The origins of hits: Cumulative advantage vs. multiplicative returns in cultural markets
Introduction
Two stylized facts characterize the careers of many cultural objects: (1) popularity is highly unequal—often a few hits are orders of magnitude more popular than the typical object (Salganik & Watts, 2006), and (2) popularity is very difficult to predict, even for experts (Bielby & Bielby, 1994). Beginning with Adler (1985), social scientists have explained cultural object popularity through cumulative advantage (CA) processes wherein the popularity of cultural objects increases the likelihood that additional adopters will flock to them. CA is a theoretically appealing model since it can parsimoniously generate both high inequality and unpredictability, and has extensive empirical support (Godart & Mears, 2009; van de Rijt et al., 2013; Rossman et al., 2010; Salganik & Watts, 2006). CA has powerful implications for the study of culture and cultural objects, in part because it implies that the public reception of cultural objects is only loosely coupled with the personal tastes of individuals (Lizardo, 2016), and because it shows how the endogenous spread of cultural objects can generate high inequality in reception (Kaufman, 2004). Recent work, however, suggests that CA cannot entirely account for the dramatic inequality in the careers of cultural objects, particularly for the biggest hits (e.g., Lynn et al. 2016, van de Rijt 2019), suggesting the need for additional models.
Other work on cultural object adoption stresses the role of durable aesthetic dispositions or “tastes” (Bourdieu, 1984; Lieberson, 2000). Although tastes form through social processes, once formed they allow individuals to ascertain the intended meaning of cultural objects and form aesthetic judgements independently rather than from contemporaneous reactions of others to the object (Cerulo, 2018; Lieberson & Mikelson, 1995; Wohl, 2015). Theories of taste give us accounts of how the properties of objects influence how objects are perceived and whether they are adopted. Yet we still lack a sociological account of how tastes and the properties of objects produce skewed popularity distributions. Indeed many studies are motivated by the difficulty of predicting the emergence of hits based on their properties alone (e.g., Salganik et al. 2006).
If CA were the only model capable of generating the unpredictability and inequality of cultural markets, then the role of taste-based mechanisms in explaining the adoption and popularity of cultural objects would be highly circumscribed—when we observe a hit cultural object, we must assume that it resulted from CA. Here, however, I introduce a formal model of cultural object adoption based on taste that also reproduces the high inequality and unpredictability characteristic of CA processes. This model captures the intuition that cultural objects are judged as ensembles of interdependent properties, such as when a book cover is carefully chosen in the hopes of matching the “core essence” of the writing in a book (Childress, 2017:133), or, in other words, that their appeal is more than the sum of their parts (Taylor et al., 2019:8). I call this model “multiplicative returns” (MR). MR assumes that cultural objects will be judged and adopted according to a multiplicative function, or the interaction, of several underlying properties. In this model, hits are “perfect storms” of complementary qualities, which interact to form an object's overall appeal.
I illustrate the MR model empirically on the popularity of baby girls’ names in the United States. Girls’ names present a puzzle for CA models in that parents often choose novel or rare names to express individuality (e.g., Lieberson 2000), yet many names still become extremely popular. The MR model helps to make sense of this. Baby names must simultaneously satisfy many conditions for parents, such as sounding pleasing or having particular gender, race, and class connotations. The MR model shows how skewed distributions can arise without CA when objects such as names are judged as ensembles of properties. That is, names can become hits based on the reception of their properties (sound, gender connotation, associated imagery, etc.) rather than because popularity begets further popularity. Both CA and MR models are generative models of inequality: mathematical and simulation models that seek to generate the patterns they explain (Epstein & Axtell, 1996). To compare the models empirically, I show that the popularity distributions generated by the MR model are closer than those generated by canonical CA models to the empirical popularity distribution of baby girls’ names. I conclude that, in general, extreme inequality in cultural object popularity is likely due to both CA and MR processes and theorize three ways that CA and MR processes may intersect.
Section snippets
Cumulative advantage and cultural objects
A CA process is one in which “current levels of accumulation have a direct [positive] causal relationship with future levels1” (DiPrete & Eirich, 2006:272). Because the “rich get richer,” CA leads to high inequality among the units subject to the process (DiPrete & Eirich, 2006). In Merton's classic study of the Mathew effect, for instance, scientists who are already well-known and respected in their field tend to
Multiplicative returns and cultural objects
In contrast to CA models, the MR model focuses on how the properties of cultural objects influence their popularity. Although sociologists of culture have long demonstrated that the properties—symbolic or physical—of cultural objects influence their reception, they have not shown if or how this explains the emergence of skewed distributions of popularity. Instead, it is clear that there is no simple or obvious relationship between the popularity of objects and their properties, and the appeal
The contexts of cumulative advantage and multiplicative returns
Although we are unlikely to find either CA or MR in pure form, features of social environments and cultural objects may favor one process over the other. Key features of the social context include aspects of network topology and contexts affecting individuals’ “taste for popularity.” Key features of cultural objects are their cost structure, how their materiality affects their durability and diffusion, as well as their social durability and visibility.
The puzzle of baby girl names
Baby names have been treated as a “model case” (Krause, 2021) for the analysis of cultural objects, in part because they are free to adopt and are not promoted by business or other organized interests (Lieberson, 2000). Yet, baby girls’ names present a puzzle from the standpoint of CA. On the one hand, the popularity of baby girls’ names is massively unequal, as we would expect from a CA process. Data from the US Social Security Administration containing the popularity distribution of all US
Testing MR against CA
In this section, I formalize the MR model and argue that, although both MR and CA generate high inequality among cultural objects, which may be indistinguishable in a substantive sense, the popularity distributions generated by MR and CA are statistically distinguishable. Unlike regression modeling, I am not seeking to parameterize the MR model, or show which specific properties of cultural objects lead to great popularity. Rather I follow a “generative” logic (Epstein & Axtell, 1996), testing
Discussion and conclusion
To explain hit cultural objects social scientists have generally invoked a cumulative advantage (CA) process wherein popularity begets popularity in a virtuous cycle. In this paper, I suggested a multiplicative returns (MR) model, wherein individuals judge cultural objects through how the (multiplicative) interaction of their many properties form an overall “gestalt” of an object. Despite having similar consequences, these processes imply very different theories of cultural consumption. In a
Acknowledgments
I thank Alison Appling, Shawn Bauldry, Philip Cohen, Daniel DellaPosta, Brandon Gorman, Omar Lizardo, Andrew Perrin, Kyle Puetz, Ashton Verdery, participants at the Culture and Politics Workshop at UNC Chapel Hill, the Duke Networks Analysis Center Seminar, the Culture and Politics seminar at Penn State, and the editors and anonymous reviewers at Poetics, for their help in various ways on this project.
Charles Seguin is an associate professor of sociology and social data analytics at the Pennsylvania State University. He studies culture, politics, and history using a variety of methods.
References (98)
- et al.
City age and city size
European Economic Review
(2014) - et al.
Binding significance to form: Cultural objects, neural binding, and cultural change
Poetics
(2019) - et al.
Power-law distribution of the world wide web
Science (New York, N.Y.)
(2000) Stardom and talent
American Economic Review
(1985)- et al.
Powerlaw: A python package for analysis of heavy-tailed distributions
PLOS One
(2014) - et al.
The political consequences of social movements
Annual Review of Sociology
(2010) - et al.
When market information constitutes fields: Sensemaking of markets in the commercial music industry
Organization Science
(2000) - et al.
Path-dependent processes and the emergence of macro-structure
European Journal of Operational Research
(1987) Competing technologies, increasing returns, and lock-in by historical events
The Economic Journal
(1989)- et al.
What makes popular culture popular? Product features and optimal differentiation in music
American Sociological Review
(2017)
Emergence of scaling in random networks
Science (New York, N.Y.)
Cross-correlations of American baby names
Proceedings of the National Academy of Sciences
Agendas and instability in American politics
How adoption speed affects the abandonment of cultural tastes
Proceedings of the National Academy of Sciences
‘All hits are flukes’: Institutionalized decision making and the rhetoric of network prime-time program development
American Journal of Sociology
Strikes as forest fires: Chicago and Paris in the late nineteenth century
American Journal of Sociology
Size matters: Quantifying protest by counting participants
Sociological Methods & Research
Portage and path dependence
The Quarterly Journal of Economics
Distinction : A social critique of the judgement of taste
‘Anything but heavy metal’: Symbolic exclusion and musical dislikes
American Sociological Review
Ability and income
British Journal of Educational Psychology
Complex contagions and the weakness of long ties
American Journal of Sociology
Scents and sensibility: Olfaction, sense-making, and meaning attribution
American Sociological Review
The mundanity of excellence: An ethnographic report on stratification and Olympic swimmers
Sociological Theory
Cultural reception and production the social construction of meaning in book clubs
American Sociological Review
Under the cover: The creation, production, and reception of a novel
Social influence: Compliance and conformity
Annual Review of Psychology
Power-law distributions in empirical data
SIAM Review
The equilibrium size distribution of freely-forming groups
Sociometry
Clio and the economics of qwerty
The American Economic Review
Why do liberals drink lattes?
American Journal of Sociology
Cumulative advantage as a mechanism for inequality: A review of theoretical and empirical developments
Annual Review of Sociology
Segregation in social networks based on acquaintanceship and trust
American Journal of Sociology
Self-organized critical forest-fire model
Physical Review Letters
A multiplicative model of optimal CEO incentives in market equilibrium
The Review of Financial Studies
Blockbusters: Hit-making, risk-taking, and the big business of entertainment
Growing artificial societies: Social science from the bottom up
Short-term projects and emergent careers: Evidence from hollywood
American Journal of Sociology
The sum of log-normal probability distributions in scatter transmission systems
IRE Transactions on Communications Systems
The winner-take-all society : Why the few at the top get so much more than the rest of us
Zipf's Law and the growth of cities
American Economic Review
Zipf's Law for cities: An explanation
The Quarterly Journal of Economics
Les inégalités économiques
Expert opinion and compensation: Evidence from a musical competition
The American Economic Review
How do cultural producers make creative decisions? Lessons from the catwalk
Social Forces
Self-correcting information cascades
The Review of Economic Studies
From Patrick to John F. ethnic names and occupational success in the last era of mass migration
American Sociological Review
A methodological framework for the sociology of culture
Sociological Methodology
Drift as a mechanism for cultural change: an example from baby names
Proceedings of the Royal Society B: Biological Sciences
Cited by (0)
Charles Seguin is an associate professor of sociology and social data analytics at the Pennsylvania State University. He studies culture, politics, and history using a variety of methods.