2002: Last.fm and Audioscrobbler Herald the Social Web

The best ”algorithm” for discovering new music was digging through profiles on last.fm back when the social functions of the site were still active. Sure, it was a lot of manual work, but the results were amazing. It wasn't completely blind, I found that people I had high similarity with, it was more likely I'll like what they like, even across different genres. Sometimes people were nice and took the effort to recommend based on my profile. I got introduced to varied music, different genres and even a bit from different countries.

The worst was Pandora, which did recommendations based on breakdown of musical instruments and elements in the song. It did what it aimed to do pretty well, only it was a bad idea. It gave you a lot of uninspiring music that sounded like a bland copy of something you actually liked.

Spotify's recommendations are not super awful, but definitely feel closer to Pandora's style. I wonder why is the result like that even though I'm sure they train their model based on listening history.

I find it funny and sad that people get so excited about those Wrapped year-end things on Spotify when these companies are basically withholding all this data all year long and then pretend like it's a special treat when they doll out a peek at it once a year.

It feels to me like "dark mode" (which is a merely single color of customization for an app). We expect so little from our software and services that even these little, previously common features are supposed to be a treat.

Anyway, Last.fm was great -- I never used it that much for discovery, but rather to get insight into what I was listening to. Largely, it didn't say THAT much about my habits because I mostly just listened to my collection on random. My top bands were, for the most part, the bands I had the most of.

I'm a big fan of last.fm.

If you use Spotify, you can download your full listening history here: https://www.spotify.com/us/account/privacy/. You get it in a pretty convenient JSON format and with a little bit of code it's pretty easy to create some visualizations.

There are also websites for visualizing this data. I'm quite fond of this one: https://explorify.link/. It allows you to do some custom queries.

Last.fm is still used quite a bit, mainly as a listening history tracker rather than a radio or recommendation engine.

Spotify is still the only big streaming service with native platform-level scrobbling. For everything else it's a lot more DIY, usually with third party tools at the device level.

A big reason it’s still relevant is the ecosystem around it. The API hasn't really changed in 15 years, which makes it easy to build tools where a username alone is enough. That kind of lightweight social integration has mostly disappeared elsewhere.

Today, the social / community side is almost entirely just Discord. Nearly every music related server has a bot that displays Last.fm stats. My estimate is that abut 10% of Last.fm their users are also active in Discord music communities.

(Disclaimer: I run .fmbot, a Discord bot that integrates with Last.fm.)

I love last.fm with all my being. I recently created a ListenBrainz (same org as MusicBrainz) account which is an open source alternative that you don’t have to host yourself. I’m scrobbling to both places now just in case.

Check out tapmusic.net too to make cool diagrams out of your scrobbled music.

Also from this era and loosely related.

https://en.wikipedia.org/wiki/Oink%27s_Pink_Palace

Still scrobbling since 2008. A lot of smaller artists used to upload their music to last.fm, and I found a lot of gems there (specifically in the swedish bitpop scene).

Ha! I am still using Last.fm 21 years after first discovering it. I would say my current music taste is largely thanks to Last.fm and its compatibility and "Similar Artists" features.

I always thought Apple missed a huge opportunity to build a social network on top of iTunes.

See what your friends are listening to, develop communities around shared musical interests, get better recommendations. Sort of like YouTube now.

When I read the negative take on "it's always somebody else selecting the music for you" I really recoiled. My favorite way to listen to music today is BECAUSE there is someone choosing it for me. I love the human stories behind the music, and it is totally missing with algorithmic stuff. I love Gilles Peterson, and Derek Smith on KMHD, for example, exactly because they are terrific and interesting people and they bring that humanity with their choice of tracks. When they interview people it is so much more interesting as a companion to the music.

My favorite thing about Napster and LimeWire was when you could find a song, and then BROWSE the hard drive of the person hosting that song. It was so interesting to find house music and be digging through the tastes of someone in London. And, then chatting with them, and discovering the live scenes, the people behind the music, etc. I loved that and nothing has ever replaced it.

Having said all this, I am interested in playing with "scrobbling." Anyone have any advice on how to get started? Do you need a music library? Is there a way to import your playlists from YouTube music? I'm not a spotify person.

I just moved my scrobbling to a self-hosted instance of Koito after switching from Spotify to Jellyfin. Very happy with the change, as I can still share all my music data with friends

Last.fm was probably my first social network, although it probably doesn't make it justice to call it that! I am still scrobbling after so many years! Loved this article. Really good memories... Thx for sharing

https://listenbrainz.org/ is an open source scrobbler, with the advantage that it leverages the musicbrainz database and connects listens to artist and track IDs instead of names, avoiding duplicate confusion. You can keep last.fm and submit to both of you like.

Made a scrobbler program for cmus when I switched to it. [0]

Comparison of listeners really nails the recommendations. Similar minds like to listen to similar things.

0: https://github.com/vjeranc/cmus-status-scrobbler

I'm still scrobbling after all these years.

I got a last.FM story. I used last.fm for years, then wanted tot use it again after a hiatus. But was locked out of my account, and the email I used of that service didn't exist anymore. I contacted last.fm saying what the old email was, and told them the profile picture was from a guitarist from an obscure islandic band, who's blonde hair I photo shopped into brown, so he looked more like me. They gave we a reset link. Very happy to use it again. I do wonder how they stay afloat.

Earlier this year I decided to move away from streaming platforms and rebuild my local music collection and serve it out over Plex. Plex supports last.fm so everything gets recorded there.

I also use the following docker containers on my home server:

Multi-Scrobbler: https://hub.docker.com/r/foxxmd/multi-scrobbler Koito: https://koito.io/guides/installation/

This allows me to share my last.fm input to both a local scrobbler (Koito) and to listenbrainz - I figured having this data in multiple locations makes it a bit more safe.

Honestly between last.fm and listenbrainz I find myself exploring more on listenbrainz - even though most of it's users don't really fit the same listening profile I do.

Still a member of Last.fm, scrobbling since 4 Jan 2007 with 283,262 scrobbles.

I miss the old last.fm. I know it's still there, but it's not the same since CBS took over and made everything rely on youtube or whatever it's doing these days.

I worked at Last.fm from 2007 to 2012. The MIR team (think: research) developed a wonderful system called "RadioQL", which allowed you to stitch together custom ratio stations from any of a huge host of factors, joined together by AND, OR, and NOT. You could select artist radios, song radios, tags, and so on, but also combine this with things like the BPM or even some sentiment analysis. It was used a little bit inside some public-facing radio stations, but nobody outside of the staff ever got full access, and that's a tragedy as it was glorious.

Part of the quantified self movement. https://en.wikipedia.org/wiki/Quantified_self

The thing with data is that you have to act on it for it to be useful, and this data is useful only to recommendation engineers. Spotify's end-of-year summary is more than enough to satisfy my curiosity.

Richard Jones is still alive and kicking https://x.com/metabrew

Been scrobbling since 2008. I found out what last.fm was thanks to installing Rockbox on my iPod nano, and seeing they had this "scrobbling" feature. Had to remember to plug in the iPod, pull the scrobble log and upload it using a website someone created.

If you need a scrobbler for Android (and Linux) I recommend Pano Scrobbler: https://github.com/kawaiiDango/pano-scrobbler

Lastfm stopped me moving from Spotify to Apple Music; none of the iOS scrobblers seem to consistently catch all listens.

I still use last.fm. Been continously scrobbling since 2004. I also export my last.fm and Spotify listening history every now and again just in case. I plan to one day make a timeline of my listens overlayed on top of world and personal events.

I made friends I still have by browsing people who had a compatible music taste to me and then reaching out to chat.

last.fm was so promising when it came out but i have to say i didn’t discover anything using their platform.

chatGPT is incredible, just giving it a single song and some context, it can recommend at a rate of something like 85-90%.

the only place i’ve gotten the BEST music recommendations were the oink and last.fm forums. humans, still, are the best recommendation infrastructure.

Memories. I wrote the initial Windows Media Player plugin for Audioscrobbler but didn't maintain it.

I filed my collaborative filtering patent in 1995, describing the core basic way that your “desert island 5 favorite albums”, and the 5 favorites list from many other people, could be used to recommend music you would like. The patent is a nice tutorial on how it is done, check it out here if you’d like:

https://patentimages.storage.googleapis.com/9d/f9/19/08ac5ef...

Here is part of the story on my website… I’ll write it up better one of these days:

https://www.whiteis.com/similarities-engine

Yeah yeah it was a software patent. If that bugs you, you can take solace in the fact that I blew it executing on monetizing it. Microsoft ended up owning it and I went on to other adventures.

Here’s a list of the 456 US Patents that cite the Similarities Engine patent as prior art: https://www.whiteis.com/cites-to-se-patent

I guess I'm gonna pop in here and mention libre.fm

The SongMeanings.com discussion boards on each lyrics page are an absolute time capsule.

Google Music killed my used of foobar, scrobbling, soulseek and probably others.

I stopped scrobbling many years ago when they messed together my top artist at the time (the lovely "alan", spelled with all small letters) with other entirely unrelated artists by the same name (but with different letter case, e.g. some "Alan" this, and some "Alan" that.) It didn't represent at all what I was actually listening to, so what was the point?

Used their API to pull tag data as part of a project not too long ago and had to spend a disappointing amount of time filtering out literal hate tags/slurs and widespread deliberate mistagging.

It caused me to not make the code public until I can ship it with an allowlist. It's almost done but I got distracted

ooo... i thought Last.fm was a rebranding of audioscrobbler; i didn't know it was a parallel project. and I am an audioscrobbler user since 2006! and I've used it to this day, i mean, last.fm.

very interesting article!

https://libre.fm/ scrobbling since 2009 built on GNU FM