A Unified Theory of Wrapped Space-Time
We get yet closer to cracking the case of who was my Top Artist in 2024.
Middling Content helps media users use media more deliberately, providing thoughtful analysis of media usage patterns. Take back control of your eyes and ears, and subscribe today!
Last week, I challenged whether a “Top Artist” needed to have balance across the artist’s catalog. Essentially, I claimed that rather than focus exclusively on the number of track listens, Spotify should also consider the unique number of tracks. This week, I’ll focus more on the other part of “Top Artists of 2024” -- that is, the “of 2024” part. And there are three temporal critiques I have of the Spotify methodology.
They can be summarized as follows, roughly:
The Year as a Research Frame
Inadequate Intra-Year Consideration
Inadequate Inter-Year Consideration
Let’s dig in!
A Year is ~365 Days
The first critique is the most straightforward.
We know how long a year is. We’ve seen Rent. The year doesn’t end in October, which is when Spotify has their cut-off; and as such, top artists around solstice times get memory-holed. What’s the impact of this?
When I run the numbers with full year data, we see that less than half the “top five” artists hold their positions.
For Raquel Sofía, that leap upward 3900 positions is indeed because of a holiday song, but the other two aren’t. And “Alles War Aus Geld” by Bakkushan is definitely one of my favorite finds of 2024, and its wistful driving optimism will remind me of a particular time and place going forward. So in my unwrapped-rewrapped Wrapped, I would definitely prefer to keep it in there.
I was surprised how much more data was provided by widening the aperture to the full year. Whereas there were 3385 artists in dataset cutting of at the end of October, the full year dataset includes 3903 artists -- an increase of 518. Nearly all artists (98.2%) changed rank by 10 positions or more. And 529 artists moved over 1000 positions.
So having the full set of data makes a big difference!
It’s possible that Spotify needs the month of November to prepare the data in Spotify Wrapped, but the most common explanation I’ve heard is that they are concerned that holiday listening will impact the rankings. But of course, there are other ways to adjust for this that don’t require sacrificing ~17% of the year’s data!
One simply needs to incorporate a more comprehensive time measure.
DAUs & WAUs & QAUs, Oh MAUs!
A tech product framework that I quite enjoy is that of the DAU: the “Daily Active User”. It is applied mostly to applications or website, but you can apply it to almost anything. What cooking ingredients do you use daily versus monthly? What books do you think about on a weekly basis? What relationships do you want in each category? For the purposes of this exercise, though, we’re curious what artists are we engaging with on a daily or monthly basis versus those who perhaps we only interact with during certain times of the year.
On most days, I listen to between 7 and 30 different artists, though occasionally I’ll have days where I’m deep in the hundreds. Often I’ll pick specific songs to loop that will help me find my productive groove that will carry me through the day.
I can also visualize day-by-day listening to see how spread throughout 2024 my artist-level engagement was, and I can sort to identify the artists that I had the most consistent relationships with in the year.
Leading the pack here is Von Wegen Lisbeth, the only artist I listened to on Spotify every calendar month of this year. They didn’t make my top artists using play count as a metric, but they were on the cusp at #6 when accounting for data through the end of October.
Red Velvet is the only artist that overlaps with the previous top five, with Ritviz, Taylor Swift and Frozen Clouds rounding out the group. There do seem to be different listening patterns for different artists, and looking at “breadth-across-year” results in a very different ranking. This top five artist list, though, has lots of overlap with other years -- shouldn’t an artist of the year be uniquely of that year?
Relativity & What Makes 2024 Special
When you download your data from Spotify, you actually get all of your data from when you first started listening to music on Spotify. Below, I charted the average number of daily hours that I would listen to Spotify, starting in 2011.
In 2017, I began shifting to YouTube Music, drawn to their deeper library of vaporwave and covers. But then as I was developing tools and solutions to help people improve their media usage habits in 2021, I realized I liked the Spotify app development environment, bringing me back in.
With this much richer and broader set of data, we can examine what made this year different from others. And on first pass, it looks to be quite different! We can see that a large chunk of artists only have listening behavior in 2024, meaning they were new discoveries this year. And I did indeed to try to find a lot of new music this year.
The artists that pop out here have meaningful overlap with the top five artists that Spotify identified for me. By focusing on the top-left quadrant, I can find the “unique” 2024 artists -- at least within the context of listening on Spotify.
But how can we bring these metrics to align with the work we did last week? Can we have a unified theory of artist depth and calendar breadth to find my true Top Artist of 2024?
Space & Time, Together on One Graph
We’re getting close to everything we need!
Below I’ve flattened out our “artist breadth” analysis from last week’s Medi(a)tation into a single y-axis measure, and I’ve put it against the “presence” metric that reflects how broadly that artist was represented across the seasons, months, week and days of 2024.
The top right corner seems to have the main contenders of interest. I’ve used font sizing to indicate overall consumption, and it’s clear that these are all independent variables. We are close to finding our “true” Top Artist of 2024!
Next week, I’ll close out this trilogy with one more dimension of critique. And If you aren’t subscribed, be sure to subscribe so you don’t miss it!
This has been a Monday Medi(a)tation, a weekly data-driven reflection on my personal media usage. Each Monday, I will take a personal media dataset of my own and explore it to get a better understanding of the role media plays in my life.
If you are interested in having any of the above graphs created for your own media data, set up a time and I’ll walk you through how to pull what you need! Or, you can just message me and I’ll talk you through the process and share the output back with you!
A clarification to last week’s newsletter:
Spotify Wrapped Swap Group: I referred to the group with whom I participate in a Spotify Wrapped Swap as “former colleagues” when indeed they are better described as amazing friends. Or, as they identify, “frolleagues” -- a term of their recent invention for “friends” that are also “colleagues”.