I assume the design contests to which you are referring were cases where participants submitted an interesting project and all of the entries were compared, ranked, etc. based upon one or more criteria. In my experience, such a process is very subjective since the entries don't necessarily have alot in common.
In contrast, I am proposing a situation where the task is very explicitly defined and all of the entries will be very similar. Everyone will be trying to generate the same thing, although possibly via different methods. Such an approach will allow for more direct comparisons between entries.
Is this idea any more interesting than standard design contests? Or, will it just be one more contest that is largely ignored by the users?