Now, of course scientists could run a reputable journal for free or on donations. However, once you have achieved a reputable status with your journal, it becomes something that can be milked for money. And generally people fail to resist that temptation.
Even if they resisted, they still have the entire academic publishing industry very scared, and as we can see, these are people who aren't afraid to use the dirtiest tactics to protect their position.
Even though the status quo is strong, it can be dismantled.
They could offer the curation without the exclusive access and I'm sure many scientists would pay for that curation.
Like the stilted writing in patents, academic writing often has a loud thundering message in the content that is never actually expressed in the writing. If you have the right context it hits you like a red brick, to the lay person is has to be explained.
Some work is like, omg this is going to change the world and another is, showing another construction of an uninteresting thing we already knew. Both are presented in the same manner. Much of this meta analysis happens on twitter and reddit, but is easy to miss. It would be nice if it was contained within the journal structure itself.