I think that the issue you're describing about `yes` was around building a buffer and issuing one big write() syscall instead of a bunch of small ones. I'm not sure about the actual implementation of the IO monad, and whether or not it coalesces call to write(), but it would definitely be possible. Also, idiomatic Haskell would make large writes, because usually (at least in my experience) you create a large string/Text representing your output, and then putStrLn or print it all at once.

In regards to assembling the stream: when the list is actually materialized, I assume it's done one element at a time, but the LLVM backend that GHC uses might be doing some loop unrolling, but I'm not sure.

Stream fusion doesn't require that all of the streams be the same type, though! Elements at different points in the stream can have wildly different sizes. For example, this would be fused:

    sum . map length . map double $ ["hello", "Symmetry"]
        where double x = concat [x x]

That will never materialize a list of Strings, even though at two points in the stream you had a stream of strings.