The annoying thing is they have a performance cost though: list comprehensions can be very fast because they stay in the C codepath of Python, whereas unrolled into a loop you lose some of that.