It has been somewhat interesting to see how the Rails community has handled the recent hubbub regarding scaling. I think Tim had the best perspective I have read, but I haven't read everything of course. The point I took from Tim was that Rails is fast enough for a lot of applications, which I agree with and had I found a job working with Ruby instead of of Python I would have been perfectly happy writing Rails applications.
In the world of Python and WSGI I think scaling might be slightly easier because there are so many great libraries and patterns to follow. In a way this doesn't seem pythonic because there should really only be one way to do something, but from an architectural standpoint I would imagine allowing many options using explicit configuration still falls under the pythonic category. The downside is that you don't get the same quick start you get with Rails. I am not saying you can't achieve the same productivity with Python because you can. The difference is that you do not automatically have a single framework that helps you make basic database backed web applications like Rails provides. Again, I am not saying that TurboGears, Pylons, Django, etc. are not just as good as Rails, but rather they are all different and help to create different types of sites.
Generally I think Python libraries allow a slightly more difficult start up time but result in a more customized and eventually scalable architecture. The sacrifice is that the initial development time could be a bit slower. Like everything in life, it is all a game finding the right balance between form and function.