Say I pay $10/month and only listen to 2 artists, each with .001% market share. IMO, each one should get $3.50 from my pool of subscription money. Using their current scheme, each would get $0.00007 while Lady GaGa would get $0.30 when I didn't even listen to Poker Face.
Your method would only increase complexity for Spotify and serve no real purpose in terms of getting your favorite artists more money.
I think it's quite reasonable to assume that differences in listening and paying habits between users would NOT average out, and a per user split would be fairer.
I don't know how Spotify is doing financially, but hopefully they make money off free users as well as paid users (but I'd not be surprised if they end up a bit like Opera did -- making money from paid users and licensing/bundling deals -- and just using the commercial breaks/ads as stick to guide users towards the paid service).
Either way I'd much prefer being able to pay for lossless records that I get to keep -- I gave up on Spotify quite early as it ended up a little like youtube -- come back to a playlist after a few months and half the songs were gone. I know they're better now, but that experience just underlined the idea that paying for licensing content in a way that leaves you vulnerable to that content disappearing is a very bad deal for me as a listener/consumer.
Things are not so simple.
Under the current spotify system, artist B would get 90% of the revenue and artist A would get 10%. Under the parent's proposal, artist A would get 89% of the revenue and artist B would get 11%.
In any case, I'm willing to bet that the parent's proposal would actually backfire. I would wager that people who listen to obscure music also listen to more music so they're actually receiving more under the original spotify proposal.
In a similar vein, there's also Magnatune and Jamendo.