"Grading is not, or should not be, about the grade"
I worked with Dr. Shermis back in the mid '90s on this (one of the professors quoted in the article) and we had our own software we were working on.
One of the very first things we had agreed upon was that grading and rating are two very different things. And when we worked on our software, it was designed to give back several scores that summarized why there was an overall score that was given.
Beyond this, *MOST* work in this area are not based around how to refine a research question or structure an argument. It was based around how to rate writing. From here, verbage could be created based around all the factors explaining why a particular metric was used as opposed to simply providing the metric. I left this field before it got that far, but its been a decade and I'm quite certain that this has happened because there were enough engines out there that could work in collaboration with one another to provide both quantitative and qualitative feedback (at the time, too many researchers were working against each other and either closely guarding their secrets or blatantly looking down their nose at the works of others because it was approaching the idea with a different vision).
That said, the #1 way to become a better writer? Write more. Even without feedback, you become a better writer simply by exercising these skills. With plain metrics? You become a far better writer simply by seeing that the computer thinks your mechanics could be improved. Maybe work on your creativity (this was by far the hardest to calculate), maybe see how improving one area weakens another as there were proven links between different areas that were almost always inversely proportional to one another except in the hands of talented writers.
And the problem with writing more? Instructors, especially writing instructors, are overwhelmed as it is. I would GLADLY grade 100 undergrad psych essays as opposed to a technical writing course. In our studies, one of the ideas was to allow more writing assignments, while giving more quality to feedback by the instructor...allow the instructor to read every 3rd paper for instance and give in depth analysis. Or maybe give more time to students that were struggling with basic concepts by catching them earlier. Almost all of this was designed to find ways to help the instructors do more with less, which is the reality of higher education these days. And when we did this, students actually got far more support than they did without our software.
Will their be lazy professors that just phone it in? Probably...but most instructors I've worked with have been passionate about their jobs, and the ones that weren't? They got this way because they were overwhelmed with the lack of resources provided to them.
BTW...every time I write about these subjects some jackass mentions my writing style. I'm not writing for publication, I'm writing for conversation (and most likely not proof reading nor thinking much more than a few words ahead...the opposite of the academic writing I use to do).