Maybe, but it's also why synthetic benchmarks are not convincing. Maybe node better handles a bad pattern, but if performance is a concern to this level we're more concerned about functionally equivalent good patterns even if the code is very different.