I already addressed this. People know peer review can be bad, but some think "good journals" still do good peer review. This is not so clear.
> In what world does the arxiv system moderate this discrepancy?
Open systems allow the scientific community to figure out ways to properly assess research quality and value more cheaply, and without passing through (often arbitrary and random) small numbers of gatekeepers that don't even do a reliable or good job gatekeeping in the first place.