The question that raises from a process improvement perspective then is “were the first 3 rounds really effective tests?” Perhaps a better solution is not more interviews, but more focused interviews conducted by the people that actually have the knowledge and power to make the decision. (And if the knowledge and the power are divided among multiple people, another great improvement would be empowering the people with the knowledge.)
Fun fact, any game dev’s financial data can be stolen if you’re capable of answering my riddles three