I think comparing Go and Haskell are like comparing two incomparable different species -- a fish vs. a cat.
Why? Because Haskell is an interpreted language while Go is a compiled one. Interpreted language doesn't care much about performance as it isn't designed for that purpose, while in the other hand, compiled language does. As a result, interpreted language tends to be more 'elegant' and has lots of convenient features at the cost of performance. A concrete example is when you talk about preventing unacceptable data type in Haskell. They could make it so in Go, but the performance cost would be undesirable.
IIRC, I read that they designed Go to be practical instead of 'elegant', the reason is so that people can learn it easily, making it a good alternative for other compiled languages like C++ whose learning curve is hugeeee and ugly!
There is no such thing as a "compiled language" or an "interpreted language", as this is a property of language implementations: interpreters (GHCi, Hugs) and compilers (GHC, JHC).
Haskell is compiled, and most of the safety features (i.e. everything in the type system) is checked statically at compile-time. Preventing unacceptable data types happens during compilation. The same could be done in Go with no performance cost.
Why? Because Haskell is an interpreted language while Go is a compiled one. Interpreted language doesn't care much about performance as it isn't designed for that purpose, while in the other hand, compiled language does. As a result, interpreted language tends to be more 'elegant' and has lots of convenient features at the cost of performance. A concrete example is when you talk about preventing unacceptable data type in Haskell. They could make it so in Go, but the performance cost would be undesirable.
IIRC, I read that they designed Go to be practical instead of 'elegant', the reason is so that people can learn it easily, making it a good alternative for other compiled languages like C++ whose learning curve is hugeeee and ugly!