It the object dynamics part of the funnel definition?

Pro:

• Let’s say I don’t care about the pre- and post-conditions. I just want the object to behave in a certain way, e.g. it should continuously rotate or spin - or to have a certain velocity (e.g. for throwing). Then the object dynamics need to be encoded in the funnel.
  

Con:

• If I don’t care about pre- and post-conditions but just about the dynamics, then I should use something different than a funnel. We never said that every manipulation automatically is a funnel.
  
• Just as a grain of sand can follow a wide range of different trajectories - including completely weird and crazy, but also roughly linear and boring ones, the object in a manipulation funnel can exhibit a wide range of different dynamics. But the funnel is not defined by these dynamics.
  
• In fact, the benefit of a funnel is that it generalizes across all the different object dynamics. It just states pre- and post-conditions, i.e. the initial and final configuration of the compliant, actuated constraints (and maybe the object - depending on how we decide on that issue).
  
• We argue that the good thing about a funnel is that it generalizes across objects. But different types of objects can follow different dynamics. Therefore, we should also argue that the funnel generalizes across object dynamics
  

[Aditya @ Google-Doc]
You need some understanding of what those dynamics do (i.e. some characterization of the preconditions and postconditions).
This understanding will turn the dynamics into a computation.

I agree with pre- and post-conditions, i.e., funnel opening and exit. But I disagree that a funnel definition should state the dynamics of the object.

You can use a funnel, although you do not understand how it does what it does. You only need to know the pre- and post-conditions. In fact, this is how we all use tools in our everyday life - I don't really know exactly how a microwave oven works. But I know how to apply it, because I know its pre- and post-conditions.

The problem is that if you define a funnel by the behavior of the object, you would need different definitions for different types of object. Because different types of objects can exhibit very different dynamics. E.g. a large grain of sand (e.g. a stone) would exhibit a rather different behavior compared to a small grain of sand when inserted into a common funnel.

It could make sense to explain what will happen when a specific funnel is applied to a specific object. But I would not include it in the definition.