Re-sorting Pitchfork’s top albums of 2010-2014

The difference between the official and expected ranking of Pitchfork's top 100 albums of the decade so far.

Pitchfork just released their rankings for the best albums of the decade so far. Unsurprisingly, hipster favorites like Vampire Weekend and Kanye West won out (as any longtime reader of Pitchfork would expect), but, surprisingly, several relatively unknown artists or lesser-known albums by famous artists sneaked onto the list, including Earl Sweatshirt’s debut Earl and Frank Ocean’s first mixtape Nostalgia, Ultra. I’ve read before that Pitchfork tends to modify its editorial opinions in order to adjust to current trends in music, and so I was curious about the degree to which the assigned ranking matched an equivalent, “expected” ranking generated by comparing the numerical score that Pitchfork gave to each album at the time of its release. The above figure is a graph of the difference in ranking of the top 100 given by the “official” Pitchfork ranking, and a ranking generated by looking up the numerical score given to each album (in the list) upon its release and sorting the albums from lowest to highest score. The order of the vertical axis is the official Pitchfork ranking, from position 1 at the top to position 100 at the bottom. The bars indicate the difference in ranking for each album, which was generated by subtracting from the official Pitchfork ranking the expected ranking based on its numerical score after release. Large differences in the position on the list thus indicate Pitchfork’s relative opinion of the piece changing substantially by the time the “official” top albums ranking was compiled.

At least two of the albums that made the list, Earl Sweatshirt’s Earl and Jai Paul’s eponymous album, were so obscure at the time of their release that Pitchfork didn’t even rank them. In recognition of this fact, Pitchfork rated them both near the bottom of the top 100, and so their difference in ranking doesn’t seem that large on the graph. But the honorific inclusion of these two albums underscores a more general trend apparent in the Pitchfork list: an emphasis on contemporary R&B and hip-hop at the expense of electronica and downtempo. In a list sorted purely by numerical ranking, Beyonce’s Beyonce would not have scored as absurdly high as it does on the Pitchfork official ranking, nor would have Thundercat’s debut Apocalypse, which is the biggest winner in the ratings. These won out over albums like Reflektor or To Be Kind, which both showed relatively large drops relative to their expected positions on the list.

Pitchfork undoubtedly sees itself as a ratings site capable of setting the zeitgeist for a given decade, and so the emphasis on new artists and movements over indie staples like Arcade Fire or Swans suggests that the website sees the newer artists as representative of the next major movement in indie music. To this end, it’s worth noting that the most recent album declared “Best New Music” by Pitchfork before the creation of the ranking was FKA Twig’s outstanding LP1, which stands at a healthy position on the official list and which generally represents many of the stylistic frontiers of emerging indie music.

The relatively large change in Pitchfork’s opinion of albums is well-captured by a scatterplot of the numerical, review-based ranking versus the official ranking released by Pitchfork (shown below, concept originally suggested by reddit user gkyshr). Surprisingly, there seems to be barely any correlation between the two variables (the line y = x, corresponding to the case where Pitchfork’s released ranking coincides with the sorted ranking, is underlaid). This variation is captured by the mean of the absolute value of the differences reported in the bar chart, which came out to 20 (a surprisingly high value, given that the maximum change in ranking for a given album data is 99). It’s almost as if Pitchfork deliberately attempted to make its rankings differ from expectations, with the only albums really falling on the line corresponding to very highly rated albums, like the number 1 album, My Beautiful Dark Twisted Fantasy:

Scatterplot of rankings

In order to make these plots, I made use of Python’s outstanding BeautifulSoup and pandas libraries.

 

d.