That doesn't work for the same reason that < and > don't work, right? They are existing operators in the language, so the parser would need unbounded lookahead. Using a different set of characters is unambiguous (and additionally is more compact).