Finally, a nearly two-year effort to re-balance my classical music listening efforts came to a conclusion.
Since 2008, I’ve been ‘scrobbling’ my listening habits (i.e., auto-transmitting them upon the completion of each ‘play’) to Last.fm. In that time, I’ve racked up (at the time of writing) 161,958 ‘listens’ to something or other, comprising the works of about 560 different composers (approximately, because sometimes what Last.fm classifies as a ‘composer’ isn’t what I would call a composer!)
Anyway, that’s all fine and lovely, except for the (to me) disturbing discovery that 10 of those 560 composers accounted for well over half of all my plays. In other words, Britten, Bach, Vaughan Williams, Handel, Shostakovich, Purcell, Rameau, Mozart, Verdi and Sibelius together accounted for somewhere around 80,000 plays. The other 550 composers had to share all the other 75,000ish listens amongst themselves! It seemed a little imbalanced for my liking: it resulted in a very top-heavy listening list, with an extraordinarily long tail of hundreds of composers who barely had more than a handful of ‘listens’ each.
So, in August 2019, I decided to do something about this wonky listening habit by implementing a self-denying ordinance: stop playing music by the top ten composers! Now, I didn’t stop completely (they’re top ten composers for a reason, after all: I like their stuff!), but I did manage to cut way back on them, so that since August 2nd 2019, I’ve only played something of their output 2,900 times (and most of that took place in November 2019 and 2020, to celebrate Britten’s birthday!) The other 20,020 plays of music since that time were all of composers who were not in my top ten. That would clearly tend to boost the proportion of listens by non-top-ten composers and to reduce the significance of the top ten.
And sure enough, today was finally the day when (with a little excusable rounding to two decimal places), I finally reached parity between the top ten and the non-top-ten:
That graph shows you the top ten getting progressively less top-heavy (in blue) and the non-top-ten getting progressively more significant as a proportion of total plays (in yellow). Today the two lines finally met: 50% a-piece!
This is good news because it means I can now unblock my top-ten! Specifically, I use my own AMP software to play music and that has been using an excludes.txt which has listed my top ten composers. The effect of that is to 100% prevent AMP from ever randomly selecting one of those composers for play. Today, I deleted my excludes.txt!
Cunningly, AMP randomises its plays by first selecting a composer and only then selecting something by that composer to play. That means every one of my 500+ composers has an equal chance of being selected by the first randomisation process: so Panufnick has as good a chance as Johann Sebastian Bach at being selected as the ‘random composer to play’, despite the fact that I’ve got at least 200 times as much recorded music by Bach as I have by Panufnick.
In other words, from this point forward, every composer -whether in or out of the top ten- has as good a chance of being selected for randomised play as every other, which means that although my top ten composers are now available for selection once more, the proportion of total plays they represent will continue to decline over time (though more slowly than they have done of late). So long as I keep using AMP to play music at random, in other words, my ‘long tail’ problem will continue, slowly but surely, to disappear.
But I shall be able to enjoy listening to Vaughan Williams and Britten once more, as of right now! Hurrah!