My preferred flavor of markdown is pandoc, so I forked and added code to add support for pandoc-ruby as a markdown interpreter. But I'd also like my site to generate on servers that don't have pandoc installed, so I've added code to allow the _config.yml file to include an ordered list of preferred markdown interpreters.
The changes can be found in the pandoc-ruby branch of my fork. A more complete description of the the changes and how they work can be found in the associated wiki. I haven't written any tests to associate with the changes because I don't know how to.
The website Math Overflow uses Mathjax http://mathoverflow.net/
The fork is probably a bit out of date. I haven't had time of late to merge in changes from upstream. Adding support for pandoc is trivial using pandoc-ruby. And I think the extra code to fall back to some other parsers was unnecessary. Also I haven't seen any indication that the Jekyll devs are interested in adding this feature to the official version, so I have not been motivated to polish it up any further. Still, I use it regularly without any trouble.
I was finally able to put together a working pandoc plugin instead, providing all the benefits of the fork but without forking. So I'm closing this issue.
Thanks for the plugin dsanson