The thing is (and this relates to issue of racket changing surface syntax) - s-expressions are not just about the editing experience. They are about the reading/mental parsing experience. The less time I have to wonder about operator associativity rules (let alone magical language keywords/formats) the more time I can think about the actual semantics of the program being expressed. It always amuses me when using a mixfix language that the first thing to do when debugging a tricky math expression is to get hella parenthetical to make sure things are working the way you intend.