One can think of compression as an attempt to increase entropy in representing the data. The less predictable the data is, the more entropy it has. Compression

One can think of compression as an attempt to increase entropy in representing the data. The less predictable the data is, the more entropy it has. Compression exploits patterns in the data to increase its entropy. So technically we shouldn't talk about compression level as an absolute. The input file will have some level of entropy and the question is what is the maximum level of entropy it could have and how close to it can the given compression algorithm get. For example, encrypted data is hard to compress because it already has high entropy. "Better" usually means "is able to identify more or longer strings of duplicate data (or other patterns) in the file". Reference: https://en.m.wikipedia.org/wiki/Entropy_(information_theory) https://en.m.wikipedia.org/wiki/Entropy_(information_theory)