"Nothing is evaluated unless necessary

head (sort ls)

The list will only be sorted enough to find the minimum"

This may be technically true, but is a pretty bad example with much hand waving - what exactly does it mean to be "sorted enough to find the minimum" given that Haskell uses a merge sort implementation internally?

A better example of Haskell's lazy nature (taken from the Learn You A Haskell book) would be:

take 10 (repeat 5) - it's very easy to comprehend taking the first 10 elements of an infinitely long list without needing to compute the entire list.