Missing the point about pricing
Most people don't have a problem with the current set-up:
Imagine two researchers at different points in their work; A is carrying out a literature review and only needs to skim through the paper, B is experiencing a problem that the methodology described in the paper could solve.
How much is the paper worth to A? - it's only one of many - so probably pennines. For B the paper is essential, so they'd probably pay through the nose for it.
A is an undergrad and they can access the material through their University library at minimal personal cost - the University pays a subscription to access the archive, not a per document fee. B works for a pharamaceutical business and his research budget will cover the cost.
A and B would both prefer not to have to pay to read the paper, but considering the cost/benefits they are probably happy enough with the prices they are presented with.
The price of an article from a journal doesn't reflect its value - you can't easily put a dollar amount on knowledge, when prices stop reflecting demand/value there's usually another reason for them.
With a lot of publications you can get unrestricted access to their archive when you take out a longer term subscription - Newspapers do this a lot.
Now consider C - he's interested in the paper for personal reasons - he's not a member of a University and doesn't have a multi-million pound budget. To him $15 to read a paper seems grossly unfair, especially when it would only cost $120 a year to subscribe to the Journal - getting access to everything they've already published for free.
The $15 re-print fee in this case acts as an anchor to make the annual subscription look more attractive.
Journals use re-prints as a marketing tool to sell subscriptions, not as a direct source of income.