In the future, estimates for individual composers can be refined as more obscure pieces are recorded (or otherwise documented) and as some works previously thought lost are rediscovered, as happened with Vivaldi’s opera Montezuma in 2002. As additional information becomes available, the sample could also be extended to include other productive, eminent composers such as Marc-Antoine Charpentier, Cherubini, Gounod, Honegger, Hummel, Lully, Meyerbeer, Milhaud, Palestrina, Alessandro Scarlatti and Telemann, reputedly the most prolific composer of all.On the other end, his sample stops with second generation modernism, and also several composers whose work was pushed into obscurity by totalitarianism. The data set should probably be doubled. However, its selection does tell us about what is valued today: super-prolific recent composers are at a disadvantage: Elliot Carter, Alan Hovhaness, Bohuslav Martinů – all examples of prolific composers whose important output was post-1940. Another blindspot are composers who went on to be film composers, or who had Jazz in their output. Sullivan, but not Frederick Loewe, Leonard Bernstein's West Side Story counted as part of his output but not Richard Rodgers or Stephen Sondheim? No Gershwin? But more on this in a moment.
But there are some insights yet to be gleaned from the data as it stands. Consider lifespan versus date of birth, giving a look at generational moments, a subject, since the Body Boomers are the generational generation, which receives attention. When do demographics come in to play most strongly?
There is the graph of date of birth against productivity, at the bottom for readability.
What can we see?
First, it demolishes the notion that complexity is the driver of production drop. Instead production rolls off the table past the early Romantic – and the beginning of the "low out put important composer" shows up in the wave of composers who would be important in the second half of the 19th century. This would indicate that economics, and life spans, are important. As it became possible to be a low output composer, and as the size of repertory grew, the pressure was on distinct work, and having some expectation of time to make it. Also visible are clusters of births, indicating that there is more than some investigation worth while into whether certain moments favored settling down to be a composer for some long term source of patronage. Again, the roll off the table effect is not based on style, the Bel Canto opera composers and supposedly early Romantics are the last of the hyper-prolific. The next note is that there is a clear gap in the 1848-1855 period, indicating, again, that outside factors come into play.
Finally the boundaries of the sample show up, with the end of the sample being essentially 1920, that is almost a century ago.
The way to remedy this defect is one of method: by using Spotify lists, calculation of a composer's complete output becomes easier to do, and crowd sourceable. Kozbelt's heroic effort in compiling this set, before the availability of playlists, must have consumed far more time than one would like to think about. I did a survey of String Quartet and Piano Trio times, averaging recordings, and then found I could use Petrucci, Sibelius 7, and iTunes to rapidly compile performance times of a vast range of composers using a simple Python script that would curl files, rip them to Sibelius and scan them, then electronically play them.
So the next wave of this project is to make it distributed and insertable, thus allowing a wide range of hands to contribute.