Yesterday, I found this software package called "fdupes" (which is probably "find duplicates"), which searches through the hard drive and identifies duplicate files. There are also options to delete duplicates, although I haven't used them yet.
My first reaction was, this package is amazing. I once actually wrote an extremely primitive version of something like this myself; mine did not work nearly as well. I wish I had found this one at the time.
Now I have found someone claiming that fdupes, although a good concept, is extremely inefficient. He (she?) spent some time optimising it, and produced "jdupes", which is supposed to be 32 times faster. I haven't tried it yet.
But even if it is really slow, fdupes has already dramatically improved the quality of my life.