They offer reputation. Career advancement is highly dependent on publication history and impact. Getting into a prestigious publication means your work will more likely be read and cited. Because highly reputable journals can charge high publication fees (because it’s in such high demand), they get to set the industry norm, which other less reputable journals/publishers get to follow. It does cost money to develop and maintain that reputation for rigour and impact (i.e. good science). But yeah it’s exploitative AF. There are attempts for less profit-motivated publications… But making those rigorous while still being democratic is hard
You could curve the proportion to income to scale impact to something more equitable. How you decide what’s equitable would be another problem to solve, but I imagine it would involve benchmarking around the middle class and poverty line. Right now fine rates are okay for the middle class, so keep the proportion similar, fine rates really fuck up poor people, and fine rates mean nothing to the upper class. So imagine you you feel would be a fair impact for a fine and scale it accordingly.