With a proper test framework, people can and are building large scale applications. There are plenty of ways to optimize performance without a complete rewrite. Python is really Python+C to a large extent.
As for your jibe, my original point was that there is something between Python and Assembly. There are many languages that provide all or many of the nice things that Python gives us that are staggeringly more efficient and safe.
Python is really Python+C if you’re willing to write C extensions, which completely negates the point of using Python. C extensions also don’t maintain safety, or solve Pythons shift of compile time bugs to runtime. It really only addresses the efficiency losses by making you write code in another language.
People can and will build successful applications in all sorts of crazy ways. It’s a fallacy argument that just because something is done a certain way, that must mean there are no better ways that could save us time, money, and sanity.
Prototyping usually involves working with existing code base and the ecosystem, although if you are building a brand new product or application, your context may be different.