Colin Morris (a self-described "unemployed programmer and deep learning enthusiast" interested in "machine learning and data visualization") recently published an intriguing paper titled "Are Pop Lyrics Getting More Repetitive?" in The Pudding, a weekly journal of visual essays. This paper takes over a reflection, started in 1977 by Donald Knuth, a computer scientist, in a paper titled The Complexity of Songs. At that time, Knuth questioned in a humorous way the tendency of popular songs to drift away from content-rich ballads to highly repetitive texts, with little or no meaningful content.
Morris's contribution literally tests Knuth's 1977 hypothesis with data. He analyzed the repetitiveness of a dataset of 15'000 songs that charted on the Billboard Hot 100 between 1958 and 2017. To proceed, he used a compression algorithm (the Lempel-Ziv algorithm or LZ) used to compress files such as gifs, pngs, and other computer archive formats. As explained by Collins, the LZ works by exploiting repeated sequences: "How efficiently LZ can compress a text is directly related to the number and length of the repeated sections in that text." The results of Collins's experiment are very clearly described in his paper through several graphics and animations. They tend to demonstrate the hypothesis according to which, since the 1960s, popular music became more and more repetitive (or, in other words, easier to compress at a higher rate):
"In 1960, the average song is 45.7% compressible) ... By 1980, the year's most repetitive song is Funkytown (85% compressible) ... An average song from  compresses 22% more efficiently than one from 1960."
Discussing the results of his study, Collins explores differences among genres and artists and establishes comparison charts, organized by decades. By browsing his paper, you'll learn that Daft Punk's (1997) "Around the World" is the most repetitive song produced during that period, Rihanna the most repetitive artist in Collins's dataset, or that rappers like J. Cole and Eminem tend to be consistently non-repetitive.
Repetition, rhythm, aesthetic value and the way they relate to society
Even if it does not assert an aesthetic claim, Collins's study brings one more piece to a long tradition of reflections questioning the relationships between aesthetic rhythms (e.g., poetry, music, dance) and the rhythmic features that characterize a sociocultural environment at a specific period. The questioning of the rhythmic features inherent to cultural production, such as poetry or music, has a long history. For Plato and Aristotle, rhythms used to refer to the principle organizing the succession of elementary and complex units composing poetry, music and dance. Their approach was congruent with a conception of aesthetic judgment privileging some kind of measure (metron). As discussed by Couturier-Heinrich (2004), during the 18th century, after the contributions of poets such as Moritz, Goethe, Schiller, Schlegel and Hölderlin, the concept of rhythm appeared again in reflections on aesthetic, privileging the inner qualities of a text, rather than its measurable attributes. During the second half of the 19th century, Wagner and especially Nietzsche reinitiated the discussion. The evolution of aesthetic rhythms was then interpreted as a sign of societal mutations, associated – among others – with the cultural and economic shifts characterizing modernity and the industrial revolution (Hanse, 2007).
Repetition and the quality of lived experience
Beside the fact that it proposes an objective measurement to describe how lyrics may have evolved during the second half of the 20th century, Collins's study brings in my opinion an additional element to the current research around rhythmanalysis. To locate it, I must first reframe it in the light of a reflection around the relationship between repetition and the quality of lived experience. Since Marx's analysis, the "tyranny of time" in capitalist society remains a recurring theme in sociological studies focusing on the role played by the rigidity, the coercion and the regularity imposed through the temporal framework of industrialization (e.g., assembly line, taylorization). As discussed by Lefebvre (1961/2002, p. 340), the relationship between alienation and repetition is both a matter of quality and quantity. Thus, different types of repetition have to be distinguished (i.e., taking into consideration the level of difference and creativity they involve) to analyze their value and meaning.
Working on an assembly line, or repeating every day the same routines within a classroom, may be experienced as alienating because repetition is lived as a source of monotony, tiredness, consumption or exhaustion (Jacklin, 2004). It dispossesses therefore the person from one’s own embodied experience. It does not let room for self-creation, plenitude or harmony with oneself and with the world. From this angle, the redundancy of the pragmatic demands of everyday life may constitute a source of detachment that separates daily actions (e.g., at work, in school or in the family) from what generates them (e.g., impulse or desire), resulting in an emptying out of meaning and the banality of the quotidian (Lefebvre, 1961/2002, 1992/2004). Alienation may come therefore from the separation between creative impulses and the repetitive rhythms of life (Lefebvre, 1992/2004). This is one of the reasons why Lefebvre’s rhythmanalytical project was grounded in the study of the rhythmic dimensions of the every day as potential sources of alienation. (Alhadeff-Jones, 2017, p.164)
Experiencing repetition and the mathematical measurement of redundancy
The contribution of Collins's study becomes particularly relevant, once it is linked to a broader reflection around repetition and the quality of lived experience. Collins's contribution translates an intuition. The intuition that the complexity of cultural production may be decreasing through time, according to some standards (e.g., the level of redundancy of information) or varies depending on an artist's repertoire. In a way, some would argue that there was no need to establish such a sophisticated demonstration to make that claim. The merit of the approach is that it provides one with an objective measurement to describe such an evolution. As formulated by Collins: "I know a repetitive song when I hear one, but translating that intuition into a number isn't easy." In social sciences, rhythmanalysis usually refers to a praxis first conceived from a qualitative perspective: the study of the qualities displayed by the experience of rhythmic phenomena. A contrario, in biology or in medical studies, rhythms analysis is based on quantitative data (e.g., the measurement of cardiac activity). What seems to me particularly interesting with Collins's approach is the fact that it demonstrates the value of using a specific algorithm to measure a dimension constitutive of the evolution of the complexity of cultural productions. By providing an analysis that goes beyond human capacity of perception, it provides us with a richer description of the world we are living in.
Computational complexity and rhythmanalytical research
From a methodological point of view, the idea of using compression algorithms to measure the level of redundancy of information opens up a stimulating avenue for rhythmanalytical research. If redundancy may be conceived as a marker of the absence of creative impulse, understood as a sign of loss of the self (Alhadeff-Jones, 2017), then its mathematical measurement provides us with a relevant tool to compare situations and evaluate their evolution through time. No need for a sophisticated algorithm to know when an activity is experienced as too repetitive, especially when the inconvenience is experienced through one's own body. Things become more tricky when we start considering activities involving discursive practices. Again, it seems that there is no need for an elaborated research setting to determine that working for instance at a call center may constitute a repetitive activity, shaped by unimaginative scripts. But once you want to compare activities, such as those involved in teaching, caring, or helping others, things become much more complicated.
Following Collins' example, we could imagine following a cohort of professionals (e.g., teachers, trainers, doctors, nurses) who would accept to have their voice recorded during a whole day, several days a year, several years in a row. Using an algorithm such as the LZ could provide us with a measurement of the level redundancy of their discourses, how it compares between professionals, between fields of practice, and for the same person, how it evolves through time. I have never been a proponent of quantitative approaches in human sciences, but it seems to me that such a tool would represent an interesting instrument to explore, through different contexts and different periods, the level of complexity of the discursive rhythms involved in one's activity.
Said in another way: In a time when standardization and quality management require people to follow predefined procedures, and adopt standard formulas, being able to measure the level of creativity inherent to one's discourses appears as an interesting way to describe how people learn (or unlearn) to resist through time to the increasing homogenization of human practices.
What about you?
When do you experience repetition in a way that seems debilitating?
What kind of strategy do you implement in order to enrich your everyday practice?
How do you know when you need to revise what you used to do in order to make it more creative?
Feel free to use the comments section below to share your feedback and questions. Thank you.