As you may guess, this is a recursive poll :) - we don't have the weighted Polls yet, so I'm asking if this sounds like a good idea, and in effect showing how different answers could express a different "level" of feedback, which we can then use to order answers and extract popular vote data.

This would be useful in a tests module in an LMS... different weight for answers, completely correct, mostly correct, incorrect etc. 

One of the reasons this has come up is that in our current process we often have trouble determining which features/enhancements are "most wanted". Often a vocal member may create a feeling that something is really critically needed, while there may be a much more important issue that gets pushed back because of this. So, I am thinking of a way to grade answers and create a running list of most-wanted improvements using Polls.