Because of that it wouldn’t surprise me much if that sped up some standard benchmarks, for example ones parsing lots of small json objects into dictionaries.
(Slightly less silly: the folks at https://github.com/faster-cpython are doing great work, too.)