We already knew how to write useful software without user-level generics, we have been doing it for decades since FORTRAN came to be in 1957, no need for Go to prove anything beyond the stubbornness of its designers.
Useful software can also be written in asm and we have the entire early software industry to demonstrate that.
That's not the same as it being a good idea
Go is about productivity. It allows writing, extending and maintaining big codebases with lower efforts comparing to assembly or some other programming language out there. This is because of simple "what you read is what you get" syntax without implicit code execution. Generics break this feature :(
Of course, there are other brilliant features in Go ecosystem, which simplify writing and maintaining non-trivial codebases in Go - tooling, standard library, fast compile times, statically linked binaries, etc.
Go showed that useful software could be written without user-level generics. I don't think any other language today would dare to do that. In fact most languages seem to be converging into the same thing.