To leave out a grade of a criteria may be considered that the total grade is not complete. It would make sense for my anyway.
Its a little bit tricky because some criteria in the rubric could crucial for the total grade, failing to get a grade on some criteria could be considered failing the whole assignment?
But it depends also how you make the transition to the total grade which you also ask about.
We (Consider we are doing this manually today) use the rubric like this: The student makes a written assignment and we grade it using a Rubric. A total score is calculated.
We put min-max score for each grade. This scores could be different of course depending on the nature of assignment. We use Bologna system which is grade A,B,C,D,E,F,Fx= 7-grade scale. A top grade like A means a full score with maybe a lesser score on some items with lower weight.
The "min-max" value is depending on different factors and the only way for us is to have table where this values are put in. It could be calculated of course but the range for different grades varies depending on which criterias on how they are weighted.
So, two ways are possible(?): a calculated total transfered to grades by some sort of rule or a table in which max-min values are stipulated for different grades. In the later case the final grade could be ordinal scale. (Im not sure if this is what you mean in the spec?)
My suggestion for some kind of cross validation origins from the fact that using rubrics scale with subjective judgements always leads to discussion about how the rubric are built up and used. If two or more teachers can compare their assessement before the grade is finalized it would be a valuable tool that not only would do grading more transparent and more fair but also help make grading a collaborative task between teachers which enhances teachers skills.
I will give all feedback you need from my perspective. I have possibilty to test things on experimental moodle site if you you want some feedback on how it works.
