Hacker News Reader

Comment by AlphaWeaver | Hacker News Reader

Really? My only experience with Discover Weekly (and any other suggestion algorithms, i.e. radios, etc...) has been pretty terrible... It doesn't seem to suggest songs I like at all, and recently it seized upon one single song in one of my playlist and only plays covers of that song and the other songs in the album for my discover weekly for the past several weeks...

I love Spotify, but I use Pandora sometimes _exclusively_ for music discovery, and then add to my Spotify.

View on HN

I posted this earlier but it got buried: I had the same experience as you and built JQBX[1] to mix it up. It's kind of like Pandora (it's a lot like turntable.fm if you ever used that) but it's user generated so you tend to get more obscure / timely recommendations.

[1] https://www.jqbx.fm

wow kinda reminds me of groovesharks dj feature, missed that.

Is there anything close to replacing grooveshark? I loved the coding station, plug.dj that tried to replace that was just some guy playing tons of video game themes

This app is simply amazing. Thank you!

Are you worried about any Spotify terms of service issues?

We love JQBX at my office. Much more friendly than our earlier solution to pass around the line in cable between everyone’s computers all day! Great work

Unfortunately large portions of your web app seem to have aria-hidden set to true on them, which means I can't use it at all with a screen reader.

This is my experience with Discover Weekly as well. Everyone seems to be really divided on how well their algorithms work, they either work really well or terribly. I think it's because the quality of signals their algorithm can key off of is so limited due to it being such a passive medium.

There's no way to know if I listened to that track in full because I liked it or because I had it on in the background while doing other work and it wasn't distracting enough (usually the latter is the case for me). Furthermore if I skip a track they don't know if I skipped it because I didn't like it or I wasn't in the mood for that style of music, or I was just sick of that song cause I've heard it enough in the past X listens.

There's no way to easily signal to spotify why I'm taking an action, and yet their algorithms are trying to make assumptions over every little action.

Add songs to your library. It's given more weight than simple up-thumb/down-thumb.

That may be why my recommendations seem to work so well, unlike a lot of other commenters on this article.

For posterity, I have a little under 1000 tracks saved from 200 or so artists (of which I am following around half), around 50 favorite album playlists and about as many "I should give this album a listen" playlists.

So not really a huge collection in absolute terms. It is very much focused on hard rock and metal, though.

I think part of the online music discovery problem is that there is zero social context for it. Using Spotify to discover new music is, at least for me, the media equivalent of eating in a restaurant alone. Sure it's doable, and it's even preferable sometimes, but you're still missing one of the core elements of the experience.

What's the normal social context for discovering music? I ask this as someone who has done 99%+ of my music discovery in the last 5 years privately.

The same as the social context for books, film, or any other kind of media: you talk about and experience it with your friends.

Oh, I get it now. I had a kid, and don't really talk to friends anymore.

That's why I prefer 8tracks for music discovery. User generated play lists still give me a much better results than algorithms. I don't always want something that matches what I already like but instead new stuff that compliments it.

I think that that's a nice idea, but in reality, my experience is that even people who like similar interests can vary _wildly_ in music choice... I think that social discovery of music just leads to less crystallization of tastes and more people liking a "mainstream" song instead of music choices that actually interest them...

Anecdotal but mine have been pretty great. Of course there are songs that I'm not a fan of but for the most part its recommendations have been spot on.

Reply on news.ycombinator.com