> They choose the journals because if they don't, their career doesn't advance.
Is this basically saying that publishing in non-big-journals (e.g. ArXiV) doesn't really "count" as citations? If so, then that's probably the right language to use here - that citations are a currency that only big (expensive, monopolistic) publishers can actually pay for papers in.
It's a classic collective-action problem: you need a large portion of researchers to more or less simultaneously stop doing it. Good luck with that.
The journal system/cartel is an effective oligopoly.