Well, the administrators have outsourced the domain knowledge to the journal reviewers and pay for this service indirectly through journal fees, so from an accounting point of view it makes total sense.

If not publication, what metric should funding agencies use to decide who gets grants? Of course they could hire senior scientists in the field to evaluate grant roposals, but that would mean hiring lots and lots of scientists across all the fields in which they disburse grants; those jobs would be sinecures and from the point of view of the grant applicant it would still be about 'who you know,' just a referee at a grant-issuing agency instead of an editor/reviewer at a journal.

I don't disagree with you about the problem, but I'm not clear on how you would address it. Making the journals less profitable by refusing to pay silly subscription prices to read publicly-funded research would be a good thing for universities, and maybe for scientific publishing, but it's not going to fix the disconnect between people who control the public purse strings and which scientists will get the grants they want.