That's right, compression algorithms depend on the content of the files. Most compression algorithms look for some kind of pattern in the data of the file, so a plaintext file for example, which has a very limited number of combinations of data (limited to ASCII usually), can be compressed a great deal. Whereas binary files, being much more random, usually can't be compressed nearly that much, at least not losslessly. Algorithms also have their own strengths and weaknesses, in terms of how much they can compress a file, and how fast they can compress it, and whether they work well on binary files or not, etc.