Thanks for your detailed answers.
#5 - I see that I forgot to give my opinion, which I am afraid is the reverse of yours. I think question outcomes are context-specific. That is, it is a property of the question in a particular quiz. To justify that, consider these use-cases:
- Teachers in different states collaborating on a question bank of maths questions, where they all have to map them to their own State standards.
- Similarly, text-book publishers wanting to sell question banks.
- Or, a question that could be seen to assess both two different outcomes A and B in one standard. However, in the quiz we are building now, we are only interested in assessing A right now, so we want to ingore the contribution to B, since it is irrelevant noise.
I can see the scope for meta-data on the questions, to help teachers set things up (e.g. "if you are adding to a quiz using this standard, the likely outcomes are A and/or B"), but I think teachers need the flexibility at quiz building time.
I do have a vested interest here. A long-standing feature request, that applies even if outcomes are not in use, is to have one quiz report multiple scores to the gradebook. (The gradebook can cope with one activity mapping to several grade-items, like the Workshop module does with the score for submission and score for grading.) I would like to implement that one day, and I don't think it should only be available when outcomes are in use.
Of course, if the quiz had that functionality, then it would help set up the outcomes thing. The quiz would set up one grade-item for each outcome, and then the outcomes code would only have ot look at that grade-item. That would give good decoupling between the quiz code and the outcomes code.
A similar bit of design might also resolve the piont David makes about different advanced grading methods.