So what if publishing research articles became more like publishing music?
When a garage band is first starting, they are desperate for anyone to listen to them, so they perform for free and give their music away. Once things start to click, then they start to think about money. They sign the coveted record contract where they assign copyright to the label and get a royalty payment for each CD sold. They also start charging admissions for shows. And if they hit the big time, then their music is pirated and they make all their money by selling out arenas for hundreds of dollars per ticket.
What if a similar business model existed for researchers? You wouldn't have the long-term record/publishing contract, but you would still have (as we currently do) assignment of copyright to the publisher. Just like a garage band, young researchers are desperate for anyone to read (and cite) their work, so they would be content with making no royalties. When they get established, then they could consider getting a royalty for copies of their papers. And if they win the Nobel prize, their papers would be pirated and they could charge admission to their talks. (Ever tried to squeeze into the room where a Nobel Laureate is about to talk? They could charge admission and still have a packed room.)
Now where to roadies and groupies fit into this scheme?