You are on point: paid journals bring nothing to the table, except that they have somehow established reputations (of varying levels), where people publish their papers. But the journals themselves do very little. It's largely or entirely volunteers who review papers and select the ones that get publish.
The journals are completely replaceable. You could also just designate a simple website as the publishing destination. The volunteers would do exactly the same work. Obviously, the website requires some funding, but a tiny fraction of what the journals take.
> You could also just designate a simple website as the publishing destination.
Just as a quick counter - you'd want to base this on something more permanent. CLOCKSS[0] or something similar is a necessity with DOIs pointing to the papers. Otherwise you're back to "the source code is here: (broken link to old academic site)"
I agree that the use of journals for the literal job of publishing seems antiquated.
(disclaimer, worked previously for the parent company of FigShare, other providers are available)
The journals are completely replaceable. You could also just designate a simple website as the publishing destination. The volunteers would do exactly the same work. Obviously, the website requires some funding, but a tiny fraction of what the journals take.