[…] Many users are complaining of hallucinated artists’ stats, songs they’ve never heard appearing in most listened lists, and more. Some users aren’t happy with the way that Spotify Wrapped looked this year, complaining that it was boring, and not as creative as previous years. […]
One of the most common complaints seems to be that Spotify Wrapped has misreported on the stats in claims to represent. Now, while this has happened in previous years, this year it seems to be a whole lot more prevalent — as you can see in this Spotify support thread. It’s filled with users complaining about top artists they’ve never listened to, songs appearing in at the top of their lists that they didn’t even know existed, or a mixture of the two.
Just anecdotally talking with people outside the internet-o-sphere and you’ll quickly find people with iffy Spotify Wrapped statistics. I have friends who listen exclusively to bizarre, underground punk acts, and their Wrappeds were topped by the likes of The Weeknd and Taylor Swift. Even on the Tom’s Guide team, we’ve seen strangeness — our own Millie Fender discovered Swift among her top artists, yet her songs didn’t appear in her top 100.
Just read people’s reactions online, especially in the trending topic on X or even the announcement thread on Reddit. Many are blaming Spotify’s pivot to AI this year for the lack of personality, and some even hold it responsible for their weird Spotify statistics. I have reached out to Spotify for comment, but I am yet to hear a response.
What can we do?
First off, there are other streaming platforms with year-end roundups. Deezer, for example, has unleashed My Deezer Year, and like everything the French streamer does, it’s filled with creativity and personality. That’s a streaming service with better sound quality too. Apple Music has its Replay feature, which similarly takes you through your year. There are options.
If you’re duty-bound to Spotify, then there is a way to really check your most-listened-to artists, songs, and genres — Track your listening with Last.FM, for example, which gives you breakdowns over the course of the year. Whatever happens, I (and many others) are hoping that Spotify Wrapped is a whole lot better next year.
Why this even calls for generative AI is beyond me. Just post my top 5 artists and top 5 songs. It’s data already available.
Mine said my top artist is Hozier, driving my Goblincore Fantasy Forest phase. I couldn’t name a Hozier song since Take Me To Church, a song that falls into Nickelback territory for me, as in sellouts that get way too much playtime. And no idea how that turned into goblins and fantasy forests.
I think that “take my whiskey meat” song is Hozier. It was popular this year. Maybe you listened to it?
Maybe, it doesn’t ring a bell but I might recognize it if I heard it.
I’ve come to the conclusion my top 5 lists are bunk. I know there is a particular song I listened to at least a half dozen times; neither it nor the artist were on my top lists. Yet somehow Hozier is my top artist? No, it’s broken.
Only the podcast style overview is ai generated. Since the two voice are not from real human
One of my top artists is “Ape Escape”, which is apparently the artist for the first Ape Escape game’s soundtrack. You wanna know what the AI podcast said it’s genre was? Italian Synth pop.
Goblincore Fantasy Forest, definitely a real human genre.
Hey don’t talk poorly of the GobliFantaForestCore community. It’s a thriving scene and just because you’re not sophisticated enough to get it doesn’t mean you can put it down
If you’re a corpo, not integrating generative AI somewhere is seen as a misstep. Shareholders want to know that their money is being used on the cutting edge.
Of course, both groups are clueless as to where generative AI may actually be useful. But ultimately it doesn’t matter.
Not seeing that Hozier criticism tbh but agree spotify sucks
They seem overrepresented in my algorithms. I don’t particularly like them, I don’t click on them, I’ll even spend limited free skips to rush past them sometimes, yet they seem to be everywhere. I can only assume their label is spending an ungodly amount of money to push them in the algorithm, making me shift from neutral about them to negative. Like how Nickelback was on every radio station despite not really being anything special. Hozier seems to sneak through despite my efforts to curate and train my algorithm intentionally.
So Hozier represents what is wrong with enshittifying music streaming apps to me. Probably not entirely fair of me, but that is my explanation for why I am confused and mildly irritated Spotify would call them my top artist.
Listen here, dummy, “AI”. Do you understand?
/s
Also had goblincore in my genre. Whatever that is…
I thinks it’s more a flavor, than a genre.
You can get “goblincore” vibes that span genres. ICP would likely be “goblincore” but so would a lot of Billie Eilish, and some Lupe Fiasco.
Think grungy, anti-establishment type of vibes, regardless of genre. And yes, anti-establishment is subjective, in this case.
IIRC, goblincore is an aesthetic related to the more ugly side of nature - think mushrooms, toads, and snakes. But it’s like a fashion/lifestyle subculture thing like “dark academia,” not a music genre.
I heard that Spotify fired the guy who invented the original Spotify algorithm for everything from wrapped to the discover playlists. I’ll look to see if I can find link and update.
Edit: not sure if is what I read originally https://blog.tunemymusic.com/spotify-wrapped-2024-bad-users-complain/
But I remember the quote
“Everyone noticing that Spotify wrapped is weak this year should remember that Spotify fired the main dude who created the algorithm for them about a year ago (for no reason) and he was also the mastermind between all the genre related data that essentially keeps the place running.
“Enshitification at its finest folks.
Hopefully tidal hires him.
He apparently made both of these sites I found every noise interesting
Well both of those are damn cool, thank you for the links!
No problem glad you enjoyed :)
Please don’t give them ideas. They have to fix core issue first - catalogue quality where albums from same named bands are often attributed to wrong ones.
They fired the whole wrapped team and replaced them with some AI nonsense.
I hope Spotify gets all they deserve from that decision.
Unless Spotify drastically changed their backend it doesn’t make sense why firing one person will make algorithms stop working
They didn’t fire one. They fired 1500. Including most of the Wapped team.
What’s so hard about tracking which song has been played how often or even for how long? I can’t imagine how you can fuck this up so badly
Even if they fire him, his code still exists and is still legally theirs.
Ya but who’s gonna maintain it? You’d be surprised sometimes how one person’s vision is not so easily grasped by another.
mine was on point
congratulations on being in the 0.001% of all listeners worldw
Hurtin’ hearts need some healin’, Take my hand with your glove of love.
Dropped Spotify after the CEO said artists music weren’t worth anything and raised their annual fee at the same time. Trash company
Switched to Napster this year. Best artist payout.
How old is this? Why is play music here?
Easily 5+ years old. This exact image was posted on Reddit three years ago. GPM shutdown in 2020. I can’t find the original page on Digital Music News for this image (source text)
To be clear, no streaming company is paying royalties on a per-stream basis, it’s basically always ((total revenue-platform cut)/share of total amount of streams).
So the artists are not really getting more money because you’re streaming on one platform or the other, they are all getting roughly the same amount out of what you’re paying on a monthly basis.
If artist payout is your primary concern, take a look at qobuz. They pay even more than napster and tidal.
Or actually buy albums lol
Napster still exists?
hallucinated
It is just fucking wrong. AI isn’t a living thing that is comparable to a person or animal, or even a plant. It is wrong a lot and that is because it has no motivation or reality or anything like that. It is pattern matching that is not reliable for factual results, and sometimes the results are just wrong.
Good callout. We shouldn’t attribute intent, experience, or any other human characteristics, to software.
Can’t report the same problem. Mine looked very sane for my listening habits.
Same. Last.fm also agreed with what Spotify told me.
Same with listenbrainz though it couldnt capture earlier parts of the year for me.
I only stick with spotify because it has the best support on linux
Can any linux users can share their experience with other platforms?
Bandcamp? works good in a browser and supports artists the most
Seconded on Bandcamp, though I don’t stream through it. I just buy albums and play them on whatever music player I have installed. Usually VLC or Amarok!
I think Bandcamp is the only platform where I have actually purchased music in the past decade. It seems like a decent model, and I like getting FLAC versions of a physical album I get.
Recently got the newest Godspeed You! Black Emperor album off Bandcamp. Great shit as always.
Tidal has no official Linux app, which is shocking considering their demographic. But a hero has made an app that gets pretty close. Under the hood it’s the web client with some add-ons to support full quality streaming. The user experience is generally fairly close to an official app.
I used soundiiz to convert all my content over, and of probably over 10,000 songs there were less than 100 unavailable, so library isn’t a concern. The increased quality is nice, but the big reason I choose tidal is that instead of doing unnecessary stuff like podcasts they pay artists better. As much as 3X according to some things I’ve read. I have not verified those numbers.
I think Deezer has an electron app or something it uses, but it’s worked great for me so far. It costs the same aa Spotify the last I checked, and has a comprable selection.
Qobuz can be installed with bottles and works pretty well.
You can also get qobuz working through strawberry and it’s pretty straight forward. Just need to grab the app secret with qobuz-dl
YouTube music works great, just gotta use an unofficial app
Do you use the official Spotify client or a third party one like ncspot? Not that the latter will remain for long now considering API changes on Spotify’s end.
Official.
Yeah, if they kill off the official app and neuter the other linux based choices then i’ll move on
I like the Spotify connect feature too. I think there are snaps or flatpaks that wrap the Apple Music web page into an application icon. A desktop app is really mostly relevant if you’re using the app to manage local music files. That said the Spotify connect feature is incredibly useful.
i listened to the new eminem album once and it said he was my #2 artist. i had a day where i listened to nothing but aphex twin and he wasn’t on the list. busted for sure.
Is there a good way to sync listening history between platforms? That seems to be a big reason why people have a hard time jumping around between platforms
Lastfm or listenbrainz
It doesn’t sync between services, but last.fm can be your tracking source and connects to both Spotify (when I had it… assuming still true) and Tidal (ways to track Bandcamp and other streaming too).
I have to wait for the full calendar year to pass in last.fm, but I get more interesting stats than what I remember from wrapped.
Last.fm is really great for tracking, it’s been around for forever too, no idea why it isn’t more popular.
Unfortunately they walled garden themselves pretty universally. I’ve been on Tidal since Spotify fired all those people and have been really enjoying it.
Maybe I’m crazy but I feel like their recommended music is also way better.
You mean the corporation you gave boatloads of personal data to is manipulating it to use against you in the pursuit of profit?! [surprised pikachu.jpg]
My Wrapped was way different this year from last year. I listen to music all day every day. I have a 24-hour playlist for every day of the week, Spotify says I listened to 326,614 minutes this year. These playlists are all different. Monday is 70s R&B Motown, Tuesday is Rock/Punk/Grunge, Wednesday is Rocksteady/Two Tone/3rd wave Ska, Thursday is Classical, Friday is Hip Hop, Saturday is Country, and Sunday is Jazz.
Last year my Wrapped had a wide range of artists and genres that represented the wide range of music I listen to. This year it was just Reggae and Ska. I had only two top artists that were not, the Rolling Stones and Glenn Miller. The Rolling Stones I can see but Glenn Miller is not even that prevalent in my Jazz playlist.
I don’t know if it’s all fucked up but I do know it’s miles different than what it gave me for the same playlists and similar listening time last year.
Edit: I fucked up Monday.
My favorite song of the year was something I apparently listened to twice. My favorite artist is some duo that I spent seven minutes listening to. I definitely listened to something else more than that.
But it did know I listened to a lot of Behind the Bastards. It got the podcasts right.
I was disappointed in the graphics, but my stats seemed right. The slide for my top podcasts had black text on a black background, but luckily all my top 5 podcasts have their name on their logo. Still looks kinda stupid though.
It looked wrong when I first clicked it, but then sort of restarted and corrected itself, almost like it started with the wrong data cached.
