It comes down to project size. Small projects with a small team tend to be fine. Bigger ones tend to break down a bit and the application becomes ossified because there's no safety when changing things. You can power through it but productivity takes a nose dive. The big ones that stay on Ruby tend to invent typing like things to solve the issue.
If there's no safety when changing things, then you've failed to define and verify adherence to contracts along component boundaries, and the lack of a proper test suite.
I've certainly seen that happen in Ruby, but I've also seen that happen in every other language - most of them statically typed - I've worked with, as very few languages have type systems expressive enough to prevent it (and it tends to massively destroy productivity to try check everything strictly enough anyway).
