For those who have learned how to work with them, S-expressions are the best notation for code representation. Interactively, they can be very conveniently navigated and manipulated (with competent editors); programmatically, they can be transformed, generated, serialized and read back with ease. When properly formatted and indented, they are also very readable, and reflect the larger structures of programs well.
Perhaps the only place where s-expressions may be at a slight disadvantage as a notation is representating short snippets of mathematical formulas. A properly typesetted mathematical formula employing the elements of super- and subscripting, stacking and all sorts of 2-dimensional layout schemes can convey a lot of information very compactly. However, they are unwieldy in terms of editability.
This mode aims to bring the best of both worlds together, where marked s-expressions in a Lisp program are rendered dynamically as mathematical formulas.