Comment by QuinnWilton

> Using this programming style requires a rather powerful type system, if you want to go past the simple examples the author shows and encode more complex validity conditions.

You need a strong type system if you want to encode complex constraints at compile time, but you can still encode a surprising number of conditions in weaker type systems. And if you can't enforce a constraint you care about at compile time, capturing that constraint in an ADT using sum types and runtime assertions can still provide a lot of value to anyone reading or maintaining your code.

This sort of style requires a lot more diligence and verbosity than the same code in Haskell would, but I think dynamically typed programs can make a lot more use of their type system than they usually do.

At the very least, humans can try to pick up the slack where a compiler is insufficient, and while designing your data model in Ruby as if it were in Haskell may not help the interpreter at all, it can give the programmer a lot more context about how the code is intended to work, and what assumptions it is making about its data.