Do you know how expensive open access publications are? I guarantee that editing, typesetting and curating could easily be performed for fraction of that price (and it's easy to prove it - just check how much Elsevier makes each year!). Also keep in mind that most of editors - which are responsible for curating - are not paid at all... And if you mean "curating" as in providing access to pdfs, then there is arxiv which does just that - for free.

You don't need to convince me. Convince all of the authors out there.

They have the power to send their work wherever they want. Really. They do.

The fact that they continue to choose the publishers after all of these years suggests to me that the publishers are doing something that the authors want.

And btw, who do you think would do this work for a fraction of the cost? Are you going to pay them? What if they unionize and demand a fair share of the grant money?

I feel like the Wizard of Oz telling Dorothy she had the power all along.

In my experience, the average Elsevier paper has worse typesetting than the average LaTeX document. At least in CS journals. The lines are too long, the font is too dense, and mathematical expressions look wrong. Springer, ACM, and IEEE are a bit better, but LIPIcs conference proceedings (where the author is responsible for typesetting) beat them all.

Curation is not access. Curation is editorial, in terms of having a journal issue contain a selection of available new works and not every single thing posted.