I'd also like to note that some parts of Markdown from the user perspective are non-intuitive and clumsy.
Such as links and images (inline).
Markdown works so well because it is intuitive and appeals to those who once saw old word processors. They don't have to worry about syntax, and can just enter their text into a textarea (free from JavaScript WYSIWYG interference and the inherent troubles of running that on old and new mobile phones, their playstation, web browsers, etc)... and it just works.
Yet some parts of markdown are simply not intuitive. Links and images are two places where I see in usability testing that the end user will constantly refer to help documentation to figure out how to do it.
Beyond getting the code consistent, maintainable, and testable I'd love to see the language itself solve some of the papercuts that trouble the lay end user.
Realising that what I was planning to do for my project (discussion forums, tumblr for forums) was to create an alternative to markdown that would resolve some of these user issues as well as parser issues, I had already decided that I would not call it markdown and that I would educate my users in something new that hopefully solves their and my needs and would remain very stable due to a well-thought out and documented design in the first place. If what you're proposing is in this vein then consider my hat thrown in, what help I can give I will... take me to your git repository.