At most it's an error of 'a-b' not in scope. Unless you have dynamic scoping or somehow have a variable 'a-b' and also want to perform 'a - b'. I think the rule

At most it's an error of 'a-b' not in scope. Unless you have dynamic scoping or somehow have a variable 'a-b' and also want to perform 'a - b'. I think the rule that binary operators must be separated by spaces is much easier to reason about then not having certain characters usable in identifiers.