My criticism is mostly aimed at the language (and its standard library) itself. The major projects I worked on, I architected from the beginning. It is when designing and architecting reliable systems that Python's terrible nature hits you in the face, especially if you're used to better languages.

Take the nonsensical but alluring Python motto "There should be one -- and preferably only one -- obvious way to do it." Assuming this was even true, which it isn't as a series of Python debacles - by its designers no less - over the years has proven, I don't want a language that shoehorns my thinking into "one"-ness. One-ness doesn't solve hard problems. I want modeling flexibility, the ability to examine a problem from multiple perspectives and explore the solution space in-parallel from multiple angles of attack. C and C++, bad as they are, let you do this. Perl lets you do that. Common Lisp is the best language I know on that front, it is _designed_ to let you do this.

I'm not going to say that Python is outright hostile to that mode of working, but it's not designed to bend / be stretched or mold with your mind in the same way that great languages are. Sooner (if you're solving fuzzy problems) or later , you'll run into its limitations and then you can't help but think that you're wasting your time.