I've used it (by using Normalized Compression Distance) in my Master degree.

It was very interesting to find out how efficient it was in authorship attribution, even having 100 possible authors.