I'm not trying to complain! I'm just saying that the type system is A Thing that people need to learn, and I think that a rigorous type system can be a hindrance to learners. As such, I think Wadler should have mentioned it as a possible downside of his hypothetical world where Abelson and Sussman used Miranda in SICP, even if only to debunk it.
"SICP in Miranda" would prepare reader for type system. So the reader of that imaginary version of SICP wouldn't encounter those roadblocks. And those who read Learn You A Haskell don't encounter ones.
I challenge both of those assertions. I haven't done LYAH, because I was learning haskell before it existed, but I did nearly every available haskell tutorial when I was learning.
As soon as you step out of bounds, you find strange corners of the type system. It's quite a complex beast, and many of the error messages are cryptic at best.
I had two colleagues who done LYAH and one who attacked a program with type level programming without LYAH (because of LYAH non-existense at the moment).
The first two didn't asked me or my colleague about anything type system related too often. After month or so of on-off work on his project one of them finished a complex translator from CPU description into VHDL.
The one who studied Haskell without LYAH did asked us. I attribute it to complexity of program he worked on.
My experience allows me to remain confirmed that proper exposure to Haskell type system greatly lessen associated burden.
I should note that we didn't encounter many errors in our Haskell code. Much less than in code in Java or C#.
