Spotify Peeved After 10,000 Users Sold Data To Build AI Tools (arstechnica.com) 17
An anonymous reader quotes a report from Ars Technica: For millions of Spotify users, the "Wrapped" feature -- which crunches the numbers on their annual listening habits -- is a highlight of every year's end, ever since it debuted in 2015. NPR once broke down exactly why our brains find the feature so "irresistible," while Cosmopolitan last year declared that sharing Wrapped screenshots of top artists and songs had by now become "the ultimate status symbol" for tens of millions of music fans. It's no surprise then that, after a decade, some Spotify users who are especially eager to see Wrapped evolve are no longer willing to wait to see if Spotify will ever deliver the more creative streaming insights they crave.
With the help of AI, these users expect that their data can be more quickly analyzed to potentially uncover overlooked or never-considered patterns that could offer even more insights into what their listening habits say about them. Imagine, for example, accessing a music recap that encapsulates a user's full listening history -- not just their top songs and artists. With that unlocked, users could track emotional patterns, analyzing how their music tastes reflected their moods over time and perhaps helping them adjust their listening habits to better cope with stress or major life events. And for users particularly intrigued by their own data, there's even the potential to use AI to cross data streams from different platforms and perhaps understand even more about how their music choices impact their lives and tastes more broadly.
Likely just as appealing as gleaning deeper personal insights, though, users could also potentially build AI tools to compare listening habits with their friends. That could lead to nearly endless fun for the most invested music fans, where AI could be tapped to assess all kinds of random data points, like whose breakup playlists are more intense or who really spends the most time listening to a shared favorite artist. In pursuit of supporting developers offering novel insights like these, more than 18,000 Spotify users have joined "Unwrapped," a collective launched in February that allows them to pool and monetize their data.
Voting as a group through the decentralized data platform Vana -- which Wired profiled earlier this year -- these users can elect to sell their dataset to developers who are building AI tools offering fresh ways for users to analyze streaming data in ways that Spotify likely couldn't or wouldn't. In June, the group made its first sale, with 99.5 percent of members voting yes. Vana co-founder Anna Kazlauskas told Ars that the collective -- at the time about 10,000 members strong -- sold a "small portion" of its data (users' artist preferences) for $55,000 to Solo AI. While each Spotify user only earned about $5 in cryptocurrency tokens -- which Kazlauskas suggested was not "ideal," wishing the users had earned about "a hundred times" more -- she said the deal was "meaningful" in showing Spotify users that their data "is actually worth something." Spotify responded to the collective by citing both trademark and policy violations. The company sent a letter to Unwrapped developers, warning that the project's name may infringe on Spotify's Wrapped branding, and that Unwrapped breaches developer terms. Specifically, Spotify objects to Unwrapped's use of platform data for AI/ML training and facilitating user data sales.
"Spotify honors our users' privacy rights, including the right of portability," Spotify's spokesperson said. "All of our users can receive a copy of their personal data to use as they see fit. That said, UnwrappedData.org is in violation of our Developer Terms which prohibit the collection, aggregation, and sale of Spotify user data to third parties."
Unwrapped says it plans to defend users' right to "access, control, and benefit from their own data," while providing reassurances that it will "respect Spotify's position as a global music leader."
With the help of AI, these users expect that their data can be more quickly analyzed to potentially uncover overlooked or never-considered patterns that could offer even more insights into what their listening habits say about them. Imagine, for example, accessing a music recap that encapsulates a user's full listening history -- not just their top songs and artists. With that unlocked, users could track emotional patterns, analyzing how their music tastes reflected their moods over time and perhaps helping them adjust their listening habits to better cope with stress or major life events. And for users particularly intrigued by their own data, there's even the potential to use AI to cross data streams from different platforms and perhaps understand even more about how their music choices impact their lives and tastes more broadly.
Likely just as appealing as gleaning deeper personal insights, though, users could also potentially build AI tools to compare listening habits with their friends. That could lead to nearly endless fun for the most invested music fans, where AI could be tapped to assess all kinds of random data points, like whose breakup playlists are more intense or who really spends the most time listening to a shared favorite artist. In pursuit of supporting developers offering novel insights like these, more than 18,000 Spotify users have joined "Unwrapped," a collective launched in February that allows them to pool and monetize their data.
Voting as a group through the decentralized data platform Vana -- which Wired profiled earlier this year -- these users can elect to sell their dataset to developers who are building AI tools offering fresh ways for users to analyze streaming data in ways that Spotify likely couldn't or wouldn't. In June, the group made its first sale, with 99.5 percent of members voting yes. Vana co-founder Anna Kazlauskas told Ars that the collective -- at the time about 10,000 members strong -- sold a "small portion" of its data (users' artist preferences) for $55,000 to Solo AI. While each Spotify user only earned about $5 in cryptocurrency tokens -- which Kazlauskas suggested was not "ideal," wishing the users had earned about "a hundred times" more -- she said the deal was "meaningful" in showing Spotify users that their data "is actually worth something." Spotify responded to the collective by citing both trademark and policy violations. The company sent a letter to Unwrapped developers, warning that the project's name may infringe on Spotify's Wrapped branding, and that Unwrapped breaches developer terms. Specifically, Spotify objects to Unwrapped's use of platform data for AI/ML training and facilitating user data sales.
"Spotify honors our users' privacy rights, including the right of portability," Spotify's spokesperson said. "All of our users can receive a copy of their personal data to use as they see fit. That said, UnwrappedData.org is in violation of our Developer Terms which prohibit the collection, aggregation, and sale of Spotify user data to third parties."
Unwrapped says it plans to defend users' right to "access, control, and benefit from their own data," while providing reassurances that it will "respect Spotify's position as a global music leader."
As a long time Spotify user... (Score:3)
...This is the first time ever that I heard of this "Wrapped" thing.
Isn't Spotify a platform for listening to music that you like?
uncover overlooked or never-considered patterns (Score:1)
Re: (Score:3)
Yes.
That's one thing AI is really good at: Finding patterns. And Spotify for example is really bad at using your data to build good playlists. They have all data that would enable them to do it, but for some reason they suck at it. If you now take your data and train a model that is able to extract patterns, you can for example try to build better playlists.
A short explainer on what such an AI finds. Let's say you have a lot of data points in X,Y. If all have the same X or all the same Y coordinate, you cle
Re: uncover overlooked or never-considered pattern (Score:1)
Re: uncover overlooked or never-considered patter (Score:2)
I would have said this is self evident?
I've been curating Pandora channels by mood for a long time.
I realized fairly early that what I wanted to listen to was context dependent, not really wanting a Chopin Nocturne while working out, nor KLF while trying to relax.
Re: (Score:2)
Sort of, but not quite. AI is not *actually* good at finding patterns. The truth is that AI models depend on humans setting up the problem first, and humans creating the class of features that will uncover patterns easily. This has always been the case since the dawn of time, ca 1958.
To state this another way, AI cannot find patterns if the inputs don't show the patterns clearly. The wildly successful applications of AI to-date have used human insight and experience to narrow down and curate the input si
Re: (Score:2)
Nah, that was the revolution of deep learning. Prior to that you designed features you deem helpful and then fit the model to use these features. Deep learning (if you want so the second wave of AI research) just said put in the data, put in a lot of layers that can learn to create arbitrary features and fit the network to both obtain a part that can transform inputs to features and a part that uses the features to e.g. classify the data. Have you ever seen a visualization of the learned CNN filters? They l
Re: (Score:2)
The deep learning revolution did not solve the problem you claim. What deep learning does is allow more complex piecewise linear functions to be modelled efficiently (if you use relu that is, which is the most popular activation (*)). That's both a blessing and a curse.
What actually happened in the deep learning revolution is that humans solved the problem of designing basic features over many generations of papers and progressively simplified the solution, discovering what is important and what isn't. Th
Nice contradictory statement there (Score:2)
"Spotify honors our users' privacy rights, including the right of portability," Spotify's spokesperson said. "All of our users can receive a copy of their personal data to use as they see fit. That said, UnwrappedData.org is in violation of our Developer Terms which prohibit the collection, aggregation, and sale of Spotify user data to third parties."
But not if you want to do something that allows you to profit off of your own data, and especially not if you get help from someone else to do it.
Re: (Score:2)
Exactly. You have the "right of portability" as long as you don't port it somewhere we don't like.
Our rights (Score:2)
You do now, all right. But I had to DMCA your ass to make you implement data takeout, after being told (after waiting for a year and more) that giving users access to their own listening history was "not a priority".
Re: Our rights (Score:2)
Curse slashdot and their lack of editing functionality. I meant of course to GDPR their ass.
Spotify peeved cause it wanted to sell their data (Score:2)
Re: (Score:1)
Usage of spotify data? (Score:2)
Unwrapped is using data their users provide. Spotify can go sue their own users if they think they can forbid them certain uses of their data. They surely have no contract with Unwrapped, so there is nothing on that side that Unwrapped could have violated. They should be careful about trademarks though, because trademark laws suck and allow for a lot of unjustified takedowns.
Users should request their GDPR data and share it. It contains (and MUST contain) a lot more data. I didn't request it, but I've read
Urm... (Score:1)