Fair enough, flexible is an extremely loose term. I was referring mostly to the ability to a language that's flexible enough to let library/tool authors create their own very high level abstractions and DSLs. In Go, lack of custom generics often makes this very difficult. You look at the kind of APIs offered by mega-popular data science toolkits like pandas and Spark, it's really hard to offer something similar in Go. You end up with a lot of inferface{} types everywhere, vectors/series/whatever carrying their type in a struct field, etc.