I'm very confused by the article. It's missing about two decades of modern non-cryptographic general hash development, it completely fails to mention the field

I'm very confused by the article. It's missing about two decades of modern non-cryptographic general hash development, it completely fails to mention the field of universal hashing which created formal criteria for collision probability and produced provably correct constructs to achieve those criteria... As a self-plug for a related topic, I think going forward we should also take into account HashDoS for non-cryptographic hash functions. All of the hashes mentioned in the article are vulnerable to this, see https://orlp.net/blog/breaking-hash-functions/ https://orlp.net/blog/breaking-hash-functions/ for some techniques.