🎵Analyzing the Top 100 Albums of 2018

and all the other ones, too

Reviewers Give Music 7.4

In 2007, The Onion published one of my favorite pieces: Pitchfork Gives Music 6.8.1 It cut deep at 16-year-old me, who would stay up until midnight every night to read the latest Pitchfork reviews2 so I would know what to think about new music. The story points out the insanity of rating music in the first place, as well as “slavishly follow[ing] the website’s advice.”

Despite having developed more of my own taste since high school, I still find value in Pitchfork and other sites to keep up with new music and see what might be worth my time. Album of The Year is one of my favorite resources for accomplishing these tasks. It catalogs reviews from 150+ websites, putting all the music and stats you could ever want in one place.

For this project, I’ve scraped AotY for all of 2018’s album reviews and end of the year lists to gather some insights about music and its criticism in 2018.

Albums in this analysis match at least one of two conditions:

  1. The release is considered an “LP” (in contrast to, say, a “mixtape”) by AoTY, has at least 5 critic reviews, and was released in 2018 (i.e. it appears on this page.
  2. A publication ranked it in their end of the year list (i.e. it appears on a list linked from this page.) Some publications put obscure albums (those with fewer than 5 reviews) on their lists, and some also include releases from 2017.

Overall, this lists are quite diverse; of the 1554 unique of the albums meeting the conditions above, 1334 of them made it on at least one of the 133 lists.

The Big List

Album of the Year aggregates the rankings of the 133 lists by awarding 10 points for a 1st place ranking, 8 points for 2nd, 6 points for 3rd, 5 points for placing in the top 10, 3 points for placing in the top 25, and 1 point for anything else (outside the top 10, or in an unranked list). Almost every album on AotY is categorized by a release data, genre, and label. We’ll look at trends among each of these pieces of metadata.

Release Date

When was 2018, Anyway?

A few sites are guilty of putting 2017 albums on their 2018 lists. Most of these placements are forgivable - Sidney Gish’s “No Dogs Allowed” from New Year’s Eve is certainly close enough to 2018 to count. Others get off on technicalities. Piccadilly Records places three 2017 albums in their top 100; however, they all saw vinyl releases in 2018. Jon Pareles’s pick of Jupiter & Okwess’s “Kin Sonic” might look egregious, but its American release was almost a full year after its European debut. See the table for all of the 2017 albums that made an end of the year list.

Artist Title Publication Date
Jupiter & Okwess Kin Sonic The New York Times: Jon Pareles Jul. 7
Cornelius Mellow Waves Piccadilly Records Jul. 21
James Heather Stories From Far Away On Piano Drift Aug. 18
Mattiel Mattiel Piccadilly Records Sep. 22
Maryam Saleh, Maurice Louca & Tamer Abu Ghazaleh Lekhfa The Quietus Sep. 22
Interrobang‽ Interrobang‽ Louder Than War Oct. 20
Mildlife Phase Piccadilly Records Nov. 2
Jim Ghedi Hymn for Ancient Land Drift Nov. 10
Björk Utopia MOJO Nov. 24
Simple Affections Simple Affections Tiny Mix Tapes Nov. 28
Nabihah Iqbal Weighing of the Heart DJ Mag Dec. 1
Neil Young & Promise of the Real The Visitor MOJO Dec. 1
MALK Death From A Love Norman Records Dec. 8
Dalham Janus Norman Records Dec. 11
Charli XCX Pop 2 Tiny Mix Tapes Dec. 15
Drakeo The Ruler Cold Devil Passion of the Weiss, SPIN Dec. 26
Sidney Gish No Dogs Allowed FLOOD, The Alternative Dec. 31

The most common month of release for a Top 100 album was June, with 18 of its 154 albums (12%) of its albums making the list. July under-performs, only contributing 3 of its 104 albums (3%) to the Top 100.

Releases in December make up only 1% of the releases for the year and none of the Top 100.

There are a few explanations for this I can think of:

  • Most lists are created in late November or early December, so December releases are not typically in consideration for best of the year lists. Occasionally, a release from December of the previous year will make the next year’s list (e.g. Run the Jewels 3, released December 25, 2016 earned 38 points in the 2017 rankings)
  • Albums released in December may not get reviews until January because many sites stop publishing reviews over the holidays.

Genre

The Top 100 features albums from 61 distinct genres. An album may belong to more than one genre, but 93 of the releases belong to two or fewer. Julia Holter’s “Aviary” spans the most genres, being classified as Progressive Pop, Art Pop, Baroque Pop, and Ambient Pop. Outside of the Top 100, two other releases this year managed to hit four genres: “Everywhere at the End of Time - Stage 5” by The Caretaker (Sound Collage, Noise, Dark Ambient, Glitch) and Moaning’s “Moaning” (Post-Punk, Noise Pop, Indie Rock, Post-Punk Revival).

It looks like Indie Rock Isn’t Dead (at least not among music publications) - it was the most represented genre in the Top 100, as well as the most represented genre overall. While “R&B” and “Alternative R&B” both make an appearance in both graphs, it’s interesting to note that a generic “Rock” does not, despite “Indie Rock” being so prevalent. This is a bit of a peculiarity of AotY’s genre classification. Rock is broken up in to many categories, with lots of releases this year: Indie Rock (118), Alternative Rock (42), Pop Rock (27), Folk Rock (17), Psychedelic Rock (15), Hard Rock (12), Noise Rock (11), Art Rock (10), Experimental Rock (10), Garage Rock (7), Blues Rock (5), Industrial Rock (5), Post-Rock (5), Stoner Rock (5), Glam Rock (4), Progressive Rock (3), Country Rock (2), Krautrock (2), Math Rock (2), Punk Rock (2), Acoustic Rock (1), Funk Rock (1), Heartland Rock (1), Rap Rock (1), Rock (1), Soft Rock (1), Space Rock (1), Surf Rock (1). “Pop” music gets a similar treatment.

Genre over-performers include Art Pop, R&B, and Alternative R&B, while Indie Pop, Trap Rap under-perform. Interestingly, there were more Alternative R&B releases (33) than “regular” R&B releases (27) this year. Perhaps a genre renaming is in order.

Label

Frankly, I couldn’t tell you what label most of my favorite artists are signed to, so my insight here is a limited. To me, the first chart looks like a list of labels that I’ve heard of and kind of like, while the second looks like a list of labels I’ve heard of, but don’t have any strong opinions toward.

Under-performers, Snubs, and Upsets

In addition to computing an end of the year ranking based on list appearances, Album of the Year also ranks albums by their average critic score. In tables below, I’ve computed the difference between the two ranking methods.

Under-perfomers

These are the albums that had high critic scores, but weren’t ranked highly on end of the year lists. These releases seem to be by established acts whose music isn’t particularly “now” - exactly what you might have expected. These are good albums that aren’t especially relevant to defining the year in music.

Artist Album Genre Score Rank List Rank Difference
Ty Segall Freedom’s Goblin Garage Rock 11 63 52
Mount Eerie Now Only Singer-Songwriter, Indie Folk 8 59 51
Stephen Malkmus and the Jicks Sparkle Hard Indie Rock 11 62 51
Grouper Grid of Points Ambient Pop, Singer-Songwriter 12 63 51
Ezra Furman Transangelic Exodus Indie Rock 8 58 50
Kamaal Williams The Return Jazz Fusion 15 63 48

Snubs

These are the albums that had high critic scores,but didn’t make it into the top 100 AotY list. I think my comments about under-performers are valid here as well - these are good albums that don’t represent 2018.

Artist Album Genre Reviews Score Rank
Sleep The Sciences Stoner Metal, Doom Metal 10 3
Years & Years Palo Santo Electropop, Dance Pop 14 3
Hookworms Microshift Neo-Psychedelia 18 3
Hot Snakes Jericho Sirens Garage Punk, Post-Hardcore 12 4
Marianne Faithfull Negative Capability Singer-Songwriter 13 4
Rival Consoles Persona Microhouse, IDM 12 5

Upsets

These are the albums that had better end of the year list performance than their critic scores would predict. These are sort of the opposite of the other two lists. Every one of these releases could be classified as “pop” - they’re good albums that are particularly representative of 2018.

Artist Album Genre Score Rank List Rank Difference
Kacey Musgraves Golden Hour Country Pop 12 3 9
Arctic Monkeys Tranquility Base Hotel & Casino Psychedelic Pop, Art Rock 23 14 9
Janelle Monáe Dirty Computer Art Pop, R&B 7 1 6
Mitski Be the Cowboy Indie Rock, Art Pop 8 2 6
Cardi B Invasion of Privacy Trap Rap 11 5 6
Ariana Grande Sweetener R&B, Electropop, Pop 17 11 6

Comparing Lists

In this section, I’ve compared lists using an ad-hoc similarity score3, which scores two lists similarities on a scale from 0 (nothing in common) to 100 (exactly the same).

The first table show the lists that were most similar to the final Album of the Year list. In general, these were longer lists (50 albums) from “major” publications. Depending on your view, these might be the most “boring” or “authoritative” lists.

Publication Similarity
The Guardian 82
Pitchfork 75
The Line of Best Fit 67
SPIN 65
Consequence of Sound 62

In the next table, I’ve compared all of the lists to each other. The most similar lists were Yahoo Music and Refinery29. They’re both short lists, with only 10 albums each - 6 of which are found in both lists. Similarly, Us Weekly and Refinery29 share 5 albums. Among the longer lists, Pitchfork and Esquire are quite similar, with Pitchfork’s 50 album list including 17 of Esquire’s 20 ranked albums.

Publication 1 Publication 2 Similarity
Yahoo Music Refinery29 73
Us Weekly Refinery29 71
Pitchfork Esquire (US) 70
The Line of Best Fit The Guardian 70
The Interns The Daily Beast 69
SPIN Pitchfork 69
The Interns Billboard 68
Vibe Complex 66
The Interns Refinery29 65
DJBooth (Hip Hop / R&B) Complex 64

Final Thoughts

This project was a much bigger undertaking than I expected. I was only able to half-finished a similar post in 2017, but I now have the infrastructure setup so that I should be able to do it again next year and start to analyze some trends. If you’re interested in doing something similar, you can download the dataset I made and look at the relevant code on my GitHub. Comment on my post on Reddit if you have any feedback or want to see additional stats or figures.


  1. By my calculations, this year, they give it a 7.2.

  2. Only look at the scores

  3. For each publication, I reverse the rankings (e.g. in a top 50 list, the number 1 album gets a score of 50 and and the number 50 album gets a score of 1. All the albums not on the list get a score of 0). Then I compute the cosine similarity between all of the list vectors and multiply by 100.

Avatar
Damon Bayer
Statistics PhD Student

Related