Maybe don’t assume other people are “wrong”? A strongly typed codebase in a larger system almost certainly has untyped boundaries (HTTP, JSON, SQL, etc.) and likely to data driven with combinatorial complexity. Going into such rewrites with a great deal of respect for any existing code that “works” is very wise.
If you view refactoring as this risky, you are I imagine using a dynamically typed language with automated tests few and far between.
In a statically typed language and with a decent automated test suite, refactoring is just another coding activity.
It sounds like people out there are terrified of going near their code with a 10’ pole. That is heart breaking to see.