If someone could run smhasher for the hash functions in the original article (and perhaps also MurmurHash 2/3) and summarize the result for the small key sizes I will give them a cookie.