I understand that some pay-walled journals include timestamp and IP address and/or institution to full-text PDFs to track copyright violations.
However, I also saw an open access journal doing the same despite allowing non-commercial, attributable redistribution and modification.
If that journal grants such rights to readers, then why does it include IP addresses on the copies they download?
Most likely, they are still doing it to track copyright violations. If the document is made available under a license that permits non-commercial, attributed redistribution, then the publisher might wish to collect evidence in cases of commercial or unattributed distribution.
It's also possible that the same publisher, or even the same journal, also has articles that follow the traditional "all rights reserved" model, and that this information is applied to all downloads indiscriminately.
Note that to give a license on a work you must hold copyright to it. Granting the license doesn't affect your copyright. So the issue is really the same as for more formal (commercial) journals.
The situation might be different for things put "into the public domain" for which you formally give up your copyright. Then the public owns it. But licensed works are still owned by their creators.
If you use a licensed work, the license is a contract backed by the copyright. Without the latter, the former has no meaning.
I can license my own work, but, of course, I can't license your work. It is copyright that makes that clear.
See this, for example:Creative Commons
Likewise, see the discussion of Copyleft at Wikipedia.
Without copyright these could not exist.
