Interestingly, nobody mentioned Mathematica's (Wolfram langauge) approach of getting Unicode characters into code. "High level" mathematical symbols can be used in Wolfram language all over the place, for instance the arrow → indicating a Rule (http://reference.wolfram.com/language/ref/character/Rule.htm...) or ∞ indicating Infinity (http://reference.wolfram.com/language/ref/Infinity.html).

As Mathematica inherits the homoiconicity from LISP, ie. the paradigm "code is data", it abstracts the code representation (called "Forms", http://reference.wolfram.com/language/tutorial/FormsOfInputA...). Every shortened code full of greek symbols can be written in a completely similar form in ASCII. Thus it is solely to the Mathematica notebook (a Qt-based GUI) to render the Unicode. The GUI also allows to quickly enter any named symbols with an approach like typing "[ESC] alpha [ESC]" and an α appears (cf. http://reference.wolfram.com/language/tutorial/SpecialCharac...).

I think this is the right approach: Let the beautification to be done by the code viewers. The approach to enter named symbols probably stems from (La)TeX where one writes "$\alpha$" to get α and is typical to Computer Algebra Systems. For instance, SageMath as well as SymPy allow to define something like a=var("alpha") and render it like α.