Their point was that by offloading the bottlenecks to C, you've essentially conceded that Python isn't fast enough for them, which was the original point made above