Adventures with Discover Weekly
Because I couldn’t find any albums coming out this month that I wanted to review for the album digest, I decided to let Spotify pick the albums to listen to. I listened to my algorithmically chosen Discover Weekly playlist one week and selected albums based on the songs that I liked the most. The album also had to be released in 2016. The selections are ones that got away.
I’ve done this before. Three of the four albums that I wrote about in August four years ago were selected by Bleep: they were in the ‘best albums of the year so far’ bundle that I bought. So this month’s album digest is bought to you by the same idea, just a different curator.
In general I find that DW is OK at finding songs that I like. It picks thirty songs a week and it cleverly doesn’t bother to change them unless you’ve listened to at least one song. You can’t decide that you don’t like the look of it, you have to give at least one song a go. That may have led me to more discoveries than anything else. Most weeks it picks at least four or five songs that I’d listen to more than once. It knows that I like long remixes and proggy instrumentals. For a while I found its selections rather morose, but it’s cheered up in recent weeks.
It’s starting to learn that I like new music too. Today I listened to an hour of this week’s DW and about seven of the songs were released in either July or August. You can see now why I am interested in what it might pick out for me in the next few weeks.
However, in the album digest I won’t conjecture as to why I ended up listening to those particular albums. I need to follow its suggestions for a bit longer before I start to analyse them. A musical prediction machines raises all sorts of questions, the kind that I like to ponder the answers to. Not least the decreasing moroseness! Is it making me happier? Am I feeding it happier data points? Or was it because I complained about it to @SpotifyCares on Twitter? Watch (or listen to) this space and perhaps I’ll figure it out.