Is its common use case to enable testing for people who aren't normally writing them, reduce total number of tests maintained for people who are writing them, or for writing additional supplementary tests to guarantee a higher confidence in interface functionality?
edit: context is prototypical M-V-C python or ruby web app development
Yes to: "tests to guarantee a higher confidence in interface functionality "
Selenium should be part of a complete suite of tests (unit, component, feature/acceptance). Selenium's design goal is a "for developers, by developers" feature testing tool -- a browser automation API with bindings for all major programming languages. Classic "perfect user" is an open-source and agile loving Java/Ruby/Python/C#/Perl/PHP programmer.
However, because of the Selenium IDE browser extension for Firefox and its "record/playback" feature, Selenium is also commonly used by people who don't normally write lots of tests. Even though those users are not "real" programmers, there's so many of them... So I try to be nice to them most of the time. :-)
edit: context is prototypical M-V-C python or ruby web app development