When Jay-Z took the stage in March 2015 to introduce Madonna, Beyoncé, and Kanye West, promising musicians fairness, no one knew that three years later Norwegian journalists would find traces of hundreds of millions of streams on a hard drive—streams that may never have happened. A story about how a platform for artists ended up at the center of an investigation into data manipulation—and why the streaming economy turned out to be too fragile for its own promises.
March 30, 2015 Jay-Z gathered sixteen A-list stars in New York. Beyoncé, Madonna, Rihanna, Kanye West, Daft Punk, Nicki Minaj, Alicia Keys—all of them signed a turquoise document called a declaration. The text promised a new music industry where artists control the platform, not the other way around. Two months earlier, Jay-Z had acquired the Norwegian-Swedish company Aspiro, which owned the streaming service Tidal, for $56 million. The presentation was theatrical: the stars stood shoulder to shoulder like a Justice League of pop music, while cameras broadcast the event live.
Tidal positioned itself as a premium alternative to Spotify and Apple Music. The subscription cost more—$19.99 for the lossless-quality version, versus the standard $9.99 from competitors. Jay-Z promised that the higher price meant higher royalties for musicians. Artists, he said, deserve more than fractions of a cent per stream—and two days after launch, speaking to students at New York University, he vowed to split money among musicians more fairly than Spotify and Apple do. The idea was simple: a platform owned by artists themselves would distribute money more honestly. But the streaming economy is built so that honesty depends not on declarations, but on data. And data can be edited.
Tidal's first exclusive was Rihanna's ANTI album in January 2016. Next came Kanye West's The Life of Pablo in February and Beyoncé's Lemonade in April. All three releases were available only on Tidal, which was supposed to attract subscribers. The company reported triumph: 250 million streams of The Life of Pablo in the first ten days, 306 million streams of Lemonade in the first fifteen. The numbers looked astronomical, but no one could verify them—streaming data remained inside Tidal. In a world where streams equal money, the absence of independent audit turns the platform into judge of its own case.
May 9, 2018 the Norwegian business newspaper Dagens Næringsliv published material that shattered Tidal's narrative. DN journalists obtained a hard drive with the service's internal data—streaming logs for the period from February to April 2016. The drive contained billions of rows: each entry recorded which user launched which song at what moment. DN handed the data to experts from the Norwegian University of Science and Technology (NTNU) in Trondheim. The researchers conducted statistical analysis and released a 78-page report. The conclusion was harsh: 320 million records of streams looked fake. Accounts of more than 1.7 million users were affected.
The anomalies were glaring. Certain accounts played Lemonade and The Life of Pablo tracks hundreds of times in a row without breaks—sometimes the same song for twenty-four hours straight. The patterns didn't match human behavior: no pauses, no skips, no switching between albums. An NTNU professor leading the analysis noted that such spikes cannot be explained by organic listener interest. The data showed that fake streams were distributed unevenly: the lion's share fell precisely on Kanye and Beyoncé's exclusives. Other albums available on Tidal during the same period showed no such anomalies. This wasn't a technical glitch—it looked like a targeted operation.
DN published specific examples. According to the logs, Washington student Tiare Faatea "listened" to songs from Lemonade 180 times in 24 hours—almost eleven hours of Beyoncé continuously, equivalent to fifteen complete album playthroughs. Oslo music critic Geir Rakvaag racked up 96 songs from The Life of Pablo in one day, 54 of which came at night—at exactly two and five in the morning, same account, simultaneously. A student from Arendal named Ida managed to "listen through" 251 Lemonade songs in 24 hours—more than twenty complete album cycles, where the track "Hold Up" played every hour with second-level precision: 13:42:22, 14:42:22, 15:42:22. Researchers emphasized: such patterns are impossible even for the most devoted fans. To honestly listen to an album a hundred times in a day, you'd need to neither sleep nor eat, pressing play in time with a metronome. The hard drive turned abstract press release numbers into an anatomy of possible fraud. Each line in the logs was evidence: either millions of Tidal users behaved like robots, or the streams themselves were robotic.
Tidal reacted instantly and furiously. In an official statement, the company called DN's investigation "a smear campaign based on stolen and manipulated data". Tidal representatives claimed the hard drive was obtained illegally and that the data could have been altered after the theft. The company stated it had conducted its own internal audit and found no evidence of manipulation. Moreover, Tidal brought in an outside cybersecurity firm to check its systems.
Tidal's arguments rested on three theses. First—data legitimacy. If the hard drive was indeed stolen, any conclusions based on it are legally dubious. Second—data integrity. Tidal claimed DN hadn't proven the logs weren't altered after the theft. Third—motivation. The company hinted that competitors interested in discrediting the service might be behind the investigation. All three arguments sounded logical, but they didn't explain the main thing: if the data is fake, why do the anomalies so precisely match Tidal's exclusive releases? Why did the manipulation concern exactly those albums whose success the company advertised loudest?
DN responded by publishing additional materials. The journalists showed that data on the hard drive aligned with other known metrics: for example, with Tidal's subscriber count during that period and with the company's public statements about audience growth. If the data was fabricated, the forger did meticulous work, reproducing the log structure down to technical details. NTNU researchers emphasized that statistical anomalies are visible not only in absolute numbers but also in patterns: fake streams are distributed across time and accounts as if generated by an automated system. Manipulation isn't chaos, it's an orchestra. And the hard drive from Oslo recorded the score.
At the end of 2018, the case moved beyond journalism. Økokrim—the Norwegian agency for investigating economic crimes—announced the start of an official investigation. The initiative came from Norwegian collective rights management organizations responsible for collecting and distributing royalties. If Tidal really inflated stream counts, it affected payouts to all rights holders included in the catalog. The streaming economy works like a common pot: the platform divides subscription money proportionally to stream shares. If one artist gets millions of fake listens, everyone else gets less. This isn't abstract injustice—it's direct money redistribution.
Økokrim questioned former Tidal employees—insiders once responsible for data analysis at the company. Tidal officially stated the company is not a suspect in the investigation but is cooperating with authorities and providing requested information. This was a fine line: Økokrim didn't accuse Tidal directly, but didn't close the case either. The investigation dragged on for almost five years—even slowed by the pandemic, which blocked the transfer of key documents from abroad. And in June 2023 the period was finally placed: prosecutors deemed the evidence insufficient and closed the case. No charges, no exoneration—just a statement that for court, guilt must be proven beyond reasonable doubt. Tidal continues operating, but the shadow of 2018 remains.
Norwegian rights management organizations were trapped. On one hand, they couldn't ignore DN's investigation—too much money was at stake. On the other—they had no tools for independent verification of Tidal's data. Platforms aren't required to publish streaming logs, and audit is only possible with their consent. As a result, rights holders depend on the honesty of those counting their money. It's like if a bank determined your account balance without showing a statement. Økokrim could seize data by court order, but even the agency needed years to gather enough evidence to move forward. Or maybe the evidence was never found.
Streaming is vulnerable to manipulation by design. Unlike physical sales or downloads, where each transaction is tied to payment, a stream is an event inside a closed system. The user pays a subscription, and then the platform itself decides which listens to count, how to weight them, and how to split the money. Spotify, Apple Music, Tidal—they all use the pro-rata model: the total subscription money pool is divided proportionally to stream shares. If an artist has 1% of all listens for the month, they get 1% of the pool. The system is simple, but it turns each stream into a financial instrument. Inflating listens means redrawing money distribution.
Technically, generating fake streams isn't difficult. You can create thousands of fictitious accounts, program bots to play specific tracks, imitating real user behavior. You can use real accounts if the platform itself controls the data—just add records to the database. Harder to do it undetectably. The anomalies NTNU experts found were too crude: the robots didn't pretend to be people convincingly enough. But if manipulation is done professionally—with randomized pauses, track variety, distribution across regions and devices—detecting it is virtually impossible without access to raw logs. And raw logs exist only at the platform.
The problem isn't just Tidal. In 2019 a journalist investigation revealed manipulation schemes in Spotify: fictitious artists with millions of streams generated by bots. And by 2025 journalist Liz Pelly in Harper's Magazine described an entire industry of "ghost" performers—studio pseudonyms under which producers upload cheap background content to editorial playlists, passing it off as real musicians and taking a slice of the pool from actual artists. Each scheme exploits one weakness: platforms simultaneously sell subscriptions, store data, and calculate payouts. They're judge, jury, and witness in one. The system runs on trust, but trust without transparency is faith, not contract. Musicians receive stream reports but can't verify where the numbers came from. They see the orchestra's result, but the score is hidden.
The Tidal story didn't end with a verdict because no verdict happened: in 2023 Økokrim closed the case for lack of evidence. DN didn't publish new data, Jay-Z didn't testify. In 2021 the rapper sold most of Tidal to Jack Dorsey's Block (formerly Square) for $302 million, retaining a stake in the company. The service continues operating, competing with giants, but its subscriber base remains modest compared to Spotify or Apple Music. The 2018 investigation became a footnote—a mention in articles about streaming problems, but not a legal precedent.
The question DN raised remains unanswered: can a platform be a reliable arbiter of its own success? When a service simultaneously sells subscriptions, stores data, and calculates payouts, each of these roles creates conflict of interest. Inflating streams for an exclusive release is profitable: it attracts subscribers, impresses investors, strengthens the platform's reputation. Counting streams honestly means showing real audience interest, which may prove more modest than promised. In the streaming economy, data is product, currency, and proof all at once. If the one counting is interested in the result, the count stops being neutral.
Jay-Z promised fairness on stage, and three years later Norwegian journalists discovered hundreds of millions of phantom listens. Maybe the data was stolen and distorted. Maybe the anomalies are explainable by technical failures. Maybe Økokrim stopped one step short of truth because proving it was impossible. Or maybe the streaming system is built so that the distinction between honest success and manipulation blurs in the opacity of server logs, where the only witness is the platform itself.