This post discusses a book, a web column, a podcast, and an academic research project that all have the same goal: to get their arms around the past 50 to 100 years of Anglo-American popular music. There have been plenty of surveys of rock and pop history, but data analysis and online platforms are making it possible to expand their scope.

The book

is Uncharted Territory: What Numbers Tell Us about the Biggest Hit Songs and Ourselves by Chris Dalla Riva, soon to be published by Bloomsbury Academic. Chris graciously shared an advance copy with me.

The project began when Chris found a Spotify playlist with every Billboard number one hit, and listened through it at random. He found some surprises there:

[W]here was Bruce Springsteen? How did Ringo Starr top the charts twice before John Lennon did it once? And why were many of Elton John’s number ones not that good? I was confused.

So he decided to listen through all of the number ones in order, to collect data on all of these songs, and to use that data to try to make sense out of it all.

Before we get into Chris’ conclusions, there are a few things you need to know about the Billboard Hot 100 as a data set. The criteria for a hit change over time as technology and the media landscape evolve. In 1958, when the chart began, it compiled sales of vinyl records, radio airplay and jukebox plays. Now it compiles physical and digital sales, radio play and streaming. The chart is a pretty good proxy for popularity, but not necessarily of cultural significance, much less quality. And it has some systematic distortions.

Chris tells the story of Soundscan, an automated sales data compilation system made possible by the advent of bar codes in 1990. Before Soundscan, the Hot 100 was easily manipulated by labels and record store owners, so it over-counted songs in genres these people preferred, for example, classic rock. The chart also under-counted songs in genres the labels and store owners didn’t, for example, rap. If you just look at the charts, it seems like rap had an abrupt rise in popularity in 1990. In fact, what happened is that the charts simply reflected rap sales accurately for the first time. Chris also points out that pre-Soundscan, Black artists had been systematically undercounted outside of rap as well. Among the first 50 non-rap number one hits of the post-SoundScan era, more than three quarters were by Black artists.

The data corrected some of my misconceptions too. In the pre-rock era, it was rare for singers to write their own songs, but since the Beatles, it has become common. I had assumed that Lennon and McCartney drove this trend, but Chris shows that it started well before them, with 1950s stars like Chuck Berry and Little Richard.

Another interesting songwriting trend: an average number one hit now has more co-writers than it used to. Before 1990, most number one songs were written by duos: Leiber and Stoller, Lennon and McCartney, Elton John and Bernie Taupin. Since 1990, however, the average number of songwriters on a number one has doubled. You might assume that this is due to the rise of sampling (writers of the sampled song are often listed as co-writers of the sampling song) and lawsuits (these are often settled by adding the person being allegedly plagiarized as a co-writer). However, Chris shows that the main driver is neither of these factors, but rather, the more collaborative nature of DAW-based track-and-hook writing.

The book is not only about popularity. Chris has a method for quantifying how good these songs, though it’s heavily subjective: he and two other people each give the song a rating from 1 to 10. The book doesn’t list these ratings, but Chris does use them to define a category of song called Argument Starters, the ones that he and the two other critics most strongly disagreed on. The Argument Starters include “Incense and Peppermints” by Strawberry Alarm Clock, “Ben” by Michael Jackson, “Hotel California” by the Eagles, and “Africa” by Toto.

Opinionated though he is, Chris is ultimately a poptimist, and his tone is cheerful throughout. This is not like listening to Rick Beato complain about how current pop is trash. The book is much more concerned with presenting an objective picture of pop history than it is with making critical judgments.

The column

is The Number Ones by Tom Breihan, published by Stereogum. Breihan is reviewing every number one song from the Billboard Hot 100 in chronological order (he’s up to 2020 as of this writing.) This would seem to be the same project that Chris Dalla Riva is doing, but there are some big differences. While Chris is a data scientist first and foremost, Tom Breihan is a critic, and a strongly opinionated one. He gives plenty of factual background on each song, but he also lets you know exactly how he feels about them too.

My favorite best installments are about the songs that Breihan hates. He says that “Rude” by Magic! “sounds like hot boiled ass.” He says of Katy Perry’s “Roar” that “you can practically visualize the board meetings that led to the track’s release.” He describes the members of Creed as looking like they manage your local mall’s knife store. And here’s how he begins his column about “Everything You Want” by Vertical Horizon:

Check this out: You know how the horizon is usually horizontal, right? Like, it’s a line that goes across? That’s actually where they got the word “horizontal” from. Because it’s like the horizon. But what if the horizon was vertical? Like, what if it went up and down? You could be watching the sunset, and the sun would be coming in from the right. Or maybe even from the left! That shit would be crazy. Would you even be able to call it the horizon anymore? Or would it be, um, the vertica?

I mean, look, I guess you could by lying on your side and looking at the sunset. And then maybe the sun would look like it was going into the ground from the side. But you’d still know that the horizon was horizontal. That’s not what I’m talking about. You’re getting my words twisted. I’m saying: What if the horizon was vertical for everybody? What about that? It’s just something to think about.

Anyway, “Everything You Want,” the only big hit from the band Vertical Horizon, really fucking sucks.

So, very amusing. But you can learn a lot from this column. Breihan goes deep into the musicians’ and songwriters’ life stories and larger cultural context. He is remarkably good at producing substantive writing about so many insubstantial songs.

The podcast

is A History of Rock Music in 500 Songs by Andrew Hickey. I am not much of a podcast listener (which is why it took me so long to start mine), but I make an exception for this one. Andrew’s project is similarly ambitious to Chris Dalla Riva’s and Tom Breihan’s, but he’s using a different data set, a hand-selected list of songs that he considers to be the most historically and stylistically significant, regardless of their level of commercial success. The selections reflect Hickey’s tastes: the Beach Boys appear in 25 episodes, while the Rolling Stones only appear in 15 so far. (As of this writing, the podcast is up to about 1969, though it’s not strictly chronological.) The depth and breadth of Hickey’s research is staggering, and he is scrupulously thorough throughout.

Hickey starts his story much earlier than most rock historians, with Charlie Christian’s guitar solo on Benny Goodman’s 1938 recording of “Flying Home”. As the episodes have progressed, they have been getting longer as Hickey widens his scope. The episode on “White Light, White Heat” by the Velvet Underground is almost three and a half hours long. larger in scope. The following episode, on “Dark Star” by the Grateful Dead, is more than four and a half hours. After that, Hickey started spreading the biggest landmark songs across multiple episodes, so “Sympathy for the Devil” by the Rolling Stones takes up four episodes that are cumulatively six hours long.

Hickey’s tone is serious and scholarly, but he has a subtle sense of humor too. The episode about “I Put A Spell On You” by Screamin’ Jay Hawkins quotes Hawkins saying he loved the army because you were allowed to kill people without being punished for it. Hickey drily comments: “History does not record exactly how many people his saxophone playing killed.”

For me personally, the most enlightening episodes are the ones that cover the pre-Elvis era, all that jump blues and hillbilly boogie. I don’t know these styles very well and I have been filling a lot of knowledge gaps.

The academic research project

is A Corpus Study of Rock Music by Trevor de Clercq, David Temperley, Ethan Lustig, and Ivan Tan. I will be referring to them as Team Corpus for brevity’s sake. Like Andrew Hickey, these researchers are examining a list of five hundred rock songs, though they did not choose them. Instead, they are using a list of the 500 greatest rock songs compiled by Rolling Stone magazine in 2004, drawing input from 172 musicians, critics and others. They are transcribing these songs and analyzing their melodies and chords. The best entry point into the research is this paper from 2011 by de Clercq and Temperley.

This project has very different goals than the other ones listed here, because it is solely focused on musicological analysis rather than historical or cultural analysis. The researchers are only looking at pitches and rhythms. They are not considering lyrics aside from timing and stress patterns, and they are not considering timbre at all. That’s a big omission for rock! In fairness, though, there is no formal framework for analyzing timbre, so there’s not much you can do quantitatively.

So, what does the data tell us? The conclusions won’t be interesting to a normal person, but if you teach or study music theory, they are richly significant. To understand why, here’s a brutally oversimplified summary of American music theory pedagogy’s attitude toward pop music.

Phase one: Popular music is not worthy of serious study, and we don’t need to bother teaching it.

Phase two: Fine, we can study and teach popular music, but we can use the existing toolkit of Western tonal theory.

Team Corpus is trying to find out just how well Western tonal theory does explain rock. My approach to this kind of discussion is to cherry-pick examples: Western tonal theory explains Billy Joel pretty well, but it’s hit or miss with the Beatles and no use at all for Muddy Waters. Team Corpus wants to be more systematic and objective.

In Western European classical music, the most common chord preceding the tonic is V, a G chord in the key of C. The V-I cadence is the central harmonic foundation of Western European classical. If you take a traditional music theory class, you study V-I for a long time before you talk about any other chords, and entire systems of analysis are based on the assumption that everything but V-I is decorative.

Team Corpus finds that the Rolling Stone corpus works differently. Those songs use V-I too, but they more commonly precede the tonic with IV, an F chord in the key of C. It’s also common for rock songs to precede the tonic with bVII, a Bb chord in the key of C. This chord movement is vanishingly rare in Western European classical. If we are going to teach pop music in theory class, we evidently can’t just treat pop as an extension of classical; the rules and conventions are different, and our pedagogy should reflect that.

I recently went to a talk by David Temperley in which he discussed some related research he’s doing. He transcribed the number one song of every year from 1900 to 2000 and did some statistical analysis. He found that the songs’ harmonic complexity of these songs declined steadily over the course of the century, while their rhythmic complexity increased steadily. My intuition and general knowledge tell me that a typical pop song from 1990 is going to be less harmonically interesting and more harmonically interesting than a typical pop song from 1940, but it’s nice to see some data backing me up.

The NYU music theory and history program is currently developing a new class on the history of 20th century popular music. I am very much hoping to teach this class. I have woven some pop history into most of the classes I have taught up to this point. I haven’t had any consistent or rigorous definition for what counts as a historically significant song. I talk about the Beatles a lot because they are undeniably important and well-documented, but my main motivation for including them is that I love those songs and know them well. I don’t talk much about the Beach Boys, even though they arguably just as important as the Beatles, because I don’t like their music. Is that a responsible choice? I could teach the Beach Boys thoroughly if I had to, but in a one-semester class, I would rather direct my and the students’ attention elsewhere.

I admire the authors and researchers I’ve listed here for their ambition to cover the whole spectrum, to give as complete a picture of the pop landscape as is feasible. Chris Dalla Riva and Team Corpus are aiming for objectivity using quantitative data. Tom Breihan is taking a predefined corpus of songs and giving his opinions about them. Andrew Hickey has chosen his own corpus of songs and is telling their stories as objectively and fairly as he can. I don’t have anything like the discipline to do anything so systematic or complete, so I’m glad to have all of their work to refer to, and to enjoy.