Grading: in academia and martial arts

Consider the typical way we grade students in a course. We give a number of assignments. We grade each. Then we average those grades to get a course grade. That course grade is then taken by various outside agents - employers, grad schools, etc. - to be a measure of accomplishment. But there is one very obvious irrationality in this system. Consider two students: student 1 has had the benefit of excellent prior training at, say, a top private HS.  2 has gone to much weaker schools. Maybe 1 has already had some philosophy and certainly has good training in writing papers. 2 does not. Thus, on the first assignment, 1 ends up with a much better grade than 2 because the paper is just plain better. Over the course of the term, however, 2 improves at a vastly higher rate and by the end of the term they are doing work at the same level. Even those professors who include "improvement" as a factor in the final grade are likely to give 1 a higher grade than 2.  (Let's assume that 1 does better on all assignments up to the final one, but stipulate that one's assessment at the end of the course is that they are now producing work of equal quality.)

This seems to me to be wrong and to account for one of the many ways in which I have always been uncomfortable with grading.  At Georgetown, I have students with a very wide range of preparation - not to mention other benefits as in the difference between students working full-time, or engaging in Div I athletics to pay for school vs those who have it all funded by parents.  We don't have the bottom end that one gets at some schools.  All are basically competent at fundamental college skills, but there is a huge range anyway. And it is close to inevitable that those with a worse preparation are going to do worse than those with better preparation. And that will have unjust effects on their future. I mean "unjust" in the most straightforwardly meritocratic sense. Even if we ignore all other factors, the uptake of a B for one student and an A for another is that the latter is "better" at the course material, or at least did better in the course. But that is just false if they both end up with equal competence and understanding at the end. (It is no good, I think, to protest that these grades should be given a more sophisticated uptake. First, they won't be, and second, it is utterly impractical for, say, an employer to dig into the details of what went on in each course.)

Compare the "grading" in traditional martial arts. Here there is no schedule for belt tests. One tests when one is ready.  And there is no penalty for failing other than that you go back, train more, and test again later.  Receiving a given belt, thus, speaks directly to one's current accomplishment level, and says nothing at all about the path one took to get there.  An awkward, 50 year old, with no athletic skills who earns a brown belt after many years of training has, in principle of course, demonstrated the same competence as a super-athlete 20 year old who earns it in a year.

Wouldn't it be better to grade on the martial arts model rather than the university one? One implementation that would be problematic would be to place all the emphasis on a final exam or paper.  But this is not the martial arts model.  This would be like saying that in n weeks, everyone has a belt test and you get it or not once and for all.  All sorts of extraneous factors can make someone do poorly on a single day, and as well, there is the issue of a fixed time-line.

So here are two proposals, one ambitious and one modest: on the ambitious end, departments don't issue grades at all for individual courses. Rather, they issue accomplsihment level grades in the field upon graduation. Letters corresond to levels of accomplishment for undergraduate philosophers, say.  We certify that this person has achieved A-level competence in philosophy, after a holistic evaluation of where they are at now, with input from all their professors. (Of course along the way individual professors would give lots of detailed feedback on how the student was doing.) This is ambitious because it would radically change the way university evaluations are done.   

The modest proposal is that we do this by course. At the end of the course,  you simply assess how good the student is at the various skills relevant to the course. No averaging at all. Indeed, if someone gets to A-level understanding from a starting point of D-level, that should suggest far more in terms of future promise than someone who moved from B to A.  (Of course we need to pay attention to intentional sand-bagging, as we call it in chess, at the outset.)  Again, this is much different than merely giving some credit for improvement. I think a set of papers that move consistently from D to A should, if anything, get a higher overall grade than a set that move from A- to A, other things being equal.

Finally, none of this is to take any stand on more radical critiques of grading.  I'm just assuming here that we are in the business of putting these one-dimensional assessments on work. But if we are going to, and if the uptake of them is going to be in terms of how good someone is at our field now, after the course, then we should at least try to make them measure that as much as possible.  And averages over a term just don't do that.