>It all does strike obviously inefficient though. In my day to day research I almost never benefit from work the publishers did. I don't even use them to download my papers because the paywalls are so annoying that I usually go directly to the author's home page (google scholar will even find these for you). Every half-decent CS researcher makes their work available on their personal website, at no cost,
Different fields have different needs. You don't need too much trust in CS or math, because you can just evaluate the arguments and proof. But other fields require much more, and you do benefit from the work they did in only showing you certain papers and not others.
In all fields, the peer review process is performed by volunteer researches at no cost. The publisher DOES NOT evaluate arguments or ensure that the contents of the paper are correct. The only thing the publisher does is own the name of the prestigious journal.
> Why is it considered hard to get into those journals
Journals only publish so many papers in a given time. Its not just a matter of submitting a good enough paper.
Dunno why this is the case for journals but when it comes to conferences (where lots of CS research is published) there are obvious time limits regarding the number of presentations that can be accepted.
> why do people give them prestige and won't give a new startup journal prestige?
For historical reasons and perverse incentives, sadly. Young researchers cannot afford to publish on new startup journals because advancing your carrer depends on publishing in prestigeous journals (universities evaluate researchers based on the prestige of where they publish). Meanwhile, tenured professors are under less pressure when it comes to career advancement but they also need to publish on prestigeous journals to secure grant money to fund their labs and to accomodate for their younger co-authors (who are still trying to advance their carrers).
It also doesn't help when the "journal prestige categorization" is defined by the government, as happens in my country[1]. Not only are our journal ratings always outdated (punishing startup journals and conferences) but they also emphasize publishing in journals over publishing in conferences (which is terrible for CS research in particular)
Once again: the curation is not performed by the publisher, but by the editorial board and the reviewers, who are researchers, and are not paid by the publishers for this work.
Paid academic journal editors are rare exceptions.
From one of your own links:
Carol Barnes, professor of psychology and neurology at the
University of Arizona in Tucson, is a reviewing editor at
the Journal of Neuroscience, a 3-year appointment for which
*she receives no compensation*.
(…)
Barnes has received funding from the university to hire an
assistant to provide clerical support to assist her with
the manuscript review process. She considers herself
fortunate, because "without this help, I would have had to
decline this position."
Not only she was still paid by the university for whats seems to be a full time job during 3 years according to the article, but in addition to that the university, not the publisher, paid for an assistant to this job!
Different fields have different needs. You don't need too much trust in CS or math, because you can just evaluate the arguments and proof. But other fields require much more, and you do benefit from the work they did in only showing you certain papers and not others.