730–80.

White, Harrison C., and Cynthia A. White. 1993. Canvases and Careers: Institutional

Change in the French Painting World. Chicago: University of Chicago Press.

Wohl, Hannah. 2019. “Creative Visions: Presenting Aesthetic Trajectories in Artistic Ca-

reers.”Poetics 76. https://doi.org/10.1016/j.poetic.2019.03.003.

Wood, Simon N. 2017. Generalized Additive Models: An Introduction with R, 2d ed.

London: Chapman & Hall.

Categorical Conventionality in Music

285

Wray, Richard. 2007. “Facebook Challenges MySpace as Place for the Cool Set to Hang

Out.”Guardian, June 21.

Wry, Tyler, and Adam R. Castor. 2017. “Opportunity, Status, and Similarity: Exploring

the Varied Antecedents and Outcomes of Category Spanning Innovation.”Pp. 355–89

in From Categories to Categorization: Studies in Sociology, Organizations and Strat-

egy at the Crossroads, edited by Rodolphe Durand, Nina Granqvist, and Anna

Tyllström. Research in the Sociology of Organizations 51. Bingley: Emerald.

Wry, Tyler, and Michael Lounsbury. 2013. “Contextualizing the Categorical Imperative:

Category Linkages, Technology Focus, and Resource Acquisition in Nanotechnology

Entrepreneurship.”Journal of Business Venturing 28 (1): 117–33.

Wry, Tyler, Michael Lounsbury, and Mary Ann Glynn. 2011. “Legitimating Nascent

Collective Identities: Coordinating Cultural Entrepreneurship.”Organization Science

22 (2): 449–63.

Wynn, Jonathan R. 2015. Music/City: American Festivals and Placemaking in Austin,

Nashville, and Newport. Chicago: University of Chicago Press.

Zhao, Eric Yanfei, Greg Fisher, Michael Lounsbury, and Danny Miller. 2017. “Optimal

Distinctiveness: Broadening the Interface between Institutional Theory and Strategic

Management.”Strategic Management Journal 38 (1): 93–113.

Zuckerman, Ezra W. 1999. “The Categorical Imperative: Securities Analysts and the Il-

legitimacy Discount.”American Journal of Sociology 104 (5): 1398–438.

———. 2017. “The Categorical Imperative Revisited: Implications of Categorization as a

Theoretical Tool.”Pp. 31–68 in From Categories to Categorization: Studies in Sociol-

ogy, Organizations and Strategy at the Crossroads, edited by Rodolphe Durand, Nina

Granqvist, and Anna Tyllström. Research in the Sociology of Organizations 51. Bingley:

Emerald.

Zuckerman, Ezra W., Tai-Young Kim, Kalinda Ukanwa, and James von Rittmann.

2003. “Robust Identities or Nonentities? Typecasting in the Feature-Film Labor Mar-

ket.”American Journal of Sociology 108 (5): 1018–73.

American Journal of Sociology

286

Cultural Tariffing: Appropriation and the Right to Cross Cultural Boundaries

Article

Full-text available

Feb 2024

Why are some acts of cultural boundary-crossing considered permissible whereas others are repudiated as cultural appropriation? We argue that perceptions of cultural appropriation formed in response to the emergence of cultural omnivorousness as a dominant form of high-status consumption, making boundary-crossing a source of cultural capital. Consequently, the right to adopt a practice from a culture that is not one’s own is determined on the basis of the costs and benefits one is presumed to accrue. People express disapproval at boundary-crossing if they believe it devalues or extracts value at the expense of the target culture. We call this process cultural tariffing. We test our theory in a between-subject experimental design, demonstrating that individuals who enjoy a privileged social position, as inferred from their social identity or socioeconomic status, have less normative latitude to cross cultural boundaries. This is explained by perceptions that these actors are either devaluing or exploiting the target culture. While symbolic boundaries and cultural distinction theories are inconsistent with our results, we find that Americans who are disenchanted about group-based social mobility are the most likely to be outraged by cultural boundary-crossing. Cultural tariffing, we therefore posit, is a form of symbolic redistribution.

Artistic referencing and emergent standards of peer recognition in Hollywood, 1930–2000

Article

Apr 2024
POETICS

Katharina Burgdorf

The Polarization of Popular Culture: Tracing the Size, Shape, and Depth of the “Oil Spill”

Article

Dec 2023

Recent research suggests that political polarization has spilled over into otherwise mundane areas of social life. And yet, the size, shape, and depth of that spillage into popular culture are generally unknown. Relying on a sample of 135 widely known movies, TV shows, musicians, sports, and leisure activities, we investigate these issues. We find the “oil spill” of polarization into popular culture is large but loosely organized into multiple fairly shallow pools. Cultural polarization is also asymmetric. Liberals like a wide variety of popular culture, do not dislike conservative popular culture, and their tastes are more rooted in their sociodemographics. Conservatives, on the other hand, like a much narrower range of popular culture, dislike the culture created and liked by Black and urban liberals, and their tastes seem to be more directly rooted in their political ideology. Potential implications of an asymmetric culture war, and ideas for future research, are discussed.

In the mood for odd? The role of affective factors in the evaluation of categorical atypicality

Article

Dec 2023
POETICS

The ties that conform: Legitimacy and innovation of live music venues and local music scenes

Article

Oct 2023
POETICS

Yun Tai

The performative power of cinema: theorizing successful performances of realism in cinema

Article

Full-text available

Sep 2023

Jessie Dong

This article proposes a performative model for the interpretation of cinematic realism. It theorizes “fusion” as a performative achievement in which audiences interpret a cinematic performance’s underlying “realism”, or collective representations. It also conceptualizes the “cinematic gap” as an experiential space of both deep cinematic immersion—and detachment from one’s own lived experience. Using the case study Parasite, the 2019 South Korean satire on class disparities, this paper answers the question: How can cinematic performances enable audiences to experience fusion with realism, despite the cinematic gap? How can Parasite effect deeper understandings of class relations? It answers these questions by way of three theoretical elaborations on the cultural mechanisms of cinematic performance: First, there are “two levels of fusion”. The first level resides at the level of aesthetic experience. The second level goes beyond aesthetic surface and reaches collective representations, or “realism”. Second, the elements of performance can be organized according to a typology that effects varying forms and degrees of realism. Third, the organization of performative elements create various cinematic gap “sizes” —the larger the gap, the further the cinematic performance distances viewers from their lived experience. In the case of Parasite, it is shown that a cinematic gap—a sense of distance from lived experience—can actually bridge viewers to deeper understandings of social reality in unexpected ways.

Organizations and the structure of culture

Article

Full-text available

Dec 2022

Recent work emphasizes how organizations and organizing are suffused by culture. Ignored, however, is the question of how organizations and organizing shape cultural understandings. In this paper, we draw attention to how organizations impact the creation, maintenance, and matching of cultural associations: The “structure of culture.” Cultural associations are distributed understandings of “what goes with what,” diffused via their co‐adoption by individual and corporate actors. We argue that the process of associative diffusion is heavily influenced by organizational forms, routines, and operations. While organizations shape culture production, they likewise shape resulting associations between cultural items, actively modifying the structure of culture. We conclude that the role of organizations in shaping the structure and dynamics of culture should be a central topic of study moving forward.

Scenescapes: How Qualities of Place Shape Social Life

Book

Full-text available

Jan 2016

Harmonizing Small-Group Cohesion and Status in Creative Collaborations: How Songwriters Facilitate and Manipulate the Cowriting Process

Article

Full-text available

Sep 2019

Rachel Skaggs

Because of economically induced changes in record label contracts, recording artists are now required to cowrite their songs. Despite this obligation, many recording artists are unskilled as writers. For their professional songwriter teammates, using face-work to save the recording artist from the embarrassment of being an unskilled songwriter can be at odds with status maintenance in the larger social community if the collaboration results in a bad song. Interviews with 39 country songwriters revealed two key strategies for orienting the songwriting process toward writing a song that will allow for status maintenance among peers while still attending to group cohesion. Of the two strategies for achieving this goal, most common was a process of commercially oriented ''bespoke facilitation'' to match songs' lyrical content to artists' personal branding goals. In a second strategy, referred to as a ''manipulation dance,'' songwriters prewrite songs or portions of songs to guide the creative process toward content that will contribute to their status among industry peers. Both strategies rely on songwriters using face-work to be perceived as a ''good hang'' to keep the recording artist happy while still ensuring that the song is well crafted and will help to build and maintain their status. This study contributes to understanding how an individual working in a collaborative group might orient efforts toward maintaining status in a wider community while attending to group cohesion when the group has mismatched goals, skills, or tastes.

Music Genres and Corporate Cultures

Book