If you want to make a game for a particular platform, use the tools designed to make games on that platform. You'll learn a lot and it will save you a lot of time learning one technology, and then having to go back to relearn a new technology when you reach a certain stage where your development capacity is limited by the tools you are using.
I take it you are advocating learning C/C++ to make games before learning Python? If so, consider why anyone who already knows C/C++ would bother embedding Python or Lua in their game... That would also be "going back to relearn a new technology when you reach a certain stage where your development capacity is limited by the tools you are using". And if you need to know both C and a scripting language to make a game, it doesn't really matter which you learn first... except that by learning the scripting language first you can prototype earlier and faster.
(And, considering the article's context of indie games, being able to rapidly prototype game designs is more important than having them perform well from the start.)
EDIT: link: http://www.grimfandango.net/?page=articles&pagenumber=2
Building a game in a week is an invigorating challenge. I learned python during one of the PyWeeks and I'd never made a game before. The best part is that because of the short deadline it isn't some huge amorphous task. Even if the game is crap, you learn something, have some fun, and get to see how more experienced people get things done.
I also like to take a holiday from the languages I have to use for work. I mean, Flash games are pretty hot, but spending some time in a different language is good for the brain. Regardless of tool analogy.
