> Generalized Star-Height at Most Three

This was an open problem in automata theory I worked on for more than one year before giving up. I'm very curious about their claimed proof.