A well written python app will be just as fast as a C or C++ app unless you optimize the C or C++ app in some way (beyond just best practices). Also python is easier for maintenance so theres a bump in its direction. This flies directly at the same age old argument... do you use a custom written hand tuned algorithm thats fast, but a nightmare to maintain. Or do you go for a slightly slower one, that still gets the job done, thats easier to maintain? Personally, I prefer longterm maintenance benefits from easy to read code.
Python also has the ugly problem of its huge memory foot-print. Though not a vis-a-vis comparison, for the longest time I called python Linux's Java (back when Java was a PITA to get working on Linux and there was no IcedTea to cool things off). You may argue that RAM is a non issue... For desktop use, try the confinements of embedded or terminal systems with limited amounts of RAM, then the snake rolls over those limited resources. C/C++ is much a much better solution, for all. Prototyping in python would be a good idea, sadly usually the process stops with the prototype and loses the 'proto' part to be the 'type'.
As for the issue at hand, I don't think that we should waste any more e-pages beating on this dead horse.