I think the comment here is misleading:
`-/ι4 ⍝ (1-(2-(3-4))), not 1-2-3-4!`
Because it is exactly what you get if you type 1-2-3-4 into the APL interpreter. The reduction is completely consistent with APL’s general right-to-left associativity and lack of distinct precedence. It would be better, I think, like this:
`-/ι4 ⍝ 1-2-3-4 by APL rules, i.e. (1-(2-(3-4)))`