Python is slow. For loops containing arithmetic, array indexing, function calls, and similar primitive operations that also exist in C, it's maybe 100× slower than C. But programs written in Python are usually not 100× slower than C programs with similar functionality. They may be 10× slower or 3× slower or, very rarely, faster; that's because their inner loops tend to be in operations like regexp matching, array arithmetic, data compression, string hashing, and the like, and Python's implementations of those things aren't themselves written in Python — they're the things you're likening to "gzip" here.
There's a selection bias here. If you write your compiler or interpreter or pathfinding algorithm in Python, it's going to be a lot more than 10× slower than if you wrote it in C (or Java or OCaml or whatever). So people don't write those programs in Python, so we don't get to compare them.
I've been writing about Python because I know more about it than I do about Ruby, but I think most of the same things are true of Ruby, as well.
Sure -- minimizing the number of Python statements is a decent optimization heuristic in Python (and, as a bonus, that can help you optimize for human readers as well when not taken to extremes).
