Okay, so I like Casebash's approach of taking SO as the template but, as Grigory says, there is perhaps more necessity for a divide here*, there is also more scope.
Certainly, as Casebash says, the broadly discursive but well motivated should be hit with the wiki hammer (and by implication, those with poor motivation are closure fodder), as should big-lists.
But notation, terminology and math history questions, as Grigory points out, provide an interesting borderline case- for my money:
- What does the turnstile symbol mean in logic? bad question
- Intuition for the turnstile symbol? CW
- What is the difference between the turnstile and double turnstile symbols in logic? Rep
- Is 0 a natural number? bad CW question
- Why might one consider 0 not to be a natural number? good CW
- How much of arithmetic is possible without 0? Rep
- Who invented the integral? bad question
- How has calculus changed since its creation? CW
- How did Newton's definition of the integral differ from Leibniz's? Rep
Notice how it is possible to make repworthy questions in each topic, but also useless ones. In most cases the CWs are the ones that are broadly discursive, but in some cases the rep ones can be a little also. What is key is that, if a slightly discursive question requires special knowledge and care , we should offer rep as an incentive.
Also if a question on a discursive topic is asking about a very specific and well defined subtopic, I feel we should deem it a fair question.
*It has been mentioned elsewhere [by @Noah Snyder?] that the proportion of people using math professionally to those using math in general is always going to be lower to the corresponding proportions of those using programming.
Edit: The problem of this question, which is both specific/well-defined and requiring no real special knowledge or care might be seen as something of a Godel number to my Principia Mathematica... That said, I think we can leave it rep and the votes will take care of it ;)