In math, at least, these transfer-of-copyright forms are increasingly allowing by default posting of (final submitted—see juretriglav (https://news.ycombinator.com/item?id=8194544)'s link for the importance of this modifier) papers to the arXiv, or even just to one's web page. This seems such an elemental freedom now that it can be hard to remember that it wasn't always the case. I remember, but can no longer find—does anyone know the source?—a post a while back from an academic who asked (probably) Elsevier for permission to post an article on his home page, and was denied it, and who then essentially dared them to sue him for posting it anyway.
Whenever I've been published, I've asked to retain copyright and that has been granted -- I'm not sure it's a grant, though, since it's mine to start with. The usual negotiating starting point is a contact from a publisher that automatically claims copyright. I simply delete that clause (and possibly insert "The author retains copyright", so that there's no doubt).
Clearly, that's not happening here and I don't understand why, unless publishers are refusing to publish without being given copyright. In which case, alarm bells should be ringing very loudly.
It's not about handing over copyright but it's generally agreed that publishing multiple copies of the same work is a bad thing. This is typically done by researchers to boost their paper count (+ a bit of self-citation), and can make a hypothesis seem better supported than it is.
So only dodgy journals will re-publish work that has been published elsewhere.
