Tuesday, June 5, 2012

Has Music Quality Gone Down Since File Sharing Began?

Joel Waldfogel of the University of Minnesota and the National Bureau of Economic Research recently released a pair of papers studying the quality supply of music since the beginning of Napster and file sharing have become common.  Anyone who is at all interested in music should be quite familiar with the file sharing issue.

In 1999, Napster became a dominant player in the MP3 delivery market, but they were allowing users to share music without charge or permission from the relevant owners.  Because of the copying aspects of this so-called sharing, it was actually an international network of gift-givers.  Music that was a commercial product became, in large part, a free good because it became impossible to exclude the freeloader.

There are some aspects in which the music industry has brought this upon themselves.  For one thing, the dominant medium at the time was the CD, which was actually comprised of music files that could be easily transferred from one computer to another, and consumers began to expect this functionality as a part of the purchasing price.  For another thing, music makers were very slow to adapt to this new medium which consumers preferred for reasons additional to its zero price.  The iTunes Store, which is currently a dominant seller of music globally started in 2003, four years after the beginning of Napster.


Economics predicts that when the revenue from a product declines, that producer would be less incentive to create it, and the supply would decline into a new equilibrium.  Joel Waldfogel investigates this supply issue in his paper, "Bye Bye Miss American Pie?  The Supply of New Recorded Music Since Napster".  He finds that the music supply has not declined since the introduction of file sharing.

Music studios were once a barrier to entry.  Could they now be a barrier to high quality?  (photo: David Boyle)

He notes an important factor being the simultaneous decrease in cost of bringing music to market.  The marginal costs of creating music declined at almost the same time as demand (in the form of purchases) declined.  Studio time is much less expensive now, and passable sounding recordings can even be made with just a few microphones and a laptop.  Distribution also used to be much more difficult, and it was the main reason that artists went through a small set of so-called 'major' labels to sell their music to a wide audience.  Distribution on the internet has become simpler, and even promotion has become less stratified from a small number of artists reaching huge audiences to a larger number of artists reaching smaller audiences.

Joel Waldfogel also examines the issue of quality.  It is difficult to examine music in terms of supply because it is not at all uniform, and supply suggests homogenous products.  Quality is difficult to examine because it is very subjective.  If you ask ten people what songs they like, you might get ten very different answers, and they'd all be correct.  While there is a certain amount of overlap, some songs are much more popular than others, and that can be taken as an indicator that many individuals find that song to be of higher quality than the other.  Music criticism can also be taken as an indicator of music quality.  They often rank, or attach numerical values on their reviews.

Waldfogel forms metrics based on music critic reviews to ascertain whether music quality has declined since Napster in his paper "Copyright Protection, Technological Change, and the Quality of New Products: Evidence from Recorded Music Since Napster".  He finds that there has not been a decline in music quality since Napster and actually notes an increase in quality since 1999, coincidentally when Napster was founded.


I am very interested in music, and have been purchasing it all of my life.  I was also an early adopter to iTunes, which allows users to rank their songs in a 1-5 star metric.  I have always ranked my songs, and after reading Waldfogel's paper, it made me wonder what my personal data would say on this matter.  Has music quality declined according to my musical rankings?

One thing that strikes me about these decade by decade music rating proportions is the difference in number and proportions for music in the 1990's and 2000's and earlier decades.  The increase in numbers can be explained easily as those were my main years of music collecting.

In the 1990's and 2000's, I was old enough to understand the cultural context of music, but not much before those years.  Much of my understanding of earlier music has been from a range of ad-hoc educational sources.  That might be music radio, friends, bands that I was aware of citing influences or performing cover versions, books, etc.  The point is that I cannot fully appreciate the greater cultural context of the songs at their release as I was not a part of their initial audience.  I have a different understanding of them.  Perhaps it is more musically based, and/or perhaps it is based on a different cultural context.  The point is that it is definitely different from people who were a part of the audience at the time.

My number of songs for the 1960's, 1970's and 1980's and almost identical, and the proportions are similar as well.  The 1990's show growth in total songs, but especially in three star songs, and the 2000's show yet more growth, especially in two star songs.  I don't think that this suggests a degredation in music quality, but rather that my collection was less selective for those decades as I experienced them personally, and collected music in those years.  The piece of evidence that I do think could suggest a deterioration of high quality music is in the number of five star songs.  Not only does the proportion of five star songs deteriorate through the years, but it also declines in real numbers after their peak in the 1970's.

As with most music lovers, I am constantly attempting to collect high quality songs.  My five star songs represent ~ 7% of my music collection.  While I certainly do not have any pre-conceived bias towards the music of the 1970's, but by my own inimitable criteria, I have chosen more five star songs from that decade than any other.  Despite the fact that I own almost twice as many total songs from the 2000's, I have almost half as many five star songs.  I think there are two ways to view this data.  Music of very high quality has declined according to me, or I am more reluctant to judge current songs as five star songs.  It could be a bit of both.


I think there are two questions that this raises.  First, even if the general quality of music has not declined, is it possible that high quality, or five star music has declined, or is this a feature of my own bias?  Second, I think this suggests a clear difference in the way that I appreciate music that Iwas alive for and aware of when it was released as opposed to music that I learned about as a part of musical history.  I wonder if music critics have the same bias.  I argue that because Waldfogel has relied heavily on music criticism by music critics that are younger than much of the music in his dataset, this alters those critics' perception of that music.  It does not render them incapable of judging music quality, but their reviews of music released throughout history is likely very different than the music released currently, as their perceptions of the two are not the same.

Waldfogel's conclusions were that music supply and quality have not significantly declined post-Napster (1999), and his pursuit of these hypotheses is a welcome addition to economic and music appreciation literature.  My efforts here are not to undermine his efforts, but rather to add a few notes to their foundation, which I believe is already quite strong.  I think that it is an interesting note, that while perhaps music quality in a general sense has not declined post Napster, perhaps high quality music has.

1 comment:

  1. Interesting perspective.

    A related hypothesis you might subject to the data: Has the threshold in music quality declined over time?

    Imagine you have to meet a certain threshold in forecast music quality to go through with making a record. High quality music (of any era) meets that threshold, but average quality stuff hasn't always been worth making (and you never want low quality music).

    The most striking part of the stacked histograms to me is how many more average and above average songs were made in the later decades, which is consistent with this idea that the threshold for recordable quality has gone down (subject to your critique of what quality really is).

    By the way, in taking the *fraction* of high quality songs as your measure, you automatically put the number of average songs in the denominator (and as a result, attribute the rise in average-ness to a fall in high quality). I would be curious to see a t-test for a difference (or an F-test across decades) in the counts of high quality by decade. Visually, it is smaller, but I'm not sure it is significant.

    For what it is worth, you would want to look at fractions if you thought that there is no screening process on quality that censored the distribution from below (or alternatively, if you think that music quality is realized after the decision to make the music), but I think censoring is probably important, especially in earlier decades when you couldn't just record your own album and upload it to YouTube.