On this Mathy Monday I reckoned it might be fun (depending upon your definition of the term) to take a peek as some data related to music consumption patterns online. The purpose of this post is pretty simple: to return to the question of the general shape of consumer behavior as that behavior relates to cultural goods, in this case music.
Reeling the clock back a few years, remember that the underlying premise of so-called Long Tail Theory was that broader availability of and access to products online would lead “the Tail” of this inventory accounting for the majority of consumer interest. The days of super-hits — what I like to call Pinhead theory — would likely be over. Furthermore, there might also be a sort of Pot Belly Theory, one that suggests consumption of media will expand in “the Belly” of these sorts of distributions.
Will Page and Eric Garland engaged in a similar inquiry a few years back to the one taking place in this post (my name showed up in there as well, in the acknowledgments). In that work, data from peer-to-peer network behavior was under the microscope.
This time around, I reckoned it might be fun to return to these prior questions of Pinheads, Pot Bellies, and Long Tails with some music video data as the raw ingredients. Comments on the underlying data can be found at the end of this post. These data come from a moment in time after the work of Will, Eric and others. And I hope to follow this post with observations of subsequent data, from different slices in time, to see if any changes are afoot (or, a’tail).
Ultimately, at least in my mind, (a) culture is what we all share, so these data in some way reflect culture, and (b) its quite easy to glimpse at these patterns of consumption in simple percentage terms, no logarithms or bio-rhythms required. Per protocol, I am not taking a side in the debate over whether any general pattern of consumption is good or bad or otherwise for the industry or society. Instead, I reckon we might as well be permitted some glimpse at these patterns so we can have some sense of what “is” at some moment, at least in some corner of the internet.
And so, below you will find a couple graphs to visualize what I observe in these data:
In the first graph, the left-side Y-axis is the Total Number of Views for the viewed videos. These data are graphed as a continuous bar, from left to right in the X-axis queue, from the most viewed to the least viewed video. The right-side Y-axis is the Cumulative Percentage of all Video Views. For example, the cumulative percentage of video views for the tenth video in the queue would be the sum of all views for videos 1 through 10 divided by the total number of views for all videos.
In short, 1.05% of the available videos account for 90% of all video views during the week in question. 2.67% of the available videos account for 99% all views.
In the second graph I simply try to provide some sense of scale were the first graph to include on the X-axis the entire queue of available videos — those viewed during the period and those never viewed.
Frankly, only 4.2% of the available videos were viewed during the week in question, so the graph would have to be 20x wider if I were to include in the analysis all of these videos that collected digital dust during the week.
The underlying data for this inquiry shall not be disclosed, neither in terms of the source nor the metrics of size (e.g., number of videos, number of views). Therefore, I simply ask the reader to accept that this blob (i.e., sample) of data is sufficiently large — in terms of available videos and videos viewed — to be at least representative of general patterns online during a single week of music video enjoyment. Furthermore, the reader might guess that the nifty accoutrements assumed to lead to broader consumption of media are present in the underlying video venue.