Getting a JIT right is hard work because time spent by the JIT is time not spent doing actual work, and finding the correct tradeoffs is very much an empirical problem.