UNIX has traditionally had a simple privilege model where the "root" account bypasses all system protections, therefore gaining root access is the big prize for an attacker.
Because the "root can do anything" model provides poor security containment, there has been a general tendency to move all non-critical functions away from the root user.
Common ways of doing this are:
Some more recent UNIX versions have more sophisticated ways of limiting privilege, such as capability models that grant individual privileges (such as the ability to bypass file permissions, or the ability to change UID) to individual users or programs, or the OpenBSD "privilege escalation" mechanism which can indicate on a program-by-program basis which system calls it is allowed to execute with privilege.