People have been noticing this a lot recently but what nobody seems to know is that this is a form of money laundering/“scamming.” I know because I used to be active on crime forums and talked to some of the people who engineered this scheme.
People will set up fake Spotify artist accounts with stolen identities and bank accounts, pay a musician for songs that pass as music, and then bot millions of streams on them. At this point there are so many of these fake profiles and songs that the music, which is simple “mood music” normally (which happens to be easy to make), is appearing on real playlists and being recommended to real listeners.
That's really interesting - although when I read that story I thought the point was the broken recommendation system.
Happens to me a lot on Youtube too - you watch just one thing - and now your recommendations are full of things like that one thing. Or Facebook/Youtube, somehow: I watched a single video on Youtube about Viking sword fighting and suddenly my people recommendations on Facebook were at least 20% militaria fans, always visible from the profile picture already. It finally stopped but it took well over half a year.
That's why I thought this story was about recommendation systems recommending either only narrowly what you already know, and finding something new is not really well supported or not better than random, or that one outlier can skew your recommendations for a long time. Worst is there is no way to tell the system "stop recommending me this kind of stuff" at least for the second problem, no manual way to make adjustments for the user.
> Happens to me a lot on Youtube too - you watch just one thing - and now your recommendations are full of things like that one thing.
Spotify's recommendation system is much better than youtube's (at least for me). I frequently "joke" that it is one of the few applications of ML that actively make my life as a consumer better.
Having been following Spotify's "Discover Weekly" for several years now, I'm actually really impressed how it manages to blend my long-term taste with recent moods. If I've been listening to one type of music for 1 or 2 weeks, there will be a noticeable uptick of it in the recommendations, while still mixing in less recent tastes.
As a father of a young child, Spotify's algo's are completely useless to me. If their ML is so smart it should be able to determine that Row row row your boat doesn't mix well with Anthrax. Wish I'd be able to toggle 'don't recommend kids music' somewhere.
I used Spotify to play music at a children's party. I am still getting recommendations for "Happy Birthday" months later. According to Spotify, it goes very well with Sheryl Crow and R.E.M.
I would argue for the vast majority of people language is not the primary indicator of what they consider good or bad music. I would assume it doesn't go into the algorithm at all and I certainly would not want it too. I have discovered some really cool music that I'm not even sure what language they are in this way.
Well i think the statistics are very strongly in my favour because English pop music is very popular all over the world and there are plenty of countries where the majority of people don't speak English (and still listen to English music).
When I was young we (in all the kids my age) listened to plenty of of English music even though we didn't understand anything. And let's not even talk about the plenty of examples of songs in English where even native speakers don't understand the majority of the lyrics.
Are you American by chance? Feel free not to answer, but living in the US I was surprised to learn friends in other countries commonly listen to music they can’t understand (most commonly English, sometimes German).
Put things in perspective for me - we really live in a bubble here. I guess Spotify could help reinforce that if that’s what the market wants.
I think it's likely you're right, although I haven't looked for any hard numbers.
But anecdotally, I only speak English and I listen to lots of Finnish, Swedish, German, and Norwegian music.
I don't know if Spotify just started recommending this stuff based on the few metal bands I knew from those places, but it started recommending other genres too and I really like lots of it. Not sure how I'd find this music otherwise, actually.
That's exactly why I enjoy listening to foreign pop music: I cannot hear how vapid the lyrics are.
But besides that, great music does not need to spell it out to convey meaning and emotion.
If the appreciation of a foreign song keeps growing, I will eventually look up a few translations. Their mixed interpretation usually only enhances an already enjoyable experience.
Can be helpful to turn off watch history during such sessions, or to remove the videos from your history afterwards. Agreed that there should be an automated solution, but you can avoid it today will a small amount of work.
Are we talking about the same Spotify? Because the only reference I found to removing songs from my recommendations is a thread from 2020 that says it is not possible.
I set up the browser session we use as a TV with profiles entirely to prevent any more of my spouse's yoga videos from leaking into my carefully curated feed of machine shop footage and construction videos.
Spotify does have an incognito mode to prevent this, but you have to carefully enable it beforehand. Why you can't just delete stuff from your history is baffling.
I agree. There are some playlists and songs I only listen to in the gym, and similar songs constantly float to the top of Spotify's recommendations for me due to gym being 3x week.
I did consider creating another "gym-only" user on our Family plan, but Spotify should really have a way to create named contexts for the one user. e.g. commute, gym, run, working from home, on a plane, etc.
I actually watched someone give a presentation at spotify on exactly this when I worked there. This was probably about 5 years or so ago. I have no idea what happened to that project. Probably the biggest issue would be getting people to give you location permissions but they also talked about working around that. It's a big org though they kill off ideas all the time.
There's a lot I like about Pandora and a lot I don't. The "stations" you mention can be pretty good. The problem I have is with the app itself. It's buggy and learning to use it well can be frustrating. A simple example would be when using the back arrow to return to where I was in the app. If I try to go back to a previous screen, it shrinks the app and when I resize it, Pandora restarts from scratch. This seems to take forever and it happens just about every time I use it. The only reason I haven't deleted it and switched to another app, is that I can't be bothered with all that while I'm working or driving (the only time I use it). Of course, there's other, more common issues like searching for a favourite song and only finding live sessions or worse, finding out they've stopped carrying a certain artist on my playlist.
Seems to be an unsolved problem to train the algorithms for recognizing different situations. One recommendation for all roles the user has. Though, thinking about, it's probably unsolvable as long as the interfaces remain simple and focused on satisfying only the one user, not the different roles, which would complicated the interfaces.
I think this is fairly well solved from a mathematics point of view (high-school level k-means clustering). The unsolved bit is simply how we get the Spotify et al. product managers to care.
All they need to do is filter out content with genre=kids from recs. They could create a separate kids recommendation item if people actually want that.
This wont happen whilst they have family and kids’ account types. Charge parents more for the premium option of not filling their carefully curated genre recommendations with nursery rhymes and Ed Sheeran.
It's fairly trivial to keep the recommendations consistent depending on the latest request, not just the logged user. Pandora does a good job at this. We have it on our Alexa and at dinner, we take turns with the kids requesting songs. If we stop requesting, it keeps playing an internally consistent series based on the latest song we asked for. So if it's the kids' choice, it's a never ending row of Kids Learning Tube :-)
From a technical point yes, but profiles are too rough and hard to use for this. What I mean is more some way to automatically maintain a kind of sub-profile, for each different aspect of a user, but exposed user-friendly and effortless. Most people won't maintain a separate user profile just for recommendations, as it's too much work for too little benefit.
It should be noted that in a Tesla, changing the Driver Profile (stores settings like seat and mirror positions) also changes the logged in Spotify account.
Both Spotify and Apple Music algos have no idea about music mood vs time of the day vs activity. Like late evening before sleep being perhaps not a good moment for heavy metal. And running not a right time for slow classical music.
Which is weird in case of Apple Music because Apple knows exactly if I am sleeping, running or driving a car - just from reading my watch.
If they're working on it then the point still stands that they don't currently have it as part of their product.
Time of day aware recommendations is something YouTube seems to have had for years. It always knows what to give me up top based on if I'm sitting down for dinner or lunch or if I'm looking for an audiobook for bed etc
Or 3) they run loads of A/B experiments to optimize engagement or some target metric and they've reached a local minima and are unable to escape it without a lot of political will or Product Managers willing to stick their necks out.
Spotify has supported this for many years - it's called Spotify Premium Family[0]. You get 6 separate accounts for $16/mth. It is worth the price of admission just to keep my listening history / algo feed clean.
That's a good tip though I'm not sure that there is an option for it in the car when I most often get a request for nursery rhymes, or more recently Disney songs.
I'm trying to avoid this outcome by playing as many different genres other than "kids music" as possible to my baby. Probably won't know whether it worked for a few years but in the mean time I get to listen to real music.
That sounds like a great argument for easy profile switching, along with an option in your play history for "move this play history item to this other profile".
I keep hearing this, but it hasn't happened to me yet. I wonder if it's because my Spotify history without kids is ~10 years vs ~2 years with?
We listen to a lot of Disney music, etc in the living room on my account, but I've never had that type of thing show up in my Discover Weekly or any of the Daily Mix playlists.
I will be very displeased if/when it does happen though.... Discover Weekly is a major reason I've paid Spotify for so long.
Ha, my life exactly. Ever since my kids used my Spotify account to listen to music my discover weekly list has unbearable amounts of paw patrol, baby shark and others mixed in. Now to be fair it still puts in music that fits my taste as well (and often my kids also like that music too), and considering that this is one profile I don't know what it could do better.
A lot of the time it's not an algorithm in charge. Most of the time now recommendations are based on who paid to promote their song or podcast.
Algorithms cannot be left running totally when sponsored ads can be purchased on the fly by content creators and musicians unless a site is lying to ad buyers about ad effectiveness.
The easiest thing for them to do is simply exclude Kids music from their recs. Kids music doesn't benefit much from it anyway. Apple Music does this as do most video services.
Don't these end up in separate mixes, though? My 13 year old makes a lot of grime requests on road trips etc. and now I have a separate grime mix under my top mixes.
It does it relatively OK with the daily mixes, but daily drive and top of the year don't. That's why I don't really get it, they already know fairly well what kids music is so let me just toggle it off?
My brother who has young kids did the thing where you can make a combined playlist with me. He has indeed kids music littered all over in the shared playlist but moreover he also listened to a 100 track audio book on Spotify (which was not a great experience since it was published as an album and did not have bookmark support). The algorithm thought: oh he must REALLY like that album because he keeps on playing tracks from it, and so chapters from this book were also in our combined playlist
Can't really complain if they offer Spotify Family which gives you 6 separate premium accounts, I think it's only double the cost of a normal Spotify account.
So the solution is paying double so I can log out of my own account and into an account I created for my 2 year old so we can listen to a disney song? And when I want to listen to my own songs again then I have to log out of my kids account and back into my own?
Yes, if you want multiple accounts you pay for multiple accounts. It's a solved problem you just have to pay the $9 a month.
I have two children, myself. I just installed second account on an old phone which has no SIM card in it, it has all the kids playlists and I don't have to worry about them getting recommended any of my 90s thrash metal. :)
That unfortunately doesn't solve it. There will always be moments where kids will want to ask for songs on your Spotify, like you're renting a car and have your phone hooked up.
> Spotify's recommendation system is much better than youtube's (at least for me). I frequently "joke" that it is one of the few applications of ML that actively make my life as a consumer better.
I've had the opposite experience with Spotify - I'd say my discovery of new music has withered to almost nil since switching when Google Music shifted to YouTube.
The algorithm just churns stuff I've already listened to, or suggests artists with (consistently) two songs that feel like Spotify's own fake artists - session musicians generated to capture royalties in-house.
I have the same problem with Spotify, just the same stuff over and over and over. YouTube Music has a much more diverse recommendation approach than Spotify does in my experience.
True, but it makes for another good use: building a live playlist. Youtube would cram all random things with the couple of novel interesting clips, while Spotify will keep the mood until I decide to switch.
Spotify's algorithm works really well until you let someone else use the account. I wish there was a "child mode", "party mode" or whatever to disable updating recommendations.
Pandora gives people granular access to the songs that are being used as the basis for recommendations. You would just delete the songs used by guests.
I turned off youtube recommendations years ago so I can't comment on that, but spotify's have pretty much always been terrible in my experience.
Anything suggested on the front page is either songs i already have in my liked songs or completely out of place. And the songs that i've already liked are from a few specific genres and artists, which seem completely random.
A few examples:
* Ratatat and Röyksopp both are in my top 5 of all time according to Spotify's own stats, yet I never got any suggestion about E.VAX, Kunzite, or Röyksopp's releases (their Lost Tapes playlist and their latest album, released about 4 months ago).
* the 'recommended for today' section is regularly filled with random synthwave when it's a genre i barely ever listen to, and a lot of 'electronica/trance/organica/deep house/whatever you want to call it', which i don't really listen to either (at least this kind of electro music).
* still, the worst offender has to be podcasts. I have zero interest in podcasts, have never clicked on any and likely never will, yet it's always the first thing on the app homepage, just below 'recently played'.
I've tried to use the browsing categories but these are most of the time just as poor, they only contain a few playlists and the latest big releases of the genre.
Highly subjective, but i'm tired of personalized suggestions & feeds. Just because I watched or listened to something does not mean I want more of it. And imho 'just use a no history session' or 'click not interested' do not solve the problem, especially since the argument in favor of recommendations seems to be that it is 'more convenient for the user'.
Why can't I just browse instead? Spotify has a very extensive way of categorizing songs based on multiple characteristics (https://developer.spotify.com/documentation/web-api/referenc...), so why can't I just use that when searching for new music, directly in the normal app?
All these recommendation engine problems people aren't discussing here aren't bugs or problems, they're features. They're either pushing the content that makes people engage with the platform the longest, or that promote another feature that the platform is trying to push on you. With YouTube, it's MrBeast style videos, influencer bait, and shorts. With Spotify, it's for sure podcasts.
Podcasts are so heavily pushed by Spotify because they're trying to make themselves the centralized one stop shop for your audio consumption. They didn't pay Joe Rogan millions for no reason.
As a man under thirty, I see the JRE logo on the front page of my Spotify at least five or six times a month. I have no interest in listening to podcasts on Spotify. I very much dislike Joe Rogan. I hit the option to not request it to me again, and yet there it is.
They're not recommendations based on your taste. They're a mix of just enough of your taste so that you trust and buy-in, mixed with whatever flavor of the week the recommendation engine would love to sell you.
> Spotify's recommendation system is much better than youtube's (at least for me). I frequently "joke" that it is one of the few applications of ML that actively make my life as a consumer better.
I hear this all the time but Spotify just plays stuff I've listened to. It's not a discovery service for me, it just plays the hits.
I'm wondering if that's just what it is and everyone likes it because it's playing stuff they already like for twelve tracks and then one new song.
> Happens to me a lot on Youtube too - you watch just one thing - and now your recommendations are full of things like that one thing.
I've heard reports of this happening but for well over a year, YouTube was still recommending relatively high quality videos to me that went along with my interests.
Until last week.
Somehow, somewhere along the line I must have clicked on some sort of "influencer" video. Now all I see are hundreds of hyper Minecraft videos, "I filled my house with 1 million packing peanuts" videos, etc. I have tried manually searching for some of the topics that interest me again, but these influencer-type videos still overrule whatever I try to manually teach it.
I don't know what happened, but it has ruined it for me.
You can review your history on YouTube and remove videos (or at least tell it to not use them for recommendations). You may find one obvious culprit a few weeks ago.
I have the same problem. I don’t know what I did to tell YouTube I’m interested in Minecraft (never tried it, never watched someone try it).
I can vaguely see why YouTube is pushing right-wing conspiracy videos on me, but Minecraft? Mr. Beast? I’ve blocked perhaps two or three dozen of these channels. Take the hint, YouTube.
Something I noticed after I watched an episode of Stoic Finance[1] about a possible "impending crash of the Chinese economy" is that YouTube started recommending about a dozen "independent" channels.
I put the word independent in air quotes because they all have near-identical looking title cards with the similar fonts, some variant of the phrasing "The collapse has begun!", and the same-ish content. Different presenters, different channels, same message. Over and over. And over.
Reminds me of the "This is Extremely Dangerous to Our Democracy" clip that edited together dozens of apparently independent local news channels saying the same script, verbatim.[2]
You've got to wonder if there's someone with deep pockets trying to influence public opinion on China? Or is it just the same thing as the "local" news channels, where a bunch of YouTubers have been "bought" by a corporation that send them material to read?
I would love to know if anyone on Hacker News knows something about this kind of thing...
The average Chinese person needs to save 100% of after-tax income for >30 years to have the down-payment for the average Chinese house. And, then, the interest payment alone would still be higher than the person's after-tax income.
Either China has to indefinitely grow at 10%+ per year (it's becoming obvious that's not going to happen), or the Yuan or Chinese home prices (~50% of Chinese wealth) are massively overvalued.
>You've got to wonder if there's someone with deep pockets trying to influence public opinion on China? Or is it just the same thing as the "local" news channels, where a bunch of YouTubers have been "bought" by a corporation that send them material to read?
These are the least likely options.
The most likely case is that the channels are all copying each other. Even if they were all put out by a couple content farms, it's extremely unlikely that the content farms care about anything else but making money. The only thing they likely care about is that people watch videos like those.
People trying to make easy money is, by far, more common on YouTube.
The Youtube algorithm bothers me a lot because they have the subscribe system but as soon as you do subscribe to a channel that channel basically disappears off the recommendations/front page.
This is also the only way I can make sense of twitter, skip the 'feed' or whatever it's called[0] and go right to my curated lists.
[0] as an aside, I only just made the connection that 'feed' is likely an abbreviation of 'newsfeed' but also brings a strong and stark connotation that I'm livestock to be fed
Funny, maybe it's a language thing (French speaking) but to me the first thing that comes to mind when speaking about feeds is "RSS", so I genuinely never made the connection.
Just opened youtube.com and 6/8 of the videos on the top section of my front page are from channels I'm subscribed to. That happens for me every time. Sorry your subscribe button doesn't work.
Spotify's recommendation system is so bad, that when I find a song I like, I put it into Pandora's free tier to find more songs. I hate how Spotify likes to prioritize every other song from that artist, and every other song from that genre, over songs from other artists and genres that sound similar to the song I want recommendations for. Pandora seems to prioritize the sound of the song, and actually finds similar songs.
Sites like Pandora, and many other "underdog" sites, operate as expected because they are trying desperately to please users in a bid to capture market share. Once they gain market dominance they begin to operate in an unexpected/unfavorable manner, just like Spotify does now, because "tailored" recommendations (strategic content recommendations) generates the most annual revenue for them.
The saying "power corrupts" applies to this circumstance perfectly.
Spotify manipulates recommendations to what makes them the most money, not necessarily what is best for user satisfaction. They do it just enough to prevent users from cancelling their subscriptions as well (usually).
> Spotify manipulates recommendations to what makes them the most money, not necessarily what is best for user satisfaction.
I was once a heavy user of Pandora, I am now a heavy user of Spotify, and I'm considering switching back. I think you're right.
But, having experienced both services at the peak of their quality, I think "stronglikedan" is correct, that Pandora has a greater ability (in addition to a greater desire) to analyze the way that music sounds.
The YouTube recommendation engine seems to overweigh tangent content. For example. If your recommendations are all puppy videos (awww) and you go watch Jonathan Blow talk about why programmers aren't productive anymore, you'll get a whole new feed of "How to learn Rust in 100 seconds" crap.
>Happens to me a lot on Youtube too - you watch just one thing - and now your recommendations are full of things like that one thing. Or Facebook/Youtube, somehow: I watched a single video on Youtube about Viking sword fighting and suddenly my people recommendations on Facebook were at least 20% militaria fans, always visible from the profile picture already. It finally stopped but it took well over half a year.
>That's why I thought this story was about recommendation systems recommending either only narrowly what you already know, and finding something new is not really well supported or not better than random, or that one outlier can skew your recommendations for a long time.
This kind of problem is exactly why we are building alternative YouTube recommendations. Search a channel to find more channels making similar content:
Is it money laundering in that "dirty" money pays for the bots that rack up the streams, which generates "clean" money paid from Spotify to the fake artist?
ok so the problem for the money launderer is that as spotify gets better at detecting bots and spam accounts they will have to pay increasing amounts of dirty money to get out clean money, at what point does it become a losing proposition to them, that is to say when this form of money laundering is less efficient than older forms?
A bigger problem would be that Spotify pays very little. And to have the bot "listen" to a song (to be eligible for payment) takes minutes of streaming, which turns to tons of time and a big bandwidth bill, since they'll need to do thousands of them in parallel...
This is where malware run bots come into picture ... it solves a lot of problem like not having to pay for bandwidth, evading detection by distributing access to the songs among varied IP and geography, etc.
It's not about efficiency, it's about avoiding getting caught for criminal acts. Since criminals don't pay taxes on their earnings and don't have to pay for all sorts of things others do, they're already more "efficient"
Sure, but if you're going through a distribution service that has a 60/40 split with Spotify, are you going to be okay with 40% of your laundered proceeds getting eaten by Spotify? How does that compare to other laundering opportunities? And that's not even counting the costs imposed by your distributer. For example, if you have 1,000 tracks, on TuneCore that's $10,000 dollars just for track distribution, as well as a $50 per-album yearly fee. It does seem somewhat inefficient, and very easy for Spotify to notice and crack down on. (But they don't have a lot of incentive to look too closely, since they're getting paid handsomely for it...)
Interestingly, searching for Danni Richardson on Tidal yields the following apparently real artists:
Deanie Richardson
Danny Richardson
Dani Richardson
So, they are typo-squatting as part of the scheme.
I asssume money laundering would work by setting up Spotify accounts with stolen money (that they used to buy gift cards?), then streaming the fake songs.
Does spotify pay artists differently for paid vs free accounts? If not, then presumably they use free accounts for the bots, so it isn't really money laundering.
I guess if you get some streams from real people, then using paid bot accounts might still be a reasonably cash efficient form of money laundering.
Don't assume they're just targeting Spotify. Spotify is (or used to be) the best at catching algorithmic spam, the other streaming services hardly care at all.
Likely they use a front service such as TuneCore which gives them access to all the streaming services. Just using Spotify is unnecessary risky if the goal is simply money laundering.
(However, if the goal is to do pick up some royalties from all those background plays, and I think that's more common, then Spotify is clearly better)
For an example of spam/fraud, look up the album "Angry man" on Deezer. Or rather, the albums. There are 200+ of them, all with the same stock image album art. That spammer is easily recognizable, he's been doing a similar bulk upload several times per month for maybe 10 years now. He's an example of a spammer who is present on very many streaming services. Spotify has kicked him off, though.
Huh... this is pretty interesting. My screenshot shows "15 results" at the top, which a reload of the page just now is still showing.
But when I go to load the site through my VPS in Germany... I see the "292 albums" you're talking about! And yep, umpteen copies of the same album art just like in your screenshot.
Wow. I think this might be due to Australian consumer protection laws dictating that faulty products MUST be returnable. (Buying stuff in Australia is actually kind of cool because if I don't get exactly what I paid for I don't generally have to make very much effort to get a refund.) IIUC this policy has taken impressive chunks out of pretty big players.
...And so it would seem I'm simply not seeing anything I might reasonably and correctly regard as faulty.
Well that's an interesting database column then.
(I unfortunately have less familiarity than I probably should about this status quo (given I live here :D), but I can at least offer the "Consumers" and "Business" headings (and all the individual sections in the associated dropdowns) at the top of https://www.accc.gov.au/ as a cursory indication of the level of pedantry and protection involved.)
It's entirely possible I have the wrong end of the stick here and something else is the axis point. I wonder what would happen if you loaded the site through an AU proxy/VPN. I'm also curious if any other countries cause this... different presentation.
This is just conjecture, but I think it may have to do with compilation rights. The spam albums contain a random selection of music from the actual artists. Maybe the spammer has some very broad license to make compilations - I can totally see record companies accepting this sort of spam as long as they get their cut. But such rights would perhaps be geographically limited?
There are also problems with this theory, though:
* "Their cut" should, from my understanding of copyright law, be 100%. I don't understand how the spammer would make money off it at all.
* From the metadata, the spammer's one-off label claims to directly own both the music and the recording rights.
* The spammed artists do not seem limited to one of the big three's back catalog.
* The spammed artists include some which are/were well known for not allowing third party compilation albums with their music.
The criminal gångs in Sweden are using Spotify to launder drug money buy buying streams on artists in their circle who they also to some extent control. One of those were just murdered last year.
How do the scammers get the money if they use someone else's (stolen) identity and bank account?
Also, how do they succeed in keeping this scheme for more than a few days after the identity theft is noticed and reported to the police?
And why would they not just take the very music they just bought and paid for, and publish that as any other artist do? There's no crime in buying the rights from another artist -- that's exactly what all music labels do.
My take was that it's about getting _some_ of the dirty money out. So the dirty cash they are spending to fund this campaign is returning some % worth of Spotify commission. Even better if it somehow hooks the algorithm to recommend it to non-botted accounts.
> How do the scammers get the money if they use someone else's (stolen) identity and bank account?
Because if you have a stolen identity and open a bank account, as long as that account isn't trying to defraud the bank (e.g. write bad checks, commit ACH fraud), it can go on living for a long time before it's discovered and closed.
In this case, they opened up a bank account with stolen identity, and then used that to get their Spotify payouts, and then eventually transferred that to another account (cash, crypto, whatever). From that bank's perspective, unless they were specifically informed of the bad account, there wouldn't be a ton of unusual activity that would flag the account.
> Also, how do they succeed in keeping this scheme for more than a few days after the identity theft is noticed and reported to the police?
Why would a stolen identity not being used to steal from the person whose identity was stolen, but just to process legal payments, be noticed in a few days? Or even a few years? Maybe at some point the IRS might find the account that used my identity and send me an angry letter, but that seems like it would be years down the line.
I think I came across a YouTube video a while back that said even the platform was sponsoring fake artist accounts to recoup/claim streaming royalty payments.
After years of grinding to break ground on spotify and seeing the names they promote taking the lion share of revenue it becomes very apparent that it's a cash grab that just can't be trusted. There are literally dozens of other sites that do the exact same schemes and only pay out a fraction of the revenue they rake in for doing little to nothing for artist and music discovery.
The social Internet just seems more scammy than ever, and that's why I keep most of my music work offline until it gets worked into a project. Eventually there will be a correction hopefully, but the vast amount of free/unrewarded work that is being hustled out of people from these sites is not sustainable, and eventually it will turn into a content creator drought/strike if it goes unchecked too long.
That almost makes sense, except that the money is paid out to the artist, so if you're using a stolen bank account, it remains 'dirty' after the round-trip.
What you'd want to do is use the dirty money for the bots that drive up the revenue, and the clean account for the artist to collect the royalties. This way the artist has plausible deniability, and if the bank account paying for the bots gets busted, it doesn't matter, you can get a new one.
This is right - the whole point with money laundering is to have a legitimate business where extra money can come in, apparently from real but untraceable customers, that is really dirty money you control.
I had my Spotify password compromised a while back and random fake generated music would play sometimes from my stereo. Very strange experience until I figured out what happened.
How is this money laundering? If you use dirty money to sign up to Spotify and listen to specific songs so that you get some of the money “clean” back wouldn’t that in case of Spotify be the dumbest and most inefficient way of laundering in the history of money laundering? I mean doesn’t Spotify keep the majority so you’re essentially donating dirty money into the stock price of Spotify for the most part, no?
Not necessarily. Let's say you get a burner phone and hook it up at home. You need internet and electricity of course.
Buy a Spotify gift card for $10 USD cash, use that to get Spotify Premium (first use the three months free of course).
Play your (artists') songs, 30 seconds each (considered one stream by Spotify). That's 2880 streams per hour, 86400 streams per month.
Spotify pays around $2-$4 USD per 1000 streams, so that's ~170-340 USD clean money coming in. Some of that goes to Amuse/TuneCore/etc. Then music production, admin, etc. But there seems to be quite some money to be made, especially once you manage to fake your way into recommendations and get legit streams as well.
Sorry, but wow you got this wrong. I’m not sure if it was just a guess or something, but look at the comment below for what it actually is. The money laundering is in the form of purchasing botted streams.
As another comment [0] suggests, this is common at least in Sweden. I have it on first hand that a successful local rapper has the whole apartment full of phones streaming his songs. Floor, shelves, sofa, all full of phones.
I'm sure mixing it with bot traffic is also popular. I also assume that Spotify (et al) are more successful in filtering out bot traffic from central Asia than legit phones in the middle of the target group.
Who launders pocket money? Anything less than hundreds of thousands of dollars is just not worth laundering. Nobody is going to ask you where you got a few 10k from.
Spotify supposedly pays around 70% of subscription and advertising revenue to labels and rights holders. Since payments are not pro rata per listener but are aggregated over all listens, if your fake customer streams more than the average you could quite possibly make an actual profit from this technique.
I wonder how much of the traditional music and theatre businesses has been money laundering. One could put out a record (which presumably has a nice margin on it) and pay folks a small amount of money to buy many many many copies of the records with cash, turning the dirty funds into clean ones. Likewise with a theatre production and ticket sales. Seems like any business with significant cash inputs would be vulnerable to this.
And of course it would make it very difficult for legitimate businesses to survive: they would lack the implicit subsidy that the illegitimate businesses get.
By theater, what type of theater are you meaning? Stage performances like plays? If you put on a crap play, the critics will lambast it for the sham it is, and then no more ticket sales.
That...Makes a lot of sense, now that I think about it. I guess it wasn't really apparent to me before it came to my attention like this, but making online content and artificially boosting it, so you can blend your funds within what you've obtained from ad revenue is so simple, yet so brilliant.
Makes you wonder what other media out there might be laundering schemes, too...
There is a shitton of crap movies in Hollywood. I wonder...
Of course Hollywood accounting is already "This movie took $100 million at the box office, we spent $98 million on 'marketing', using our own internal marketing company.".
I guess making crap Hollywood movies loses you money rather than laundering it...
> There is a shitton of crap movies in Hollywood. I wonder...
The movie doesn't need to be crap to be a product of money laundering! The Wolf of Wall Street[1] was infamously funded by money laundering (from the Malaysian Sovereign Wealth Fund),
And now as far as we can tell there's a cottage industry built around algorithmically-appealing (and probably algorithmically-generated) mood music. What a weird outcome.
There is a similar problem with the YouTube recsys on cheaply generated videos (e.g. a physics + 3d rendering model to generate colorful videos of Spiderman and Batman driving colliding cars).
See this older piece by Natasha Lomas on Tech Crunch:
I watched 1,000 hours of YouTube Kids’ content and this is what happened…
http://tcrn.ch/2iPXpIA
Somewhat similar, I'm not really a tik-tok-fan, but after watching a LinusTechTips video with "KallMeKris" I was hooked on her content (one woman plays like 20 characters, yea yea judge me later :P ).
Anywhoo I was surprised to find close to 20 Youtube accounts that only has like stitched-together 1 hour videos of all her content. With non a none negligible view count.
This thread has quietened down now, so I'd like to ask about those forums you were talking about. I'm really curious about them and would love to have a closer look, but I completely understand if you don't want to divulge that kind of information.
I don’t doubt that people are using Spotify to launder money, but I really don’t think this is particular method is a viable approach. It would be a much better business model if you could buy off iTunes or something, where are you know that most of the money is actually going to the recipient.
You have to wonder about the economics of this scheme. The music probably isn't expensive to produce and clone, but Spotify pays so little per stream, how is it viable to purchase bot streams in large quantities?
I worked in music distribution and maintain a very active interest in this space. This is actually fairly unlikely to work, from three different standpoints: operational logistics of how money moves around; technical logistics; and economics.
I’m not saying that it never works - and I’m not saying that this was not happening in the past (in fact, it almost certainly was) but the chances of it happening today are really pretty low.
Here’s why - and I’ll also talk about what is more likely to be the reason for these tracks.
1. Operational Logistics
Although streaming activity is reported in real time Spotify accounts usage to labels and distributors in the middle of the month following the month in question. The label/distributor then invoices and receives payment. Depending on the deal the label or distributor has with Spotify they will be paid on a defined cycle. This may well be NET 30 EOM - meaning that when the distributor raises the invoice it will be paid 30 days after the end of the month in which it was raised.
2. Technical logistics
Both Spotify and distributors identify and flag unusual streaming activity. If you bot 1 million fraudulent streams this activity may well be removed in between the streams happening and payment being made. If a distributor sees sudden spikes in activity they may well hold payment until they can identify whether or not the streams are genuine.
Streaming fraud is a hot topic - and it’s obviously fairly easy to identify “non-human” activity when you have end to end control of the streaming platform.
Spotify has historically significantly reduced play count on tracks they think are botted: today I’d be surprised if many botted streams are even counted against tracks.
3. Economics
Essentially this is an arbitrage situation - can you get enough streams to drive more revenue than the amount you are paying for the fake streams?
The answer is almost certainly no.
Let’s say you’ve got 1000 accounts that are going to be your bot farm. There’s a lot of work is going to have to go into getting those accounts to stream in a human-like way.
Because Spotify does not have a user centric royalty model (where the subscription payments of a user go proportionately to the artists they stream) there is no fixed “amount per stream” - only broad indicators.
Streams from premium accounts pay a far higher rate than streams from ad supported accounts.
Roughly speaking a million premium streams is going to bring you somewhere in the region of $4000-5000 dollars.
If your bot farm is 1000 premium accounts then that’s costing you $10,000 a month. You’re going to have to have a certain amount of hardware - or hardware emulation - as if the streams on a certain track are all coming from web, and from the same network, that’s much more likely to flag than streams distributed across web, Spotify desktop app and Spotify mobile.
Now, sure, you can pay someone who already has the bot farm - but there are then three risks: first of all, that they take your money and run; second, that their bot farm gets flagged and the streams don’t pay through; third that Spotify doesn’t pay out on the streams.
So assuming that you’ve set up your own bot farm investing in 500 cheap android phones and doing the rest with emulation there’s a lot of cost - at least $10,000 a month.
Can you drive 5m streams a month? Well, each node in your bot farm would need to stream the track 5000 times a month. Is that going to look strange? For sure.
There’s also the question of finding distributors who are going to pay out easily and quickly if you can even convert the fraudulent streams to cash.
What is actually happening here? Well, mood music - ambient music, ambient electronic - is a really important part of the “make music happen” sell of Spotify. A lot of people don’t want to hear the hits. They want music to play while they work or run or cook or have a shower. And for these kinds of “activity driven” or “event driven” music it’s important that the flow of a curated playlist works.
You might have ten great tracks by known artists, but the transition between those tracks doesn’t work. Spotify has (according to media) begun commissioning songwriters to write these “filler” or “transitional” tracks. There was a huge tabloid storm whipped up about this by a music journalist some time back who said he’d unveiled a whole stable of “fake artists”. Spotify wants people to keep listening to music. I suspect - but have no direct insight into this - that people who listen to a lot of ambient playlists are particularly valuable subscribers - they will almost certainly be paying (as you don’t want your flow interrupted by ads) and they will almost certainly be over indexing against usage. They are probably also less likely to churn because of the lock in of saved playlists etc. It’s also quite likely that a lot of these custom transitional tracks are not available on other platforms so you may not even be able to replicate a saved playlist.
So the interesting bit is that there are hundreds of thousands or millions of Bot accounts streaming music on Spotify? Wouldn't this have massive implications for their Ad Revenue model?
The scam part here is that listen traffic is bot-generated? Irrespective of that, on the payout side, why do you need stolen identities? It is Spotify paying out legitimately, isn't it?
See also Twitch streamers, Amazon listings with weird pricing, etc. Give someone poor dirty cash, let them set up accounts and buy/subscribe. You lose a percentage but you get clean money.
It costs money to run the streams but Spotify pays some of that back out to the "artists"; that money sounds plausibly pretty clean. Until there was a thread at the top of HN about it and everyone suddenly knew how it worked.
What does it cost an artist to run a stream? I thought Spotify hosted the content. Do you mean it costs money to run the bots that listen to the stream and get the artist recommended and heard by more people?
There are services that offer bots that increases your listening count on Spotify. People with dirty money create a bunch of artists and songs on Spotify, and use the services to increase the listening count. Spotify pay out royalties based on those numbers. So they pay the service with dirty money, and get clean money from Spotify
You need some piece of hardware for the bot to do the streaming on so it sneaks past the spotify bot detectors, and you probably also want to pay for a premium account for each bot so that you get a bigger payout on the other side, since premium users generate bigger royalties.
Making it expensive to operate an effective bot farm is part of the way that spotify tries to discourage botting. Spotify's bot detectors are good enough that it's not lucrative to just make a bunch of bots to give yourself streams for the free royalty money, but apparently not yet quite so good that it's not feasible to use bots to turn a large amount of dirty money into a lesser but still substantial amount of clean royalty money by way of funding a bot farm.
Getting out, say, half of what you put in is a losing prospect for somebody who wants a money printer, but might be acceptable to a money launderer.
You need some kind of publisher/label to get on Spotify; an individual can’t do it themselves (unless that’s changed recently). There are labels that do this for a reasonable amount of money, Like $30-50/year or so.
(In my opinion)
Some hacker provides a paid service in which you have either hacked spotify accounts or well generated fakes that can run streams on fake artists accounts.
Some random mafia pays these hackers to maintain their bots and to launder their money.
You can even pay the hackers with the finally laundered money.
Now the hacker has to launder the money somehow. Fine, it's much less money, but still it doesn't feel like laundering rather than just straight up scamming.
In the “Release Radar” generated for me last week, there was a 40 second song consisting of rhythmic mechanical noises with a strange title and album art of an alien head. I didn’t like it at all. I think it was recommended to me because the publisher listed its artist as “Fanny Mendelssohn” — a classical composer whose work I definitely enjoy and have probably pressed “Like” a few times on — even though the song had nothing to do with her.
You’d think, she shouldn’t be releasing any new works, you know, on account of being long-deceased and all. Therefore Spotify could block releases claiming her as an artist. Unfortunately, there are new renditions of her work being performed, recorded and published regularly, by many different groups. All of them have the right (perhaps even an ethical duty) to put Fanny Mendelssohn as one of the artists. These are works I’d like to hear, so Spotify’s recommendation algorithms are on the right track.
How on earth can Spotify distinguish real Fanny from fake Fanny?
Wasn’t the whole point of Brexit to have less regulation on things like region of origin? British Fanny is just as good as continental Fanny, even if it isn’t as popular. It’s not “fake Fanny”, it just gets less exposure.
To be fair, Spotify often conflates lesser-known artists that have the same name, particularly if they both inhabit some kind of "alt" genre, even if the two genres are totally different alt genres. Are they two different artists, or did they just change their sound?
Sometimes it's obvious (e.g. techno DJ and death metal band with same name), but sometimes it's not even easy for a human to decide (two unrelated indie bands from the early '10s with the same name).
So it's not too much of a stretch to imagine that it can be exploited for money laundering and/or algorithm revenue farming.
Probably serious enough, since the album they describe exists. Spotify is smart enough to know that it isn't "the" Fanny Mendelssohn though - at least in the artist database. It could well be that Release Radar is dumber and is fooled by the matching name.
Release Radar by its nature will try to recommend tracks with very few plays (if any). It is supposed to solve the "bootstrap" problem in recommendation systems.
In retrospect, I think it's more likely that Fake Fanny Mendlelssohn did NOT inform Spotify that they were a different person, and Spotify only split it out after a lot of Felix/Fanny fans clicked "I did NOT like this" when the track was put in their Release Radar.
The Release Radar playlist is supposed to contain new releases from artists you already listen to. I think you might be right about the release radar being a method they use to separate out bands with the same name. I've definitely noticed bands with the same name in my Release Radar. It would be useful if Spotify exposed a method to report different bands under the same name. The do not recommend is a different signal since maybe I also like the new band. I probably mess up Spotify's recommendation algorithm for me since I use the heart button to flag albums as listened to regardless of my enjoyment since there isn't a another way to mark albums as "listened" like there is for podcasts.
Fanny Mendelssohn was personally involved in roughly 0% of the recordings that come up on Spotify when you search for her name, but I'd bet any fan of hers would be OK being recommended most of those recordings, or most any new ones that came out using her as a search term.
Classical music categorization (where the composer of the music is often of more interest than the performer) is very difficult to reconcile with pop music (where the opposite is more likely to be true).
This is incorrect, but it's principally incorrect because we all abuse the artist field to contain composer information. This is done, I surmise, because figuring out when to display the composer information (performances of composed works) and when not to (bands) is too hard while delivering consistent UX.
And on the file level, it is! But then do you as a music player / streaming service display that all the time, when most people are not listening to classical music and do not care about the composer? Do you make the user manage which columns are visible? (Almost every HN user misses when more software did this. Almost every normal user found it confusing/annoying.) What about publishers who do not give a damn that Johann Sebastian Bach did not perform this digital recording because they know it'll SEO way better to have him in the artist field? Now are you displaying all that inconsistency?
I don't mean "add composer to the metadata", I mean add a field "primary artist key" which could have the value "composer", "performer", etc.
How many fields are displayed is up to the UI, but having metadata to indicate what the most sensible field is would allow a simple UI like Spotify to just show that by default.
Specifying on an individual level avoids trying to lump together genres and categorize which field matters most for each.
Isn't it principally an industry thing to hide the composer, they don't want us to know that the pop is from a "factory" and picked for the artist, they want the public to believe the myth that the manufactured band sit on mountaintops with their instruments coming up with new grooves (or whatever).
Can't you just let the user decide "artist + track" or "artist + composer(s) + track" or ...
I was trying to express the idea that it was pop industry who made the decision that composers didn't matter rather than pop consumers and maybe that given the information pop listeners might well be happy to know where the pop comes from.
They should (and probably do, behind the scenes) have separate data fields for composer and performer. At least, I believe that labels will input those separately for each track, to enable correct tracking of compensation.
Then Spotify will lump them together into “artist” n the UI, I assume. To make it easier for a casual user to find what they want.
The “performer” Fanny Mendelssohn should probably not be confused with the Composer of the same name, and they should ensure this by using a unique id for each person rather than just a string field.
I like how you think labels correctly label anything. I happen to work at a music streaming service (not Spotify) and I can definitely say that those datasets are insanely noisy and generally you can't really trust labels to, well, label their data correctly for anything.
I was about to say something similar. I don't even work at a streaming service, but I know label-provided metadata is enough of a steaming pile that there are third party start-ups which promise to help with the problem and such.
I don't think it's nefarious, just incompetence. I like Poppy, the bubble gum pop youtube girl turned metal, turned grunge artist. Her name is pretty generic though, and I've seen multiple artists with the same name have their stuff put under her identity on spotify briefly. The artists I've seen are clearly not fake artists trying to sneak into her brand.
Composer / performer is a whole nother metadata thing that should happen.
What if there was a new artist with the same name? Not too crazy. There was another kid in our high school with the exact same first, middle and last name as my brother. No relation.
It happens; what last.fm would do is add the country of origin to the band, like Shining and Shining (nor) to differentiate the Swedish depressive black metal band and the Norwegian extreme avant-garde jazz metal band.
The post was interesting, but I also noticed something in the blog that had the same flavor.
Even though the link was for a specific post, the page layout neatly flowed into another post. So smooth I thought the second post's title was a subheading of the original article. A strange thing, but maybe that's just me making connections that aren't there.
I listened to the playlist and it's definitely the same song over and over again with slight tweaks. Imagine finding dozens of blogs where posts are all the same, but with very slight changes to vocabulary used in each blog.
There has actually been a lot of that showing up in my google search results for the past few years… there are thousands of sites out there that were created by scraping news articles and blog posts and replacing random words with thesaurus alternatives.
The results vary from incomprehensible to some kind of uncanny valley “is this just…really terrible writing or am I having a brain problem?” territory.
I’ve actually encountered lively discussions of these sorts of cloned articles with nary a mention of the weird writing.
With what we’ve seen from AI text and image generators recently, it’s only going to get weirder soon.
> The results vary from incomprehensible to some kind of uncanny valley “is this just…really terrible writing or am I having a brain problem?” territory.
I’m glad it’s not just me! But for me, it’s not just random searches. It also happens with news articles from major outlets. Can they really not afford to pay a single person to just proof read the story and make sure whatever’s auto-generated is at least coherent?
I listened to the playlist too and it's definitely the same song with small variations. Imagine finding dozens of threads where comments are all the same, but with very slight changes of words used in each comment.
I listened to that playlist as well and it is nearly the same song with slight variations. Imagine finding dozens of threads where comments are all the same, but with very small changes of phrasing used in each comment.
I listened to that comment thread as well and it is very similar to the playlist, only with slight variations. Imagine finding hundreds of threads where the colors are all the same, but with very small changes in the pattern as they are all woven into a blanket.
I listened to the users also and it's nearly the same comment with minor variations. Imagine finding dozens of threads where the jokes are all the same, but with very unsubstantive changes in phrasing as the site is turning into Reddit.
YOU ARE IN A MAZE OF TWISTY LITTLE PASSAGES, ALL DIFFERENT.
YOU ARE IN A LITTLE MAZE OF TWISTING PASSAGES, ALL DIFFERENT.
i remember getting lost in that maze, and didn't realise that the subtle change in wording for the room description was the trick to identifying them, so i dropped objects to help me make a map - which is what you are supposed to do in the other maze
YOU ARE IN A MAZE OF TWISTY LITTLE PASSAGES, ALL ALIKE.
I listened to what gave it away and it's definitely the same obvious random word shuffling. Imagine finding users who all have been listened to, but only partially and in the same way, perhaps with slight changes.
I listened to what it gave away and it's definitely the same random obvious word shuffling. Imagine users finding who have all been listened to, but only partially in and the same way, perhaps slightly changed.
I listened to mgdlbp also and it's nearly the same comment with minor variations. Imagine finding dozens of posts where the rants are all the same, but with very unsubstantive reasons in phrasing, as the old userbase complains about the new userbase ad nauseam.
That was the web circa 2010, when people were still using spinbots and link wheels as an SEO technique. They were the same articles, but with occasional words swapped out so it wasn't duplicate content in the eyes of Google. The Panda update for Google was largely responsible for cutting down on that behavior.
What's fascinating about natural language models for me, is that I assume it is already being applied to "spin" articles, and I imagine that is part of why Google kinda sucks to use right now.
The "horn instrument" being reused in every song is a giveaway that they're using the same generator, without even changing the parameters of the instruments.
It's like they're basically using a fancy "arpeggiator" to generate these songs.
I recall reading an article on HN a while ago about how recipe/cooking blogs are basically becoming just that—centrally run networks of independently branded sites that just serve to drive clicks on affiliate links. But then again, I can't find that link, so it's probably a conspiracy theory I just made up
With the success of AI content generation for text (GPT-3 etc.) and images (DALL-E etc.), it seems inevitable that music will soon be targeted as well, if not already.
What's particularly interesting in the music sphere is that there are already well-established trends towards building a sort of ambient, atmospheric, generated soundscape. (For example, the famous "Lo-Fi girl" stream.) AI-generated content is a very natural progression here.
Regarding the broader pop music industry, a "GPT-3 for music" would likely further inequalize the relative power of labels and musicians. If people who control music distribution can easily make hit songs without needing to hire songwriters, arrangers, or performers, they surely will do so. I can imagine a lot of music-related occupations potentially having to pivot to rely much more heavily on live performances to make any money.
They always have relied on live performance (and to a greater extent, education). Talk to some professionals in the industry some time, they're not going to talk about records as a source of income.
I could write a ten page rant on why AI and automation aren't a threat to musicians but it basically boils down to the fact that music is a human spectacle and we will continue to grow the industry through performing live music much like a robot that can mimic Tom Brady isn't a threat to teams selling tickets to see Tom Brady play live and people to watch it on TV (which exists, by the way).
At the end of the day, technology that lowers the barrier to entry for records and distribution is a massive boon to the industry at all levels. The rising tide lifts all boats. The companies doing interesting things with AI in music aren't just generating old shit, they're making tools to give to the next generation of creators to create new shit that no training can replicate, because it has never been done.
I don't think it does, if you are alluding to electronic music that's also just as much about the human aspect of it. Lets be real, at the commercial end of the spectrum electronic music is trivial to make, but no one wants to hear "Song seed 2aslk3j25lh" they want to hear what their "Music Hero" has made. Another aspect is DJing, DJs don't even pretend to have made the music yet people will come out just to hear a specific DJ play other peoples songs. It's almost entirely about the human figurehead, the popularity contest, the status and the fashion of it all.
Some labels might try to present artists that have AI music, and no doubt it would still be consumed. But it's a huge risk when it to the human aspect of it as people want authenticity.
Music is consumed in a huge array of contexts though, where I think AI music will end up is as music in movies, backing tracks in adverts and youtube videos, as filler music for all kinds of other media.
Let's be realistic here. Hatsune Miku was created before 2010. While I dislike the genre, they've completely sold out all tickets to these festivals where "she" "performed".
You're really not thinking it through if you think any genre isn't going to fundamentally change once the industry starts to push virtual artists. They can be perfect and relatable to teenagers. You really don't need physical people to pull off a good festival, a well orchestrated 3d avatar is likely even better because they can be bigger and seen from the back
Hatsune Miku kind of helps my argument here, because they still needed a personified identity for the music even though none of it's real. So then we ask, if Hatsune Miku could be manifested into the real world, do you not think their fans would be absolutely ecstatic? Behold, humans, the solution to that problem.
I know it's possible to get some people excited about an avatar, but I'm going to argue that the vast majority of people would prefer humans for as long as they can get them.
You might want to check out a Hatsune Miku concert on YouTube because humans are actually entirely redundant if you can just generate the song. (You'll probably want to mute the audio though, otherwise your ears might start to bleed)
There is a 3d avatar dancing on the floor. The holographic technology in the context of concerts is incredible at this point.
As a matter of fact, its likely going to be in favour of AI if you consider VR headset etc, as the coming generations will be able to interact with the virtual artists, giving the producer am even easier time to get money from the consumers.
I'm not looking forward to that future to be honest.
/Edit: After thinking about it some more: I think I agree that the potential music generation isn't going to change anything by itself. It's just another building block that will enable the music industry to eventually remove real humans from the equation. while we're slowly progressing on that path, the music generation alone won't push us to the logical conclusion.
Yes, it's inevitable. Some parts of this process are harder than than others, though. I think I've handled the hardest part, which is making catchy melodies (https://www.youtube.com/playlist?list=PLoCzMRqh5SkFwkumE578Y...). Live music is already over 50% of music industry revenue, so the transition could be a bit easier for musicians than for visual artists.
In fairness, I'd bet that most of your favorite tunes would become very un-catchy if they were stripped of harmony and percussion, and played by a quantized, expressionless square wave
AI that imitates well-known composers goes back much further, I remember attending a concert of "virtual mozart" works at UC Santa Cruz back in the late 80s/early 90s
OpenAI (the people behind GPT-3) already make one. It's called JukeBox. It can even make new unique songs in the style of existing artists with a simulation of their vocal style.
Considering that significant portion of the pop music produced in the last 30 years is made by a few people like Max Martin, one can assume that there's a formula for writing music that people love. Seems promising for AI to be honest.
That's the plot of Norman Spinrad's book Little heroes.
It starts with the idea that Music record companies can only make bad automated music, and in order to make a hit, they hire real musicians.
I think music, especially pop music, is in a completely different ball park than text or images. An AI can't just make a catchy song only with a training model and some rules. Successful musicians have a feel for what we will like and what will be trendy. An AI can't do that, at least not for a while to come. Well, thats just my thoughts, I'm not an expert on this.
Music can be algorithmically generated since it's formulaic, what I think machine learning models will struggle with at first is maintaining those formulas throughout a song. I imagine AI produced songs will/do sound winding and unhinged at first. All the right ingredients but stewed up in a pot.
Like Dall-E. What you'll get is music which apes music tropes really well, but strangely intentionless and disassociated. For genres like ambient where the intention is to be intentionless and disassociated (or can be), the result is like 'oh hey, AI is here!'.
It's weird. You could extrude endless amounts of The Caretaker, "Everywhere At The End Of Time" if you specified the recipe, with a 'deterioration dial' and some coding to determine how you vary the output. But you'd be free-riding on a pre-existing artistic intention that was originally implemented in dramatic, bold manner. To extrude endless amounts of this stuff is both fairly trivial and missing the point completely…
This is the TikTok-ification of music. The required response is the same here, as it is to the consumption of any other media, which is the same as when ordering food.
Don't order food out of a random dark kitchen. Order from a brick and mortar restaurant you know.
Don't use algorithmic feeds to discover video. Find a channel/producer you like and watch that.
The same applies to music. Assume everything fed to you is a tiktok style snippet of short term gratification that has no real value other than keeping you around for a few more seconds until the next snippet, while costing as little as possible for both producer and delivery platform.
Don't waste time being upset that everything is TikTok. If you scroll the TikTok feed, Watch Instagram reels, Order Dark Kitchen food, or listen to generated Spotify playlists, you are part of the problem.
the comment seems to place the blame and responsibility squarely on the consumer
yes consumers create the demand, and yes corporations are simply fulfilling that demand, but no, changing consumer behaviour is not the solution
it is unpleasant to say, but most (all?) people are not rational actors, and no matter how much proselytising to the choir on obscure message boards goes on, that is not going to change
given the free opportunity, especially with the urging of self-interested corporations, most people will become addicted to cigarettes, or slot machines, or micro-transactions, or heroin. the solution is not to teach some bible of consumer behaviour to a group of people that probably don’t do these things anyway, it is to use public organisation to force these large, perversely-incentivised groups of people to follow a set of rules that maintain a healthy media space
It is an awful side-effect of products getting bazaar-ified, where technology reduces friction in entering the marketplace and new "stalls" appear en masse. Human curation is a clear remedy - I can only listen to so much music or eat so many meals a day, so do I really need an algorithm to "optimize" that quantity?
I bet many listeners believe the Spotify-presented playlists are being curated, as if the stock image models in the thumbnail actually are jamming together and meeting with musicians face-to-face. So much of the recommendation service seems obfuscated to customers that I honestly don't understand what the end game is. Everyone is racing to just be "stickier" than their opponent that I can imagine a sudden pivot in customer demands where "Made for you" becomes a technology mistake akin to "Facebook Games"
Yes, you are part of the problem. But you are also a victim to the problem. You can only expect so much individual resistance against conveniences that are being launched into your brain by massive industries. Succumbing to these conveniences is highly understandable.
I mean I will use the algorithmic lists to discover new music - with limited success in recent years, I might add - and the autoplaying 'radio' after an album helps to keep me in a flow, but I usually start with a specific band / album or playlist.
Doesn’t it come down to „be a critical thinker instead of a mindless consumer“? From that follows that you don’t just scroll through the homepage but should use the search function, in an informed way.
Which in turn already fails at the majority of people being too lazy to type, and not knowing how to search for things they want. They quickly give up, and just take the easy route of consuming what’s served to them.
It’s the modern equivalent of falling on the sofa, switching on the TV, and just watching whatever’s on until they fall asleep.
Getting a majority of people to care about their media consumption is probably a hopeless cause.
Music Business Worldwide’s story explains how the moneymaking trick worked. “Soulful Music” had 467 songs by virtually unknown artists — which is to say, artists who may have been created for the purpose of this alleged scheme. The vast majority of songs were about 30 seconds long, which is the minimum length a song needs to be to count as a monetized play on the service.
The most probabl explanation for all this is that someone or someones in Bulgaria set up 1,200 computers with premium Spotify accounts, then had them play the songs on “Soulful Music” constantly. While it would cost $12,000 to set up all those accounts, the payoff would be worth it.
Yes, it's very important piece of Spotify's rev share model, it has arbitrage built in(!), most people are not aware of this.
Spotify pays PER STREAM, meaning that a bot account can funnel a massively disproportionate amount of revenue compared to a human one. In fact, you can generate more revenue than your subscription costs. Artists have complained about this forever, because 24/7 playlists at the gym with Justin Bieber on repeat would "steal" from the indie enthusiasts. And they are right! If I use my account to listen to one band only, my $10 contribution should go to them (modulo taxes, margins etc).
I'm sure they have "patched" some of these holes in recent times (ie some half baked abuse detection system that scammers can circumvent easily). But the per-stream principle remains, and it's such a massive incentive fuck-up from every angle, beyond salvation. And now they have (predictably) content farms and money laundering at their necks, and they're still not patching it.
> If I use my account to listen to one band only, my $10 contribution should go to them (modulo taxes, margins etc).
I'd argue that pooling the funds and divvying it up evenly across all played songs is actually the most fair. If anything, your implementation would be more prone to scammy content farmers (e.g. spammers using trials/locale pricing arbitrage to play only one $SCAM_ARTIST song per account).
Most importantly, however, your model is just not how memberships work in the real world. HBO doesn't (and shouldn't) reward a TV show with more budget because you only binged one TV show despite having access to all the show. Nor do gym memberships. Do you think Costco shouldn't allocate your membership funds for their gas stations because you have an EV? What about cable TV? What about your local/state/federal taxes?
> your implementation would be more prone to scammy content farmers (e.g. spammers using trials/locale pricing arbitrage to play only one $SCAM_ARTIST song per account).
Wat? You're saying that if you misimplement it and count all users including free/trial as equal paying you could get problematic incentives? Well, yeah. The point is today you have correctly implemented per-stream and there is a massive open arbitrage. A bot account can control 100-1000x of revenue compared to a human.
> Most importantly, however, your model is just not how memberships work in the real world. [...]
Spotify doesn't produce music, they act as middle man between you and the artists you listen to. There's a reasonable expectation that Spotify takes a cut between you and the artists you listen to, similar to an app store. Splitting your contribution between the artists you're listening to makes sense and importantly does not have arbitrage for content farm spammers (modulo a tiny bit of value from stolen accounts).
HBO and Costco aren't just funneling money around, they're producing physical goods and content, so it doesn't apply directly. You could make a more convoluted argument that they should follow the same principles but it's really not clear what that would mean, so I'm not gonna defend it. Same with taxes, I don't know what either per-stream or per-customer would even map to in that space.
i cannot fathom how your example here supports the idea that Bieber playing 24/7 in a gym is stealing money from the niche artist who are listen to by only a handful of fans?
I think the idea is that there are two Spotify subscribers in the whole world.
You, who listen to IndieBand™ for a total of an hour a week or so.
The Gym™ who plays Bieber 24/7 every day, all month.
You each pay $10 a month, so $20 total. Spotify allocates based on plays, so Bieber gets 720 "played hours" and IndieBand™ gets 4. So Bieber gets 720/724 of the $20.
The OP is saying in this case, Bieber should get $10 and IndieBand™ $10 (minus fees, etc).
Exactly. Per stream means that accounts that play music 24/7 are allocating orders of magnitude more revenue than yours or mine, much more even than what they paid in subscription fee. It's an open arbitrage incentive model, ripe for minimum risk exploitation by gray hats. Of course, this practice has the ability to erode actual artist revenue, which is already very low.
Spotify is actually half-decent here. The other streaming services seem to mainly compete on cutting costs, and many are little more than frontends over 7digital's services (who, surprise, also do virtually nothing to prevent spam).
That particular spammer has been doing this thing for years. He's easily recognizable: one characteristic is the way he uses colour filters. If you search qobuz (one of these terrible front-end streaming services which does no QQ) for the label "piano to go", and say, the album "Joker Games", you'll notice it's uploaded once for Bill Haley, and once for Bill Haley & his Comets. That was apparently close enough to be detected as a duplicate, so he put a blue image filter on the latter.
Most of this guy's albums eventually get deleted (or delisted), even on sites like Qobuz - probably more for lack of plays than for being detected as spam, I'd bet. So you don't get his full history. But I found out the French library service made some effort to catalogue streaming releases a while back, and with access to search in those, you can track his evolution. He was a lot sloppier in the past, for instance he didn't bother to come up with a new label name for every dump. But he's been at it since at least 2015.
I recently discovered YouTube videos that use a crappy spoken intro with a real human face in it and then proceed to display scraped StackOverflow answers as text with music in the background. the intro/outro is always the same as it is very generic.
Song spam does not seem to be endemic yet, but there are tons of low quality covers. Many classic rock songs have a couple of decent covers and half a dozen abysmal ones.
iirc those are the same "4 chords" used by the Axis of Awesome [1]. Certainly sounded like that to my ear from memory, but don't really have enough music notation/theory knowledge to know for certain and I may have been mis-remembering.
Look at the Peaceful Guitar editorial playlist on Spotify. (It’s one of the biggest and most streamed in the world.) Every artist on there that doesn’t have a Bio is a fake artist. Meaning Spotify owns the music. They populate their playlists with these artists they own and that way don’t have to make royalty payments to them. It’s very shady.
This doesn't make sense though - Spotify doesn't pay per stream (no-one does) but by a revenue share. So they would still pay out the same more or less - i.e. your premium subscription is put into a big pool and then divided out, it's not like your individual subscription goes to what you listened to.
I guess it could help shift some streams to indie artists with a less favourable contract compared to the big 4 record companies, but it hardly seems worthwhile?
The less 'real' music you listen to, the fewer royalties that Spotify has to pay to the record labels. If you're listening to more 'fake' music, then this saves Spotify money, and increases their profit margins. This is the exact same incentive behind their push into podcasts, another kind of audio content they don't have to pay royalties on. In this case, either because it's shared freely, or because they've acquired the original publisher themselves, e.g. Gimlet Media, Parcast and The Ringer.
This is crazy. Every single one I looked at was like this. Are these even real people playing guitar? If it were piano I’d say it’s 100% just MIDI arrangements.
I’m gonna look into this more. I have friends trying to break into the industry and gaming the Spotify and TikTok algorithms is (unfortunately) a big part of that these days. Having fake, phantom artists filling up your flagship playlists is so anti-creator I’m truly appalled at Spotify.
I think about these kind of activities and their motivations a lot.
Money is too valuable and rewarding that many people will constantly chase these hustle practices no matter the ethical implications or if they are shady. Even more so if they are legal.
With the current potential of AI and automation this just makes the resulting impact bigger for the 'good citizen consumer', like the sheet amount of autogenerated videos on YouTube that just read out loud some scrapped content, or even the stolen and reposted content, all in seek of finding a semiautomatic money cow that they can milk. After all, if that can give you easy scalable money, why not trying?
The bad part is that now consumers have to deal with a polluted search result and recommendations list, that is really hard to filter by the media owners because its a tough issue.
I understand the motivations, but I hate that the product degrades the quality as a result of a few actors .
Sometimes I wonder if this will always be the status quo in our world and I'm actually missing out on joining these practices as well and see if I can hit some jackpot, but I hate these side hustle practices.
This reminded me-- years ago, I ran into an old high school friend and asked him what he had been up to. He had gotten into graphic design and wound up partnering with a couple of guys who had a recording studio. They had a niche product and needed cover art -- and lots of it. He explained to me that they would release albums of fart noises on iTunes -- like many, many albums -- and the tracks would get a ton of plays, and people would purchase the albums. The partners made enough to pay him a good salary for just illustrating fart album cover art.
> A single kitchen operating under many names to increase its algorithmic “surface area”; another shape of things to come.
I find this infuriating. There's one dominican restaurant in my city that's listed on DoorDash, Grubhub and UberEats under over 20 different names with different subsets of their menu. All of these duplicate listings for the same place make it harder to search or browse the actual restaurant options in the area. Any time I come across a restaurant that looks worth trying out, I have to look up the address to see if it actually exists, or if it's just IHOP masquerading as a burger bar.
I wonder if that's a consequence of the "gig economy" working in the wrong direction. Perhaps the relationship should be between the delivery service "gig fleet" and the restaurants themselves. The ratings and feedback information would actually be more useful and would allow different restaurants to post different contract rates through the service based upon these ratings.
That would actually provide a service that doesn't dilute their brand. As it is, these restaurants have just found a way to counter this negative outcome by diluting this third party marketplace itself.
It's really bewildering when this happens. I've accidentally picked up food from one of these places and I nearly could not find it. It was a faceless nondescript commercial building that looked abandoned until I saw someone else walk out the door with their food order. Then I walked in and the place looked like some industrial building versus a restaurant. There was just cement floor in a small room, and I gave my order through a metal lined hole in the wall and they gave me my bag. Very oddly, in the background, I noticed Chic Fil A also being prepared in this ghost kitchen by people in tshirts. Such a strange concept. I made sure to never order food from a place that wasn't a brick and mortar again, and to call up restaurants directly.
In my experience, it's inferior in every way, perhaps because they have no accountability the same way a standalone restaurant has. For example, a pizza place near me opened multiple ghost kitchens, including a Saladworks and Guy Fieri's Flavortown.
There's no Saladworks if you visit their address, just the pizza place. They offer the same menu as Saladworks, the chain of fast service salad shops, but don't have the same supplier or training or equipment as a Saladworks. This pizza place just attempts to recreate the same salads using inferior ingredients, pizza toppings, generic dressings, and generic plastic takeout containers. The result doesn't taste like a Saladworks salad and is smaller than a Saladworks salad. I've tried it twice and neither recreation of Saladworks was faithful, missing ingredients both times, with a different lettuce mix than Saladworks uses, etc.
There's no Guy Fieri restaurant if you visit their address either. The food from this brand of the ghost kitchen is just inedible. It has 5 reviews on Yelp locally so far, and all are one star. They're microwaving burgers, sending out raw pasta, sending out easy mac with pizza toppings as a fancy loaded mac & cheese, using the same generic plastic takeout containers they do for Saladworks.
And yet at least they have a brand to burn. What I was actually frustrated with in my comment a few levels up is the places that list their own menu, sliced and diced in different ways, under 20+ different names. If someone complains about the quality of "Simply Crepes", they can just remove that listing, since those crepes are also available under "Morning Times Cafe", "Morning Breakfast Sandwich Bar", "The Daily Fare", "The Coffee Creperey", etc.
Haha, right on. I love me some Indian food, and there's ten different Indian "restaurants" near me, all located at the same address. I looked it up on Google Maps, and it's basically a hole-in-the-wall restaurant space that can barely hold one restaurant, never mind ten. Yet Grubhub et al. don't seem to care one way or another
Isn't it really easy to avoid all these fake restaurants, since the fakes only appear when you're searching for takeout? Just pretend you want to dine in, find a restaurant that you want that way, then pull it up in your takeout app of choice.
On the delivery services, you can search all of the menus of places that deliver to you at once. If I want a club sandwich, I can find what 32 restaurants will deliver me a club sandwich. There is no other place I know of that will tell me what restaurants in some radius of me sell club sandwiches for dine-in, that I'd then have to cross-reference with which are available on my preferred delivery service anyway. It's a lot more work.
I don't mind it myself. If you like the food it's OK if a downtown bar and grill is also sidelining as a chicken restaurant online. They have slow times and a paid kitchen staff. Go for it!
And some of the other ghost kitchens / home address food prep is pretty good. Like there was an ice cream place that delivered until midnight ( out of somebodies garage ) their ice cream was cooled down to well below zero so it was not melted when you got it.
I remember looking out of the window with my girlfriend trying to find a restaurant that Grubhub claimed was across the street. It was a ghost kitchen inside of another place and the food is pretty good, but I have generally stuck with places I've at least seen with my own eyes since then.
Seriously wonder why this isn't taken down. Should be pretty trivial given overlap of physical or payment addresses if they're registering the same restaurant several times on the same platform. Probably even possible to do it by hand.
If it generates revenue, it's not going down. There is absolutely no incentive at all to remove it. In fact, it gives the consumer more choices, and more ways to give money to the gig delivery service, and acts as a bulwark against being bullied by chains.
The equivalent on Spotify are not generating revenue, though. The existence of these algorithmically generated spam artists doesn't factor into anyone's decision to buy premium or not - at least not in a positive way. And still they're there.
(Shrug) This is how hotel restaurants, including some very good ones, have always worked. What difference does it make what other ethnicities/varieties of food the restaurant delivers?
Virtual restaurants don't have the reputation to burn like a brick-and-mortar with a well-known brand does. Normally if a restaurant presents itself as primarily selling one menu item (common examples: pizza, wings, milkshakes) you expect them do do it really well. But with virtual restaurants there's no need to build up a good reputation by serving quality food since your strategy is to drown out competitors by putting up dozens of fake brands, each with a supposed "specialty". If one of your identities is tarnished it is trivial to spin up a new one.
I remember the first time I got caught by this, I excitedly ordered from a "Korean fried chicken" place on Uber eats, and when the food arrived it was heat-from-frozen chicken tenders drowned in unappetizing sauce. I felt scammed, and the food was so bad, I didn't even manage to finish what I had ordered. Later I looked up the restaurant and discovered it shared an address with an unremarkable local burger chain. I had a similar experience with another virtual restaurant and at this point I always check if it's a VR before ordering delivery from an app and treat it as a strong negative signal to the quality of food.
This is obviously a lot different from your hotel example because the hotel restaurant's reputation is inextricable from the hotel itself; not only do patrons always know where their room service is coming from, bad food/service reflects poorly on the hotel as well. The hotel has a lot more to lose than the virtual restaurant.
What? Hotel restaurants have always convincingly appeared to people as 20 different restaurants while operating at the same physical location? A place serving a lot of different types of food is not the same thing, it's still one restaurant and they're not making any effort to hide that from the hotel guests.
The multitudes of different places at Disneyworld that all operate out of the same massive underground/hidden kitchen complex might be a better example.
I think you kinda have a point when it's one physical location presenting itself as multiple different options whose food would never belong on the same menu together at a dine in restaurant in the first place.
Where I think it's shady is one kitchen pretending to be 10 different restaurants all serving the same type of food under different restaurant names, just to basically game the system into presenting them to more consumers.
The main difference between this and the hotel case is that when a business lists another "restaurant" on Uber Eats/Grubhub/whatever they're effectively taking away some advertising space from all the other restaurants on the platform. Whereas when a hotel adds another "restaurant" frontend to their kitchen, nothing really changes for all the other places in town. And it would be painfully obvious if one hotel opened up 10 nearly identical e.g. indian restaurants on the same premises.
TLDR: It's not possible for the practice to get to a predatory level in the physical world so it's less of a concern, even though it's fundamentally very similar.
> multiple different options whose food would never belong on the same menu together at a dine in restaurant.
Any halfway decent hotel in a major Asian city will happily serve up (approximations of) Chinese stir-fries, Japanese sushi, Indian curries, Italian pizza and American burgers from its restaurant.
My thinking is, if the food is good, I couldn't care less if they are selling sushi, lawn furniture, plumbing supplies, and car insurance out of the same building.
If the food is not good, see above.
And if Uber Eats is dumb enough to fall for 10 Indian restaurants with the same physical street address and controlling interest, well... party on.
But that's just it. If the food is no good, then you may avoid "Wing Bucket" in the future. But then you try out "Thrilled Cheese" and it sucks, too! Okay, mark that one down as crappy. A month later you check out "Super Mega Dilla" and damn if it isn't shitty microwaved food as well.
A real restaurant in the real world can't play this game. Sure, there are combo KFC/Taco Bells out there, but it's not some sort of subterfuge. Obviously those locations are combining the two brands into one kitchen. Do with that info what you will.
When browsing options on an app, there is no similar affordance. You have to play detective to discover that IHOP, Super Mega Dilla, Wing Bucket, and Thrilled Cheese are all the same place. And only one of those is a real brand that carries any kind of reputation. The others can be burned at will if they don't work.
What frustrates me is that this scheme seems to be working? It's been like this for a while now, and I don't understand why people keep using these delivery apps. Cold food, delivered late, for twice the price, from a phony restaurant, by a gig worker that is more often than not getting shafted on their own expenses. What's not to love?
This is a really interesting comparison! And good context. Is the hotel operating multiple restaurants from one kitchen really different (as some comments here seem to imply) because each has its own physical space? In a way that’s more “skin in the game” for a hotel vs a ghost kitchen proliferating brands bounded only by their ability to cook the food. On the other hand, that’s a pretty significant bound.
Seems like most of the discussion here is focused on everyone's own experience with this, but can I just recommend Robin Sloan as a human in general for folks to follow in their lives?
His 2 novels, "Mr. Penumbra's 24 Hour Bookstore" and "Sourdough", are both really engaging reads. His 3rd novel is in progress and it seems like it's going to be more of the same quality. He also has a couple of novellas and a bunch of great short stories (some of which he did limited printings of and mailed out and one of which was procedurally generated and unique to the person who purchased it).
His newsletter is constantly filled with great media recommendations and he's always working on something interesting in that bridges the media and tech worlds.
Thanks for the recommendation. It is nice to see. I have read both of the novels myself, as well as at least one of the Penumbra novellas (it's been a while). I really enjoyed the Penumbra stuff; I have to admit I remember feeling slightly unsatisfied with the conclusion but I was definitely hooked and kept turning the pages and it was the first time I had seen a fictional book refer to Ruby programming. I really enjoyed his characters. I recall enjoying Sourdough as well, perhaps not as much as Mr. Penumbra's 24 Hour Bookstore, but that was an absolute pageturner for me so it's hard to compete with.
I am subscribed to his newsletter but I confess I rarely if ever read it. After having read this post and your comment, I'll make an effort to give it a solid look.
This is almost certainly generated as well, found through a recommendation series from the article.
It's so creepy because it feels like something that isn't made by a human, like composition uncanny valley!
Edit: As a side note, you should be careful to not like the music in this article. I suspect it will poison your Discover Weekly and other algorithmically generated playlists as they seem to be based on other listeners' data.
This one is interesting because even the album art is generated, rather than stock. The artist names are usually all uppercase. Perhaps a different author, but the techniques seem the same.
The same "discovery" happened to me half a year ago. Suddenly my queue contained just awful music at some point and I have googled the artists and found nothing. I decided to ditch Spotify altogether, because not only was I paying premium for this shit, but also it seems that Spotify doesn't really have the situation under control: https://www.rollingstone.com/pro/news/spotify-bot-takedowns-...
The next step in this, of course, is Spotify just generating music on the fly algorithmically based on your choices. Skip the middleman, so to speak. Maybe it is Spotify behind it?
I heard a similar story a couple of years ago: some no-name artist with a fairly good amount of plays on Spotify who could not be found anywhere else online. It was less obviously ai-generated back then (more instrumental than vibe/ambient), so it was speculated that it could be a musician working on commission for Spotify, who was then stuffing their playlists with these songs to cut costs. I can't find any traces of this story anymore though.
Fascinating find! He nails so much about todays music. the mystery artist is barfing algorithmic refuse into poor unsuspecting victims' neural grooves. Seems to echo shades of the creepy algorithmic bright colored child videos youtube have.
Interesting that this Ghost Producer chose a polyrhythmic tuples over 4 meter.
That's very difficult for the average listener to "find the one" downbeat.
Wonder what the thought process was behind this. Less likely to skip the track while trying to find the groove? More monetization with the short length?
I wonder if there are perhaps "attractors" (like chaotic attractors) in Spotify's recommendation algorithm. Like some sequence where it plays song A and that recommends song B which recommends song C which recommends song A again and it gets stuck in a loop (but longer / more complex).
If someone managed to generate a collection of songs that are all similar in the right ways to pop up when people leave the recommended playlists playing, they could rack of thousands of plays without all the trouble of having musical talent and writing good songs.
This almost sounds like the start of a Call of Cthulhu TTRPG scenario where a something really grim is being fed into public media pool for some nefarious purpose. Hmm. Nice.
This sounds quite familiar to what Yuval Noah Harari mentioned in his book "21 Lessons for the 21st Century" where he talks about how smart devices attached to our bodies and AI would be able to generate and play music based on what the sensors detect our feelings are.
Interesting to notice that even when there are only little variations on the same song, the analysis (which is powered by the Spotify API) shows very different indicators for each of the songs in danceability, accousticness, valence, etc. This makes rather unexpected that all those songs could come together in a recomendation system.
How fascinating. I guess this opens the old question "What is music?"
This is all definitely generated algorithmically using a really common chord progression in western music. (Think Pachebel's canon).
So a person (or group of people?) is creating generatively created music that is designed to be as pleasing to the ear of a human so that Spotify ranks it higher in automatically generated playlists which means more plays, which means more money.
It's crappy and spammy but also really kinda clever.
The author seems to be suprised by this, not just founding it on Spotify, but the whole process. E.g.:
"I love it, because it’s so strange, so dizzying, and — credit where due–because our mystery producer is truly going with the grain of the medium, in a way that no one merely “making albums” does, at all. What could be more 21st century, more “liquid modernity”, than releasing your music as a haze of variations into the swirling currents of the algorithm?"
But tons of artists have done the same in the past, whether in streaming platforms or in the vinyl and CD era. The most famous is probably Brian Eno who has made several albums using algorithmic generation.
In general, algorithmic composition is a sizable niche, both in academy (tons of music PhDs on that), live performance, and recording settings. And yes, people release those things too.
The strange thing is, the author should already have known all this, as his previous post goes:
"My band The Cotton Modules, formed with the composer Jesse Solomon Clark, also goes with the grain of the 21st century: our process combines AI tools with human skill and imagination, metabolizing a huge archive of recorded music into something genuinely new and exciting."
Not that big of a leap. A human also created the "algorithmically created bands", which is far easier than algorithmic music (they'd just register artists with random/generated names and some random choice of an image).
These generated 'things' only pass as music if you are truly uneducated in this regard.
I'm tired of people saying, as I've read in some comments here, that this will be our dystopian future.
These generated pieces of text, art, music, etc. are mere mashup copies without sense and purpose. None of the tricks of trade and conventions the artists know and apply to their art (such as, for pop music, a standard pop song is approx. 3 to 4 minutes long, has a verse, a chorus, a bridge which are played and repeated in a specific sequence) are recognizable here. Even the deviations from said conventions are so obviously non-intentional and random, without sense or 'dramaturgy', build-up and release of tension, etc. It's not music.
If this is our future, it is only if we become less educated and, sorry, somehow significantly dumber.
And if, according to the numbers on Spotify, people willingly listen to this garbage, I only have a misanthropic "we've earned it" for you.
The only takeaway from this article is that people are staggeringly stupid and uneducated and will consume whatever trash whichever industry will throw at them.
Perhaps it's time we spent money on education again.
Local Kitchens is like this. They have a single storefront serving 19 different cuisines. I'm guessing they're waiting to see if one or two pop off with the local community? Interesting to see this "algorithmic surface area" issue popping up somewhere else. Is this standard business school training?
Ooh, I think I found something like that too , a few years ago. I searched for a song (a real one, that I already knew) called "Dyson Sphere". That search still yields a few hits.
One of those was by an artist called Adal, and while it wasn't what I was looking for at all, I still listened to it. It was slow, rythmic and very simple music (think "human music"). When I looked for other works of this Adal, they had maybe like 8 songs, all of them the same simple melody but slightly different speed, rhythm and so on. At first I thought I was listening to that same song, but there definitely were slight differences.
At the time I didn't understand why anyone would produce such a thing, but in this context it makes much more sense.
I just searched again for Adal - Dyson Sphere, but there seem to be no hits. I wonder if the artist has since been deleted.
Your memories of Adal seem unreliable Are you sure you are remembering correctly is it possible it was a different name There Is No Musician Named Adal I’m sorry your query “Adal” did not return any hits The name Adal is invalid please re-enter a real musician You have misremembered aAdDal is nothing Invalid Entry: Confirmed: No Entity ADAL In Human History.
No, they're there on all the streaming services. YouTube reveals that it got there via DistroKid. I don't think it's spam, exactly - a spammer doesn't bother with choosing a theme in track titles as such. But it is possibly fairly low effort, as much DistroKid music is...
I don't think it is likely - but I wouldn't rule out that Spotify is behind this.
Spotify has collected user "skip data". That is a massive dataset. They have research to try to train AI to generate audio that humans are less likely to skip.
These results seem a bit more midi based, and I would assume that the models they generate would work on raw audio, so doesn't seem like a perfect fit.
Though, in general, it must be tempting for Spotify researchers to test their creations on their customers.
(Note: I do not believe it is ethical to collect skip data, or any data from customers and use it to train AI without a very steep OPT-IN process. Burying some legal disclaimers in wall of signup jargon is a terrible practice.)
> Recently, Spotify announced their Discovery Mode program. This allows labels to discount selected parts of their catalog in return for increased promotion via Radio and Autoplay
Uggh. I wish payola were actually illegal in practice.
> So who’s really losing in this equation?
Possibly, older out of touch-artists who think music should always be something meaningful and “culturally important” and are perhaps – just slightly – butthurt that no one cares, or that other people have figured out another way to be successful.
I can't tell if this bit is dripping with sarcasm, or if the author is actually this contemptuous of musicians and this dismissive of the difficulty associated with getting "discovered" these days.
Either way, it is a fascinating article. Wouldn't have guessed that Sony was one of the leading spammers/fraudsters (see other threads on this article) in this space, or that Spotify was complicit.
This actually makes me angrier than the Sony rootkit fiasco.
Music has always confirmed to the medium. The standard song structure of pop no longer serves the medium's purpose. I wholly expect pop songs to get shorter on average and take advantage of what social media has trained our brains to consume.
It sounds like the Fake Artists thing which is designed to make sure Spotify gets all the money and does not have to pay royalties.
They basically either have salaried/contracted musicians or pay a one time fee without royalties to make passable songs in every genre. Then they push their own songs in lists so people will listen to them more than "real" artists.
In his book Sourdough, he spends quite a while describing the music of the Mazg - a fictional group of European migrants. For the audiobook version of Sourdough, he trains a SampleRNN model on a particular genre of Croatian music, then has the model produce Mazg music. The description of machine learning is vivid and approachable, and the audio clips are fantastic.
What "cost" to the money launderer? What is the "cost" like, compared to e.g. mules taking cash to casinos, or laboriously processing e-card visa style money?
ie how "effective" is this, compared to alternatives? I had a belief the rate of IPR payment was low, so this is probably a low rate of return, high cost, fully digital method.
Which in my theory, would also be making spotify significant profit, as the facilitator. Are they now at some legal risk if they can "detect" this traffic?
Every single business can be a vector for money laundering. From cash laundromats to high end technology, if the launderer owns the business then it has potential. I would estimate the government catches somewhere in the ballpark of 0.00000001% of all laundering. Businesses also have zero incentive to stop this activity as long as they get their margin.
I have asked people who work with preventing money laundering how expensive it is, how much you get back from a launderer in clean money for every dollar you pay him in dirty ones. They have all estimated a 20-25% premium.
That doesn't seem like it would be sustainable if only 0.00000001% was caught.
yes. but the cost side matters: if a criminal has $1,000,000 but can only recover $1,000 "safely" burning $999,000 then a method which returns $500,000 for burn of $500,000 is better, if riskier. "it depends"
I worked in an office next to an Uber clone that was clearly unsuccessful. Yet they had a significant Series A from a Russian oligarch, then got bought a few years later at something like 10x the Series A funding amount by a different Russian oligarch. There was a press release and a news story about how wonderful this was.
This scheme was so clearly a laundering play, but everyone pretended it was some amazing exit. I'm confident the tech industry is full of these shenanigans, but I've never seen anything like this uncovered.
If they can offload on the market, this might be net lossless or lower loss at scale than other methods.
I guess taking net company worth into a bank in Cyprus or something and using it as security to buy land, boats, whatever achieves the real outcome. Even if only some 1/nth can be collateralised
The experimental artist Arca released an album[1] in 2020 which consisted of 100 AI-generated remixes of her song Riquiqui. It was part pretentious artsy experiment and part marketing stunt for some AI company, but an interesting concept nonetheless.
Nice discovery! I suspected the same, since all the accounts were made a few months ago, but I couldn't find any on any of the other streaming services.
It could be that you're right, that this fake artist used to be on Spotify but was removed. But that doesn't explain why you can find e.g. Romilda Gebbia on Spotify but not on YouTube.
When I think about it, it could also be that the spammer is on multiple services, but they use different algorithmically generated artists for each service.
What was extremely useful in you finding this on YouTube, is that we can see by which distribution service they got access to the streaming services (information that all the other services hide, annoyingly). It's a French World Music thing.
Which is especially interesting, since I also traced back the "Belle Wood" spammer to a French World Music thing, although a different one (divercities/diMusic). I wasn't 100% sure if they had anything to do with the spam, they could plausibly be a victim of it themselves, but now I have more circumstantial evidence. Thanks!
(If you know of a less noisy forum where we could discuss tracking down/locking out algorithmic spammers, I would be grateful.)
My account once got taken over by one of those hackers that will play the same song over and over to boost its rating
I mostly listen to metal, but one day I noticed that on my last.fm my most listened to track was suddenly "The 1975 - Chocolate". A song that I'd never heard before or since. The track was played at completely random times, often multiple times in a row or when I wasn't awake.
I am very curious about the author's comments on whether all these variations have been made automatically.
I think that with a standard Ablenton project, for example, with several VSTs (this sounds like a Kontakt library) and replacing the midi clips by the Magenta's MusicVAE output for example, you could make infinite variations.
As a producer, I feel like this could be more related to people using Splice and other popular sample packs in order to just get something uploaded onto Spotify? I've come across a lot of very similar tracks on Spotify which utilise all the same sounds from the same sample packs and they're almost identical.
After spending some time and frustrations releasing music and trying to understand growth I believe Spotify is propped up by playlist/user fraud.
If somebody could go an audit Spotify and get to the real human numbers I think it would shock and inform how crazy our world is.
I tried listening to the provided playlist and recommendations yesterday and a few hours later Spotify reset my account password due to detected suspicious activity. So it seems like they are at least actively trying to mitigate such money laundering schemes.
I am waiting for services to give up on purely algorithmic approaches and start exploring recommendations based on other users whose tastes the user likes. Even better if you could have different clusters of selected users for different occasions.
Always nice to read a nice fairytale with the wonderful drawings by Theodor Kittelsen. The norwegian artist is famous for his depictions of trolls, and also illustrated the old icelandic book Snorre saga with depictions of vikings.
I know it was a rhetorical question about how to produce such tracks en masse, but Sonic Pi would be reasonably good at this if you ever did want to algorithmically generate a ton of tracks using the same stems.
You don't publish on Spotify. You publish on one of the services which have a distribution deal with Spotify, like TuneCore or CDBaby. Some of the dodgier services like that have features to make this exact sort of thing easy (for instance, you get to choose a new publisher label name every time).
One time I made a "Hitler finds out"* meme and uploaded it to youtube. My video immediately got 2 copyright claims on it. One was for the video and was probably by the film rights holders. The other was for the audio and it was this song! https://www.youtube.com/watch?v=RSlbcRumfwk I assumed that this was some kind of scheme. The audio claim has since been removed (not by me and I don't know when).
People will set up fake Spotify artist accounts with stolen identities and bank accounts, pay a musician for songs that pass as music, and then bot millions of streams on them. At this point there are so many of these fake profiles and songs that the music, which is simple “mood music” normally (which happens to be easy to make), is appearing on real playlists and being recommended to real listeners.