The utility does its job without a doubt. However, casual users actually expecting to improve computer performance with such a tool shouldn’t don’t bother. By principle and not by any fault or lack of foresight by the programmers, such a program is limited to detecting ALL duplicate files regardless of whether or not they serve a function.
Certain programs may make use of the same DLL library which is by no means necessarily packaged with some sort of framework (such as Java or .NET or .NET’s predecessors). Other situations call for an archival installation copy of a file and an active one.
Whatever the reason, duplicate files exist many times for a functional purpose. Creating a detection system to discern whether or not a duplicate file is a functional one would require a database of some sort which contains knowledge of all programs even if there is a cross checking routine with the registry.
In other words, duplicate file detection is only recommended for extremely advanced users searching for specific duplicates while exercising a high level of discernment and technical judgment.
As all CSC users know, Duplicate Files utility has two major options to choose from: simple scan and advanced scan. Although the options imply that one is simpler and one is more advanced, no matter which option is selected, the results presented after scan require a careful reading and a more advanced user that will know exactly what items to select, items that s(he) knows for sure that can be safely removed(as you pointed out, a knowledge database should be used as a reference).
CSC is a cleaner, which basically means that it will help you remove all kind of unnecessary things from you system. Duplicate files are part of those things. So a regular user can use this utility for removing the sort of duplicates that are popular on most computers in time: movie, music and picture files.