There seems to be a lot of misunderstanding in this thread about what MathML is for. What we are wanting, what we need, is for modern browsers to support the rendering of mathematics. To even get off the ground, we need a markup language for the browser to interpret. Since browsers already know how to speak XML, it only makes sense for the markup to be some flavor of XML. Those who are suggesting LaTeX instead are really missing the point here. We aren't solve the problem of a lack of human writable markup. That problem has been solved many times over. The problem we are trying to solve is rendering mathematics in the browser. Period. THAT is what we need MathML support for.
Again, the problem is NOT a problem of AUTHORSHIP. Authorship is easy. It's a problem of DISPLAY. And it is a serious and important problem to be solved. The web was invented to share scientific information. Education on the web is huge--and growing. Academic publishers, mathematical software, and software shims that display math in a browser all use MathML extensively. It's a ubiquitous technology precisely because it fills a need in the industry, and it fills it well. What's more, MathML is important for an accessible web.
PDF is clearly not good enough for digital consumption. PDF is great for print but totally sucks for screens. MathJax is amazing (as are the people behind it), but it is a huge, complicated, and inefficient solution to the problem of math in the browser. The author of the linked article in the submission works on MathJax professionally and is advocating MathML support in the browser. That should tell you something. (In fact, MathJax itself uses MathML both internally and as an input/output format.)