Live data from Hacker News

How bad is your Spotify?

pudding.cool

411–420 of 618 posts

Re: How bad is your Spotify?

#412
post #7

This is hilarious... > You are 26% basic. You're trying to be cool with Kapitan Korsakov, but your favorites are the same as everybody else's.. > Your spotify was please-read-my-manuscript-walmart-hawaiian-shirt-sitting-alone-in-the-cafeteria bad. Ouch. Really interesting project by all means. I hope they release the source - I'm curious about how it was built. Edit: This same link was posted 11 hours ago https://new…

> You're stuck in the early 2010s. You only listen to Obama-era jams like Nocturne in B-Flat Minor, Op. 9, No. 1 by Frédéric Chopin and Piano Concerto No. 21 in C Major, K. 467: II. Andante by Wolfgang Amadeus Mozart.

The output it spitted out while processing was pretty good though.

Re: How bad is your Spotify?

#413
I've worked on various music recommendation systems just for my own private consumption for 15 years.

My current approach at least makes me really happy. The problem is I'm likely radically different.

The general problem is so are many people. But all in different, incompatible ways.

For instance, my (current) approach is fundamentally label based. Things are organized by label and I "rate" releases as I listen to them. This leads to a generalized "label rating" which can be partially extended to (1) a rating of the artists who released on the label along with (2) albums and eps which may see a release under another label. Then there is a similarity index to other labels based on a percentage of artists who release on both labels and this leads to something going into a big "consider" pile which then gets treated randomly. It's not really a "recommendation" as much as it is a "might be related? i dunno" system.

For people who want familiar songs they can sing along with, this is a terrible no-good system. For people who like to listen to the same song multiple times a day, this is equally useless. For those who want steadfast genre uniformity and things to sound more or less the same, this is also useless.

This is designed to be exclusively long-tail and stay firmly within the confines way out at the end of the tail. But that's exactly where I want to be.

In the pareto principle, everyone in the 80% looks more or less the same but everyone in the 20% is strange in a different kind of way.

There is no 1 recommendation system that works for all customers. Users have to be segmented. It's utterly inescapable. That's why these systems try to just service the 80% and just not put any effort to the other 20%.

The problem with this approach is the 80% is no longer 80%. More people are becoming snowflake 20%ers and this trend is likely to continue. The mainstream orientation engines of radio and cable television aren't influencing like they used to and the 'service the 80' approach just won't work any more. It's over.

The single recommendation system approach is being tweaked to try to service this coalition that no longer exists. So what happens?

Everything starts to suck. Google results, youtube recommendations, spotify, they all start to be unfocused, scattered messes that just discard all the signal, amplify the noise and return piles of nearly identical derivative garbage.

Socially we've moved beyond the assumption that simply everyone in society will be in some conventional nuclear family with the wife vacuuming at home and the daughter playing with dolls but technologically we still seem to model humans using utterly conventional and outdated single purpose models that simply do not apply and are increasingly a poor fit for everyone.

The new normal is that there isn't one and it's time for tech to play catch up.

Re: How bad is your Spotify?

#414
post #332
post #7

This is hilarious... > You are 26% basic. You're trying to be cool with Kapitan Korsakov, but your favorites are the same as everybody else's.. > Your spotify was please-read-my-manuscript-walmart-hawaiian-shirt-sitting-alone-in-the-cafeteria bad. Ouch. Really interesting project by all means. I hope they release the source - I'm curious about how it was built. Edit: This same link was posted 11 hours ago https://new…

i'm 1% basic, i'm clearly the coolest in da room

I got 0%!

Re: How bad is your Spotify?

#415

It looks like the API is overloaded. With regards to taste, you usually need to find a suitable gateway into an unpopular genre to begin appreciating it. I always hated jazz until I turned 30, when I heard the melodic jazz of Avishai Cohen [1] and Phronesis [2], which led me to Eric Dolphy [3] and ultimately to free jazz outfits like Alexander von Schlippenbach and Peter Brotzmann. Having grown bored of everything el…

I had a similar Jazz experience. Thanks to Spotify, I ended up going to Bad Plus, Brad Mehldau, Avishai Cohen and similar concerts and started to appreciate free Jazz and the local scene more, too. I found this incredibly enriching, and it has made me try to play the piano again.

Re: How bad is your Spotify?

#416

Earlier quoted context omitted.

You can disable sharing altogether if you want. This is specifically for a single "session." Like if you want to listen to a guilty pleasure and don't want your friends bugging you about it

Wait wait wait, Spotify notifies my friends of what I am listening to (not a user, so honest question)?

No no, but in the UI you can see what your friends are listening to (if you're added as friends). Example I found online (the right bar is the friends feed):

https://miro.medium.com/max/3836/1*C_ImLXQdMOGljMRhLcV-Yw.jp...

Re: How bad is your Spotify?

#417

As someone who listens to spotify all day while programming, here's my secret to success: I create a playlist to fit a mood (for example, ambient electronic music with a good beat and no words), and I seed it with at least 10 songs. Then, I go down to the "Recommended Songs" section and I just play whatever it recommends. If I don't like a song, I hit the skip button on my keyboard. The vast majority of what it recom…

Anyone here remember turntable.fm? I do. Sad. JQBX.fm isn't the same either. That was a great way to get a higher "hit" ratio though.

Re: How bad is your Spotify?

#418
post #7

This is hilarious... > You are 26% basic. You're trying to be cool with Kapitan Korsakov, but your favorites are the same as everybody else's.. > Your spotify was please-read-my-manuscript-walmart-hawaiian-shirt-sitting-alone-in-the-cafeteria bad. Ouch. Really interesting project by all means. I hope they release the source - I'm curious about how it was built. Edit: This same link was posted 11 hours ago https://new…

The algorithm is more basic than the music it passes judgement on so take the verbiage with a grain of punk-vinyl-you-never-heard-before.

Re: How bad is your Spotify?

#419
post #175

Earlier quoted context omitted.

It looks like they request these permissions: user-top-read, playlist-read-private, playlist-read-collaborative and they also gets your your name, username and profile picture.

Seems a neat way to collect information for their startup. Maybe a Spotify competitor, or a datamarket for selling people's music preferences.

The Pudding is, per the footer on that page, "a digital publication that explains ideas debated in culture with visual essays". It's been around for a few years already. https://pudding.cool/archives/

Re: How bad is your Spotify?

#420

Funny, but I was hoping this might provide some insight into why my Spotify is so bad, even by my own tastes. Spotify constantly queues and recommends songs to me that are so bad, I can’t even imagine how there could possibly exist any data indicating that any significant sample of listeners has ever enjoyed hearing them. Spotify has 5+ years of my listening history, and orders of magnitude more data from listeners a…

It seems to always give me the best songs first and last in the Discover Weekly and Release Radar auto generated playlists. But in the middle there are sometimes unbelievably ill suited songs for me. Maybe it’s the artificial intelligence attempting to probe the boundaries a bit.
Post reply on HN