It says something about the performance that is available from the Python language. This is how Python has always been, starting from the interpret and builtin data structures in CPython which are written in C. That isn't cheating, that's normal, realistic usage of the tool. Why should benchmarks go out of their way to use only unrealistic Python programs just to prove that it cannot "x += 1" as fast as C? Who is surprised by that, and why would it matter? Node uses libuv and nobody blinks an eye.