Look into v-ABA from Abaraham et al.

[amiller]

This paper appears to offer an improvement on validated broadcast. Whereas honeybadgerBFT requires O(log N) expected rounds, due to N instances of binary ABA each of which is geometric distribution of rounds, it should be possible to refactor our ACS protocol so that it uses theirs and works
https://www.researchgate.net/publication/328758146_Validated_Asynchronous_Byzantine_Agreement_with_Optimal_Resilience_and_Asymptotically_Optimal_Time_and_Word_Communication