You’re neglecting what fraction of users actually listen to indie artists’ music. It does hurt payoffs when listeners listen to more (diverse) music, but the payoffs are not diluted by the masses who individually listen to less, but cumulatively listen to much more music from mainstream artists. This combined meta-user is listening to a LOT of music, of which dilutes the number of plays of indie music. As long as there are users who listen to mainstream music but not indie music, indie artists will always be better of under the proposed scheme.

My intuition says that an example of Simpson’s paradox might be hiding here, in case someone is interested in constructing the mapping and making it explicit.