Starting a Little Project – AP Music

Music recommendations are broken. Amazon can’t even be bothered to tell me when artists I’ve purchased music of before release something new. And no matter what playlist or artist I start with on Spotify, I seem to get the same recommendations sent my way (which often feels like label-sponsored recommendations based on how infrequently they fit with what I’m trying to listen to). Pandora is a bit better, but one too many like and suddenly your playlist is overly key-holed and it’s difficult to find something new.

So, I wanted to take a stab at doing it differently. In my opinion, there are a lot of different variables that play into the equation, and I’ve written about how Launchcast better understood parts of this years ago. However, besides just how much an individual likes something, there are other key factors as well, like who are they influenced by, who do they tour with, or who do they collaborate with. So I want to work with some of that data to create a more holistic recommendation approach.

I’ve had the core of this project in my mind for a while, but never got started on it for a few different reasons I’ll talk about in future posts, but one of the things that got me over the starting line was the emergence of the fediverse. Originally, trying to build one comprehensive service came across as an overwhelming task. However, the idea of having multiple servers that could support more focused communities was an exciting one.

So, that’s what I’m getting started with, and I’ll be documenting parts of the journey as I go…