It's different because if you imagine Spotify only had 2 users, Alice and Bob, and Alice listens to Indie Band X 10 times a month, and Bob listens to Megastar Y 10,000 times in a month, Y would receive almost all of the artist revenue in the current model, but in the model that GP is proposing, both artists would make the same.

Essentially A's money is going towards Y, even though she never listed to Y.

(I'm not expressing an opinion on which of these I prefer)

It can be a huge difference. Say this month I only listen to a completely unknown artist. I am the only person listening (to make the calculation easy). I've listened to 10 of his songs and this is everything I've heard this month.

Now if my 7$ are distributed to all artists I've listened to this artist will now receive 7$. If my 7$ is thrown in a global pool and split by global stream counts this artist gets almost nothing.

Of course this is a extreme example but it should illustrate why this can matter.

> If my 7$ is thrown in a global pool and split by global stream counts this artist gets almost nothing.

Not by definition. If you generate more streams this month than the average user, your streams are a larger fraction of the stream counts than your 7$ is of the revenue pool, and this artist will get paid more than if he only got your 7$.

Splitting the revenue per-user instead of per-stream can certainly make a huge difference, but switching to per-user instead of per-stream benefits artists with listeners that generate a below-average amount of streams per month, regardless of their absolute stream numbers.

Their current split favors big artists. 90% of the music I listen to is jazz and classic, and if I was on Spotify (I use Qobuz instead), 90% of my subscription (minus Spotify's share) wouldn't go to these artists, instead most of it would go to the most popular artists on Spotify. That's not fair, I am de facto subsidizing the biggest artists, even if I don't listen to them, while the artists I like receive less.

And there's the fact that the payment per stream is low on Spotify compared to Qobuz or Tidal.

It’s the weighting. If you listen to very little music, but 1 artist, your $10 (in that scenario) would go to them.

Instead, your listening is not weighted in any special way so it goes into the global pot of money and hours listened and gets diluted by the people listening more.

If I'm listening to only Primus for a month basically none of my subscription goes to Primus.

Alice pays 10$ per month and play (only) a thousand times a song of Madonna each month Bob pays 10$ per month and play (only) a hundred times a song of Queen each month

Currently Madonna will get 1000/1100 of artist's share => 10/11 x 70% x 20 = 12,73$ While Queen will get 100/1100 => 1/11 x 70% x 20 = 1.27$

With a per subscriber split => Queen and Madonna will get 7$

Right now there is an incentive for artists to create shorter songs so they get more stream = more revenue