Introduction

People like to dunk on "music these days", claiming it is too simple and watered down, but how much has popular music actually changed over time? I used data from Ultimate Guitar to quantify song structure and chord progression, then used this data to compare differences.

TL;DR

While structural complexity (chorus/bridge/verse) has remained about the same over the past 50 years, there are a few ways in which modern music has become more simplified. Modern songs often use fewer unique chords. Modern music has become much less likely to deviate from its key signature.

Song measures

• Structure: the different regions a song has, and how the song moves between them. One common song structure is Verse/Chorus/Verse/Chorus/Bridge/Chorus.
• Chord Progression: the order the chords are played in. A common chord progression in pop music is I V vi IV - in C major, this would be C G Am F.

Data Acquisition Process

Chord and structure information was scraped from Ultimate Guitar using the Beautiful Soup library. Additional information about song genre was scraped from Last.Fm.

The detailed process for obtaining, parsing, and manipulating the data can be found in Jupyer notebooks on GitHub.

Limitations

• Not all data on Ultimate Guitar is completely accurate. More popular tabs tend to be more accurate, as many people have seen them and been able to add changes, but it's likely that some of the data in this set is not correct.
• Song data was procured based on the song's popularity on Ultimate Guitar rather than its actual popularity during the time it came out. It may not be the best representation of a decade's popular music, however one could argue that a decade's most popular songs tend to be the ones people still care about years later.
• Finding the tonic chord is not always so easy. This analysis uses roman numeral notation to allow for comparison across songs with different keys, however not all songs list their key on Ultimate Guitar. For these songs, the tonic chord is inferred to be whichever tonic would be the "best fit", but it may not always be correct.