Overloading Functions

I think it's perfectly fine. I must admit, it does create complexity, which in turn can lead to confusion. But, look at this situation: You have a string class with 3 constructors: a default, one for initializing and copying a C string, and one initializing and copying another string class. How else would you go about this? And then say you have a templated function that creates a string class. It's not perfectly safe, but because you can't know ahead of time, function overloading works great in this situation.

I suppose you are right, however, given a chance to avoid it, I would!

In some cases, you can overload operators to perform what you think would be logical functionaly but would normally not be provided, thus reducing complexity overall. For example, if you were implementing a vector class, its much more intuitive to simply overload the +/- operators to add/subtract vectors, so you could do this:
vec3 = vec1 + vec2;
rather than:
vec3 = addVectors(vec1, vec2);

Overloaded functions are useful when you want to do essentialy the same thing with different data types. Though I think in a lot of cases this can be avoided by using template functions, conversions and default arguments. It's extremely useful when overloading operators though.

Essentially what BringBackFuturama said. I would also point out that in any non-trivial program, ctors are overloaded (necessarily) all the time. Function overloading is also convenient for compile-time dispatch, where, using the same name, multiple functions can be called based on a type (or trait). In this case, templates do not provide a solution, and in fact are often overloaded themselves (look inside your standard library headers sometime for some examples).