Most people, and most albums, though, are not going to outperform those economics. Simply put, $5-10 a month for access to 15M tracks is a joke and a big loss for the industry. I look forward to the labels realizing it and walking away.
Heck, for fun, let's say you were to listen to music 8 hours a day every day for a month (~30 days) -
8hrs * 30 days = 240hrs
(240hrs * 60min/hr) / 3min/song = 4800 songs.
Or, from the other side: Spotify's premium is $10/mo, which comes with no ads. They pay $.003 out to each band - let's just pretend they follow Amazon's 'agency' policy of a 70/30 split, so Spotify's making roughly $.001 on each song, and it's costing them $.004 total for that song (hypothetical - just stick with me). If we assume they're not losing money, then an average $10/mo user must listen to at most 2500 songs per month, and if Spotify's costs are higher than $.001 per track, the number goes down.
Point is, the industry's providing _ACCESS_ to 15M tracks, but they're only having to deliver ~2500/mo - but that's a different bundle of 2500 songs for each user. It MAY make sense from their end just to call it 'unlimited' and rely on the fact that the user can't consume music fast enough to really upset the economics for them.
(Incidentally, if you were to decide to listen to each of those 15M songs once, you'd wind up paying:
15,000,000 Songs * 3min/song = 45,000,000 min
45,000,000 min = 750000 hrs = 31250 days ~= 1027 months
1027 months * 10/mo = $10,270
Not sure this is any different to being able to listen to any radio station in the country for free. I don't think that devalues music.
BTW, most of those 15M tracks are deep in the long tail of obscurity that would never see the light of day without an all you can eat subscription model.
As for those buried tracks, keep in mind that it's unlikely that Spotify will cut anybody a check for $0.029, or even 10 times that amount. So unless those albums get listened to a lot, they're not any better off in Spotify.