If you look at the original comment (or at least a close as I've been able to find for it.) <http://groups-beta.google.com/group/alt.religion.emacs/msg/2988394c3082559f> the context is interesting. One of his complaints (and in that thread, he had many) is when things are kept as a byte stream, and that regular expressions force you to keep things as byte streams. In the case of a shell pipeline, what you have is predominately a byte stream.

I actually think that the "sort | uniq -c |sort -n" is more confusing to the bourne shell novice, but it is so commonly used that it eventually gets recognized as a single "sort by number of occurrences" token.