I can see value in this concept, but I’m also very leery of turning over my decision-making authority on grades to an automated system. I simply don’t know how trying to quantify the nuances of writing would work. Or if it even should work!
I can totally see this as beneficial for a fundamentals of writing course where compositional mechanics are what is being assessed. My technophilia only goes so far, however. When the machines start making decisions for me [Are you listening, iPhone4S?] then I start to get the Skynet jitters.