The permission edit_borrowers specifies: Add, modify and view patron information. However, when that permission is not enabled, staff can view patron information in a number of places, including holds queue, circulation.pl, waitingreserves.pl, and pendingreserves.pl. In the Holds Queue staff can see patron name and other info for patrons at their library, but patrons from other libraries show "patron from central branch" etc. When looking at a bib record, if an item is checked out, only the borrower number is displayed. This should be consistent, where staff without that permission only see a borrower number and nothing else. To replicate, create a staff member with only permissions catalogue ( to log in) and circulate_remaining_permissions. Check out to a patron, see the holds queue, etc and see that the patron information is displayed.
I don't think that this really makes sense. Arguably "edit_borrowers" could be changed to just "add and modify" patron information, but if we want to restrict viewing patron information, then there should probably be a new permission to control that. Of course, that wouldn't be backwards compatible, so that probably wouldn't happen. Maybe a system preference in addition to a new permission I suppose...
I agree with Donna that, since the edit_borrowers permission says "Add, modify and view patron information" it should actually work that way. If, as David suggests, the wording in the description were changed, then I would want a separate permission that controls "view patron information" because we have staff at 50+ libraries that are currently restricted from viewing borrower address information that we need to continue to restrict from viewing address information.
For what it's worth, off the top of my head, it would be a massive long expensive development with little chance of success to hide the borrower information as desired. I could be wrong, but that's my first instinct. I can also think of a few holes (like reports) that would probably be impossible to plug without completely rewriting entire modules in Koha, which would then break many workflows and expectations. But that's just my two cents. Someone else might have a good idea.