In the paper and these final presentations you need to sell your ideas on parameterization improvements. Picture yourself among a group of collaborators with whom you are planning to write a joint proposal. The money is there, but you have to convince your colleagues that your ideas are exciting and time is short!. In 15 minutes, you must capture their imaginations and swing them over to your way of thinking on this topic, or you'll end up having to follow "Deadly Fred's" tried and true method . You must explain your ideas for the parameterization improvements you advocate or justify scuppering the old scheme (Deadly Fred's favorite!) in favor of starting a new system from scratch.

Should we expect a unique solution between cloud or boundary layer structure and large-scale averages of other variables? If we could observe a number of different situations on the fine scale, is it possible/likely that they could have the same large-scale averaged quantities? Think of examples for/against this possibility.

If large-scale averages do not determine the subgrid-scale structure uniquely, this pulls the rug out from under the basic premise of parameterization! How could we design an experiment to determine the uniqueness of a solution?

Alternatively, is there any way to determine whether an overriding driving toward some new balance (as all stability related parameterizations do) is all that is required for a successful and realistic simulation, since most widely-used parameterizations have been shown to remain within the bounds of reality?

Is there a relative scale dependence in your answers? i.e. does the definition of "large-scale" and "subgrid-scale" determine the outcome?

Is using one model to validate another an appropriate approach? If so, should any constraints be placed on how/when this is done?